"Invisible Inequality: The Two Americas of Military Sacrifice," by DOUGLAS L. KRINER* & FRANCIS X. SHEN, Memphis Law Review (2016)

 Just found this 2016 article in the law review of my alma mater, now known as University of Memphis.  It was cited in a New York Times article today on Shanksville and 9/11, stating in pertinent part: "

Taliban Takeover in Afghanistan ›

Latest Updates

Updated 

Sept. 9, 2021, 5:59 p.m. ETSept. 9, 2021

Sept. 9, 2021

• Afghanistan is at risk of ‘universal poverty’ by mid-2022, U.N. warns.
• An uncertain future awaits Afghans who flee to Pakistan.
• Scenes of relief replace images of desperation at the Kabul airport.

The field was part of a new commemorative site called Patriot Park, which is about a third of a mile down the road from the Flight 93 National Memorial. That memorial, run by the National Park Service, sits on a quiet expanse of more than 2,000 acres, and includes a museum, a wall engraved with the names of the 40 passengers and crew members and an open meadow of thistle and goldenrod. It opened to the public 10 years ago and, with the 2018 dedication of a tower of wind chimes called the Tower of Voices, is now complete.'

From University of Memphis Law Review:

Invisible Inequality: The Two Americas of Military Sacrifice

DOUGLAS L. KRINER* & FRANCIS X. SHEN**

INTRODUCTION ......................................................................546

THE INVISIBLE INEQUALITY OF MILITARY SACRIFICE............548 

A. Scholarship on Veterans Affairs.......................................549 

B. What Does the American Public Know About

Inequalities in Military Sacrifice? ...................................554 

III. THETWOAMERICASOFMILITARYSACRIFICE......................556

A. Poorer Areas of the Country Bear Greater
War Sacrifice ...................................................................557

B. The Causes of the Casualty Gap ......................................564 

INEQUALITY AND THE VETERAN’S BRAIN .............................569 

A. The Wounds of War ..........................................................569 

B. Social Determinants of Veterans’ Brain Health...............574 

V. THEPOLITICALCOSTSOFCASUALTYINEQUALITY................578

IV.

A. B.

C. D.

*

Is Inequality in Military Sacrifice Different From
Other Forms of Inequality? .............................................579

How Americans React to Information About
Inequality .........................................................................580 

The Invisible Politics of Non-Fatal Casualties ................583

Non-fatal Casualties and Public Support for
the War in Afghanistan ....................................................587

Associate Professor, Boston University Department of Political Sci-

McKnight Land-Grant Professor & Associate Professor of Law, Uni- versity of Minnesota; Director, Shen Neurolaw Lab; Executive Director of Edu- cation and Outreach, MacArthur Foundation Research Network on Law and Neuroscience.

ence. **

*** We thank Ryan Pesch and Morgan Carlson for excellent research assistance, and we thank members of the University of Chicago Public Law Workshop for helpful feedback. Support for this research was provided by Bos- ton University and the University of Minnesota Law School. Shen notes that his work is Ad Maiorem Dei Gloriam.

545

546 The University of Memphis Law Review Vol. 46

VI. DISCUSSION...........................................................................590

A. Why Don’t We Want To Talk About the Inequality of Military Service?..............................................................591

B. Should We Care About Inequality in Military
Sacrifice? .........................................................................596

C. Are Veterans Already Receiving the Care
They Need?.......................................................................600

D. Are Courts Likely to Intervene?.......................................609

E. Fostering a National Debate on Inequality in Sacrifice ..611 VII. CONCLUSION .........................................................................612

Abstract: Through a series of empirical investigations— including analysis of over 500,000 American combat casualties from World War II through Iraq and Afghanistan—we show in this Article that there is growing socioeconomic inequality in military sacrifice and that the relative invisibility of this inequality has ma- jor political ramifications. Today, unlike in World War II, the Americans who die or are wounded in war are disproportionately coming from poorer parts of the country. We argue that these Two Americas of military sacrifice constitute invisible inequality be- cause the issue is routinely overlooked by scholars, policymakers, and the public. We then use seven original surveys of American public opinion to uncover a variety of social, legal, and political consequences of this inequality. With Congress unlikely to act, and courts unwilling to intervene, we argue that the best path for- ward is to generate a renewed public debate over inequality in military sacrifice. To this end, we show empirically that such a conversation could transform public opinion. Ignoring inequality in military sacrifice is both morally comforting and politically ben- eficial. But it is at odds with empirical reality, and, most im- portantly, with our American ideals of shared sacrifice.

I. INTRODUCTION

The central issues of the 2016 presidential campaign are coming into focus: addressing economic inequality at home, and defining America’s military strategy abroad amidst new terrorist threats. These issues will be debated thousands of times.

Yet these many debates will overlook a connection between the two: America’s economic downturn means that increasingly it is not the governing class, but the working class that dispropor-

2016 Invisible Inequality 547

tionately sends soldiers to fight and bears the burden of physical and mental war wounds.

For members of both parties it is politically convenient to overlook these Two Americas of military sacrifice. But in this Article, we show that ignoring this invisible inequality has not made it go away.

Through a series of empirical investigations—including analysis of over 500,000 American combat casualties from World War II through Afghanistan, combined with seven unique surveys of American public opinion—we reveal that, even more than pre- vious wars, Iraq and Afghanistan have been working class wars. This inequality is normatively troubling, but it also has significant social and political consequences. Inequality in pre-service oppor- tunities can translate into inequality in post-service health out- comes. For example, soldiers returning home to fewer resources may be at greater risk of developing mental disorders. We also show how non-fatal casualties remain largely invisible in the polit- ical sphere. This invisibility has artificially inflated public support for wars and for the leaders who wage them.

The emergence of these Two Americas and its consequenc- es are cause for concern. Yet equally problematic is the failure of legislatures and courts to even acknowledge, let alone address, these disparities. With neither legislatures nor courts likely to act without prompting, we argue that the most viable response is a renewed public debate over inequality in military sacrifice. We present experimental data suggesting that such a conversation could have real policy consequences.

The Article proceeds in five parts. We begin in Part II with a discussion of the legal literature on veterans’ affairs, finding that it typically makes little mention of inequality in sacrifice. We also report original survey data in which we find that the American public is not aware of the distribution of war sacrifice.

The omissions by scholars, and the views of the American public, would both be warranted if no inequality exists. Thus, we turn in Part III to the empirical question: who is dying, and who is returning wounded, in America’s wars? The answer is stark: more than in any conflict since World War II, today’s human costs are being borne by America’s working class.

In Part IV, we examine the social and political consequenc- es of this invisible inequality. Focusing on mental health, we dis- cuss how soldiers returning home to weaker social support struc-

548 The University of Memphis Law Review Vol. 46

tures are at greater risk of developing mental disorders. Then, we turn to the political sphere in Part V. We show with new data that Americans view inequality in military sacrifice differently from other forms of inequality, and we also show that if knowledge of this inequality is communicated, it reduces public support for war. Yet, we also find that wounded in action casualties are less visible and politically salient than fatalities. Thus, even when American soldiers return home wounded in large numbers, it is not likely to slow American politicians from sending new recruits into the battle zone again.

In light of these many empirical findings uncovered in the Article, we conclude in Part VI with a discussion of why this ine- quality remains invisible and what to do about it. We call for a national conversation on the human costs of war. The invisible inequality of military sacrifice should be invisible no more.

II. THE INVISIBLE INEQUALITY OF MILITARY SACRIFICE

American society is increasingly concerned with economic inequality as seen in the Occupy movement of 2011 and the rheto- ric and policy proposals of the 2016 presidential candidates. Legal scholarship is filled both with articles concerning inequality and redistribution1 and with commentaries on the need to aid military veterans.2

Given this attention to inequality on the one hand, and mili- tary veterans on the other, one would think that inequality in mili- tary sacrifice is a well-known and researched fact—that is not the case. As we will show in this Part, scholars do not routinely ad-

1. Gillian Lester, Can Joe the Plumber Support Redistribution? Law, Social Preferences, and Sustainable Policy Design, 64 TAX L. REV. 313, 315 (2011) (“[A] great deal of legal scholarship concerns itself with questions of inequality and redistribution . . . .”).

2. See, e.g., Benjamin Pomerance, Fighting on Too Many Fronts: Con- cerns Facing Elderly Veterans in Navigating the United States Department of Veterans Affairs Benefits System, 37 HAMLINE L. REV. 19, 22 (2014); Thomas J. Reed, Parallel Lines Never Meet: Why the Military Disability Retirement and Veterans Affairs Department Claim Adjudication Systems Are a Failure, 19 WIDENER L.J. 57, 59–60 (2009); Robert R. Gagan, Thank Veterans; Help Veter- ans, 88 WIS. LAW., Feb. 2015, at 7; Tara Shockley, Veterans Legal Initiative: Showing Appreciation for Service Through Legal Assistance, 50 HOUS. LAW., May/June 2013, at 16.

2016 Invisible Inequality 549 dress the issue (Section A), and nearly half of all Americans are

not even aware that such inequalities exist (Section B).

A. Scholarship on Veterans Affairs

There is a robust legal literature on veterans’ affairs.3 In 1989, the Department of Veterans Affairs (“VA”) was created as a cabinet-level position.4 At about that same time, VA administra- tive decisions became subject to judicial review in the United States Court of Veterans Appeals (“CVA”) through enactment of the Veterans’ Judicial Review Act in 1988.5 This Act, as well as the Veterans Claims Assistance Act passed in 2000 (“VCAA”), aimed to improve the process for administering veterans claims.6 Litigation and commentary on the VCAA is now extensive.7

3. Notably, in 2009 the Veterans Law Review was first published by the U.S. Department of Veterans Affairs. 1 VETERANS L. REV. (2009) http://www.bva.va.gov/VLR.asp. Excellent work such as THE ATTORNEY’S GUIDE TO DEFENDING VETERANS IN CRIMINAL COURT 199 (Brockton D. Hunter & Ryan C. Else eds., 2014), has highlighted a range of legal issues specific to veterans.

4. Department of Veterans Affairs Act, Pub. L. No. 100-527, § 2, 102 Stat. 2635 (1988).

5. Veterans’ Judicial Review Act, Pub. L. No. 100-687, 102 Stat. 4105 (1988).

6. 38 U.S.C. §§ 5100, 5102–5103A, 5106–5107, 5126 (2013); Terrence T. Griffin & Thomas D. Jones, The Veterans Claims Assistance Act of 2000: Ten Years Later, 3 VETERANS L. REV. 284, 284 (2011).

7. Veterans’ Judicial Review Act, Pub. L. No. 100-687. For scholarship on this topic, see, e.g., Michael P. Allen, Due Process and the American Veter- an: What the Constitution Can Tell Us About the Veterans’ Benefits System, 80 U. CIN. L. REV. 501 (2011); Laurence R. Helfer, The Politics of Judicial Struc- ture: Creating the United States Court of Veterans Appeals, 25 CONN. L. REV. 155 (1992); James T. O’Reilly, Burying Caesar: Replacement of the Veterans Appeals Process is Needed to Provide Fairness to Claimants, 53 ADMIN. L. REV. 223 (2001); Jeffrey Parker, Two Perspectives on Legal Authority Within The Department of Veterans Affairs Adjudication, 1 VETERANS L. REV. 208 (2009); James D. Ridgway, The Veterans’ Judicial Review Act Twenty Years Later: Confronting the New Complexities of the Veterans Benefits System, 66 N.Y.U. ANN. SURV. AM. L. 251, 252 (2010) (“[T]he VA adjudication system today is very different from the one that existed prior to the VJRA, but the adju- dication system has not necessarily improved.”); Rory E. Riley, Simplify, Simpli- fy, Simplify-an Analysis of Two Decades of Judicial Review in the Veterans’ Benefits Adjudication System, 113 W. VA. L. REV. 67 (2010).

550 The University of Memphis Law Review Vol. 46

There has been much written on the Court of Appeals for Veterans Claims (“CAVC”).8 Scholars have also examined the emergence and efficacy of specialized veterans’ treatment courts,9 as well as training for family court judges on how to work with those returning from combat.10

Both in and beyond law reviews, extensive scholarly atten- tion has been given to the physical and mental health of veterans.11 This has led to action on issues such as traumatic brain injury and mental health treatment for veterans, in part due to an influential

8. See VETERANS APPEALS GUIDEBOOK: REPRESENTING VETERANS IN THE U.S. COURT OF APPEALS FOR VETERANS CLAIMS (Ronald L. Smith ed., 2013).

9. See, e.g., Alana Frederick, Veterans Treatment Courts: Analysis and Recommendations, 38 L. & PSYCHOL. REV. 211 (2014); Michael Daly Hawkins, Coming Home: Accommodating the Special Needs of Military Veterans to the Criminal Justice System, 7 OHIO ST. J. CRIM. L. 563 (2010); Mark A. McCor- mick-Goodhart, Leaving No Veteran Behind: Policies and Perspectives on Combat Trauma, Veterans Courts, and the Rehabilitative Approach to Criminal Behavior, 117 PENN ST. L. REV. 895 (2013); Tabatha Renz, Veterans Treatment Court: A Hand Up Rather Than Lock Up, 17 RICH. J. L. & PUB. INT. 697 (2014); Robert T. Russell, Veterans Treatment Court: A Proactive Approach, 35 NEW ENG. J. CRIM. & CIV. CONFINEMENT 357 (2009); John Furman Wall, IV, The Veterans Treatment Court Program Act: South Carolina’s Opportunity to Pro- vide Services for Those Who Have Served, 65 S.C. L. REV. 879 (2014); Rosendo Garza Jr., Note, “The Soldier Bears the Deepest Wounds and Scars of War”: Mobilizing Connecticut to Implement a Veterans Treatment Court, 46 CONN. L. REV. 1937 (2014); C. Philip Nichols, Jr., Veterans Courts: A New Concept for Maryland, MD. B.J. March–Apr. 2014, at 42.

10. See Evan R. Seamone, Educating Family Court Judges on the Front Lines of Combat Readjustment: Toward the Formulation and Delivery of a Core Curriculum on Military Family Issues, 52 FAM. CT. REV. 458 (2014).

11. See, e.g., THE PRAEGER HANDBOOK OF VETERANS’ HEALTH: HISTORY, CHALLENGES, ISSUES, AND DEVELOPMENTS (Thomas W. Miller ed., 2012); Kathy Cerminara & Olympia Duhart, Introduction: Wounds of War: Meeting the Needs of Active-Duty Military Personnel and Veterans with Post- Traumatic Stress Disorder, 37 NOVA L. REV. 439 (2013); Charles W. Hoge et al., Combat Duty in Iraq and Afghanistan, Mental Health Problems, and Barri- ers to Care, 351 NEW ENG. J. MED. 13, 13 (2004); Matthew Jacupcak et al., Anger, Hostility, and Aggression Among Iraq and Afghanistan War Veterans Reporting PTSD and Subthreshold PTSD, 20 J. TRAUMATIC STRESS 945 (2007).

2016 Invisible Inequality 551

RAND study that labeled traumatic brain injury and related mental disorders as “invisible wounds” of the wars.12

Paul Rieckhoff, the Director of Iraq and Afghanistan Vet- erans of America, commented in 2009 that “[t]hey call brain trau- ma ‘the invisible wound’; well, there’s nothing less visible than being uncounted.”13 It is certainly the case that today, more so than even a decade ago, military leaders are publicly recognizing brain trauma, including mental injuries, as true wounds of war.

In January 2010 in Washington D.C., before an audience at a Suicide Prevention Conference, then United States Secretary of Veterans Affairs Eric Shinseki announced a sobering statistic: “[o]n average, eighteen Veterans commit suicide each day.”14 It has become common knowledge that a record percentage of Amer- ica’s returning combat veterans are committing suicide;15 that

12. TERRI TANIELIAN ET AL., INVISIBLE WOUNDS OF WAR: SUMMARY AND RECOMMENDATIONS FOR ADDRESSING PSYCHOLOGICAL AND COGNITIVE INJURIES (2008), http://www.rand.org/content/dam/rand/pubs/monographs/2008/ RAND_MG720.1.pdf. On a longer historical trajectory, we also now recognize more wounds of war than we did in previous eras. DAVID A. GERBER, FINDING DISABLED VETERANS IN HISTORY, in DISABLED VETERANS IN HISTORY 3 (David A. Gerber ed., 2000) (“[T]he visibility of . . . disabled veterans . . . has increased in this [20th] century . . . .”); see also Brockton D. Hunter, Echoes of War: Combat Trauma, Criminal Behavior, and How We Can Do Better This Time Around, in THE ATTORNEY’S GUIDE TO DEFENDING VETERANS IN CRIMINAL COURT 2–3 (Brockton D. Hunter & Ryan C. Else eds., 2014).

13. Paul Solotaroff, The Iraq War’s Invisible Wounded, MEN’S J., Oct. 6, 2009, http://www.mensjournal.com/magazine/print-view/the-iraq-wars-invisible -wounded-20131108. Rieckhoff went on to say:

The VA and DOD paid no attention to this problem the first four years of the war, and now there are all these guys in need of treatment with no clear way to get it. A lot don’t even know they have head trauma, or are too afraid to admit it. They think if they raise their hand for help, it’s the end of their service career.

Id.

Eric K. Shinseki, Sec’y, Dep’t of Veterans Affairs, Remarks at the

14.
Suicide Prevention Conference (Jan. 11, 2010), http://www.va.gov/opa/ speeches/2010/10_0111hold.asp.

15. Olympia Duhart, Soldier Suicides and Outcrit Jurisprudence: An Anti-Subordination Analysis, 44 CREIGHTON L. REV. 883, 884 (2011); Lindsay I. McCarl, “To Have No Yesterday”: The Rise of Suicide Rates in the Military and Among Veterans, 46 CREIGHTON L. REV. 393, 395 (2013) (“The number of sui- cides has increased significantly since the beginning of the Iraq and Afghanistan

552 The University of Memphis Law Review Vol. 46

many are being diagnosed with Post-Traumatic Stress Disorder (“PTSD”);16 and that the Department of Veterans Affairs needs to improve its provision of mental health services.17

In these and other ways, much that was once “invisible” is now in the public square for open debate. But our review of many related contemporary literatures finds very little discussion of the systemic economic inequality undergirding the experiences of re- turning veterans.18

There is some occasional, often tangential, mention of the issue.19 For instance, psychologist Robert Klein suggested in the early 1980s, in the wake of Vietnam, that inadequate VA care has contributed to the creation of “a whole new underclass of alienat-

wars, despite the implementation of VA-sponsored programs to help stave off deaths of our war-beaten warriors.”).

16. See ERIN P. FINLEY, FIELDS OF COMBAT: UNDERSTANDING PTSD AMONG VETERANS OF IRAQ AND AFGHANISTAN (2011); Tiffany M. Chapman, Leave No Soldier Behind: Ensuring Access to Health Care for PTSD-Afflicted Veterans, 204 MIL. L. REV. 1, 8 (2010).

17. See, e.g., TIMOTHY A. KELLY, HEALING THE BROKEN MIND: TRANSFORMING AMERICA’S FAILED MENTAL HEALTH SYSTEM (2009).

18. There is legal literature on the inequalities experienced by African- American veterans. See, e.g., BENJAMIN FLEURY-STEINER, DISPOSABLEHEROES: THE BETRAYAL OF AFRICAN AMERICAN VETERANS (2012); Benjamin Fleury-Steiner et al., From the Battlefield to the War on Drugs: Lessons from the Lives of Marginalized African American Military Veterans, 6 ALB. GOV’T L. REV. 464 (2013). For discussion of race and military participation, see Amy C. Lutz, Race-Ethnicity and Immigration Status in the U.S. Military, in LIFE- COURSE PERSPECTIVES ON MILITARY SERVICE 68 (Janet M. Wilmoth & Andrew S. London eds., 2013). However, we do not find that there is a racial casualty gap. See DOUGLAS L. KRINER & FRANCIS X. SHEN, THE CASUALTY GAP: THE CAUSES AND CONSEQUENCES OF AMERICAN WARTIME INEQUALITIES (2010). For a review of the historical literature on inequality and military sacrifice and service, see our discussion in Chapter 2. Id. There have been a few longer treatments of military service and class. KATHY ROTH-DOUQUET & FRANK SCHAEFFER, AWOL: THE UNEXCUSED ABSENCE OF AMERICA’S UPPER CLASSES FROM THE MILITARY SERVICE—AND HOW IT HURTS OUR COUNTRY (2006).

19. At the national level, New York Congressman Charles Rangel, who has supported reinstituting the military draft, has complained that “[i]t’s just not fair that the people we ask to fight our wars are people who join the military because of economic conditions.” David M. Halbfinger & Steven A. Holmes, Military Mirrors Working-Class America, N.Y. TIMES, Mar. 30, 2003, http://www.nytimes.com/2003/03/30/international/worldspecial/30DEMO.html.

2016 Invisible Inequality 553

ed, unemployed citizens.”20 More recently, in the wake of the war in Iraq, physician Ronald Glasser has observed that the “Vietnam divide between those who serve and those who are served has be- come the foundation of the volunteer force” and that we have a “country that would send off Reserve and National Guard troops without becoming engaged nor demand an accounting of wartime policy, goals, and purposes.”21 He suggests that the Iraq War was “fought without any sense of pretense of communal sacrifice”22 and that “privilege spells the difference between living and dying, between being crippled or blind for the rest of your life. Today once again, survival is a matter of class.”23

On a similar theme, veteran and Rhodes Scholar Josiah Bunting III penned a 2004 essay in The American Scholar entitled “Class Warfare.”24 Bunting observed “[t]he diminishing numbers of war dead disclose another phenomenon: the withdrawal of . . . the privileged intellectual and professional and commercial classes, and their novitiates and children, from the active military service of our country.”25 He argued that this trend “is dangerous, it is unworthy, it is wrong.”26

In 2014, United States Navy Lieutenant Commander Mat- thew Ivey identified the challenges of a shrinking all-volunteer force.27 Ivey similarly recognized that “despite the disproportion- ate burdens suffered by the current all-volunteer military, very few Americans have called for a change to the current way of staffing the military.”28 Journalist Jorge Mariscal penned a 2007 online essay describing what he called the “poverty draft”:

Exactly who will have to fight and die in those wars will be determined by economic class. In order to

20. ROBERT KLEIN, WOUNDED MEN, BROKEN PROMISES 199 (1981).

21. RONALD J. GLASSER, WOUNDED 122–23 (2006).

22. Id. at 128.

23. Id.

24. Josiah Bunting III, Class Warfare, THE AMERICAN SCHOLAR (Dec. 1,

2004), https://theamericanscholar.org/class-warfare/#.VQmPTOGm0UM.

25. Id.

26. Id.

27. Matthew Ivey, The Broken Promises of an All-Volunteer Military, 86

TEMP. L. REV. 525 (2014).
28. Id. at 528. For further history, see id. at 530–40.

554

The University of Memphis Law Review Vol. 46

accomplish their goals, the recruiters and politicians will exploit the hopes and dreams of mostly well- intentioned youth from humble origins who are looking for a way to contribute to a society that has lost its moral compass. As they did in Vietnam and again in Iraq, young women and men will serve their country. But how well will their country have served them?29

Yet apart from these and similar exceptions, contemporary aca- demic scholarship has not seriously explored inequality in military sacrifice. Indeed, as recently as 2013 a sociologist writing on the topic speculated (incorrectly) that “socioeconomic disadvantage has been associated with war-related mortality, although the same may not be true of the current wars.”30 If scholarship is generally not concerned with this inequality, can we say the same thing about the American public? We turn now to that question.

B. What Does the American Public Know About Inequalities in Military Sacrifice?

Since 2007 we have been conducting studies in which we ask the American public about how wartime casualties are distrib- uted across the country.31 And we have regularly found that a large segment of the population mistakenly believes there is shared sacrifice.

In a nationally representative sample of Americans polled in 2011, we asked each respondent: “Thinking about the American soldiers who have died fighting in Iraq and Afghanistan, what parts of the United States do you think they are coming from?”32 Re-

29. Jorge Mariscal, The Poverty Draft, SOJOURNERS, June 2007, at 32, 35.

30. Alair MacLean, A Matter of Life and Death, in LIFE-COURSE PERSPECTIVES ON MILITARY SERVICE, supra note 18, at 213.

31. See KRINER & SHEN, supra note 18, at 92–103; Douglas L. Kriner & Francis X. Shen, Conscription, Inequality, and Partisan Support for War, J. CONFLICT RESOL. (forthcoming 2016).

32. We embedded our survey question in an Opinion Research Corpora- tion CARAVAN omnibus poll administered in May 2011. CARAVAN is a twice-weekly telephone survey that employs a random-digit dialing (“RDD”) methodology to ensure a nationally representative sample of 1,000 adult Ameri- cans. This survey, which produced a sample of 1,010 respondents from the

2016 Invisible Inequality 555

spondents were then asked to choose one of the following options (or say “I don’t know”): (i) More casualties are coming from poorer, less educated parts of the country; (ii) More casualties are coming from richer, more educated parts of the country; or (iii) There is not a significant difference in the share of casualties com- ing from rich/high-education and poor/low-education parts of the country.

The results, presented in Figure 1, are striking and paint a portrait of an evenly divided public. Just under half those surveyed (45%) believe that the country is equally sharing military sacrifice. This is roughly the same percentage as those who correctly believe that there is inequality.33

Closer analysis of our data suggests that rather than basing their answers on knowledge of the facts, many Americans simply adopt the position of their preferred political party. The strongest predictor of a respondent’s answer to this question is his or her partisan affiliation. A clear majority of Republicans, 57%, believe that shared sacrifice exists. Only 35% of Republicans believe that there is inequality in casualties. By contrast, for Democrats, the numbers are reversed, with 30% believing that shared sacrifice exists and 60% believing there is inequality. Given the lack of reliable information concerning casualty inequality in the public sphere, many Americans simply draw on their partisan priors to inform their guesses.

continental United States, was conducted from May 19–22, 2011. National news outlets such as CNN rely on Opinion Research Corporation because of its reputation for reliably providing truly nationally representative samples. CARAVAN data is also regularly used in political science research requiring nationally representative samples. The averages reported from this survey in Figure 1 and discussed in the text are the unweighted averages. We also ran a similar experiment with subjects recruited from Mechanical Turk in 2015. Even in this Turk sample, which is younger, more highly educated, and more liberal than a nationally representative sample, a sizeable portion of respondents be- lieved there is not a casualty gap. Thirty-three percent of respondents said casu- alties come from rich and poor places equally. We also find evidence of a parti- san gap. Republicans are more likely to believe a casualty gap does not exist than Democrats or independents.

33. In addition, eight percent chose the “I don’t know” option and three percent responded that more casualties were coming from richer/more-educated parts of the country.

556 The University of Memphis Law Review Vol. 46

The result is that much of the public believes—as we will show, mistakenly—that American localities are sharing the human sacrifice of war equally.

Figure 1. Percentage of Americans Who Believe in Inequality versus Shared Sacrifice in War Casualties

What to Notice in Figure 1: We asked a nationally representative sample of Americans whether American soldiers who have died fighting in Iraq and Afghanistan were coming equally from rich and poor parts of the country (“shared sacrifice”), or more from poor parts of the country (“inequality”). The data presented in Figure 1 is striking because it shows that nearly half of all Americans believe there is shared sacrifice, even though the empirical data suggest otherwise.

III. THE TWO AMERICAS OF MILITARY SACRIFICE

In Part I we established that inequality in military sacrifice is rarely discussed in scholarship and often not acknowledged by the public. Given the widespread public uncertainty over how mil- itary sacrifice is shared across the country, in this Part we turn to

2016 Invisible Inequality 557

the actual data and ask: When America goes to war, who fights the battles, who dies, and who returns wounded?34

We began to ask these questions in 2004, and in this Article we make a novel extension of the work, as for the first time we examine non-fatal casualties, including casualties from the conflict in Afghanistan.

We find that both fatal and non-fatal casualties in Ameri- ca’s wars have come from parts of the country that are lower on the socioeconomic ladder.35 This Part explains why these Two Americas of military sacrifice have emerged, and how this distri- bution is more unequal than in past wars. Details of the statistical analyses are presented in the Appendix.

A. Poorer Areas of the Country Bear Greater War Sacrifice

While concerns about inequality and military sacrifice have periodically arisen since America’s founding, empirical research to determine the existence of such inequalities and changes in them over time has progressed haphazardly since World War II.36 Prior

34. The discussion in this Part builds on previous discussion in KRINER & SHEN, supra note 18.

35. The analyses discussed in this Part show strong evidence of a socio- economic casualty gap between rich and poor, and high and low educationcommunities. We do not have access to individual-level data, but it seems most plausible that there is also an individual-level gap, i.e. soldiers coming from poorer backgrounds are disproportionately bearing the costs. While plausible, we acknowledge that we cannot conclude definitively from the community-level casualty data alone that poorer individuals or individuals with lower levels of education are dying at higher rates than individuals with greater socio-economic opportunities. To do so would be to commit what social scientists call an error of “ecological inference.” From aggregate-level data alone, we cannot make inferences about processes at the level of individuals. To address this, we have previously carried out a series of additional analyses—all suggesting that in fact there is an individual level gap. See KRINER & SHEN, supra note 18.

36. JOHN CHAMBERS, DRAFTEES OR VOLUNTEERS: A DOCUMENTARY HISTORY OF THE DEBATE OVER MILITARY CONSCRIPTION IN THE UNITEDSTATES, 1787–1973 (1975). Studies focusing on military enlistments, recruits, and personnel have been conducted by sociologists, historians, think tanks, and the popular press. CHRISTIAN G. APPY, WORKING-CLASS WAR: AMERICAN COMBAT SOLDIERS AND VIETNAM (1993); SUE BERRYMAN, WHO SERVES? THE PERSISTENT MYTH OF THE UNDERCLASS ARMY (1988); ROTH-DOUQUET & SCHAEFFER, supra note 18; Wilson, infra note 40; Tim Kane, Who Bears the Burden? Demographic Characteristics of U.S. Military Recruits Before and

558 The University of Memphis Law Review Vol. 46

analyses have varied significantly in approach and scope. Some have found strong, if limited, evidence that socioeconomically dis- advantaged communities have borne a disproportionate share of the nation’s casualties.37 Others have yielded mixed results and uneven empirical support for assertions of a casualty gap.38 Still others have produced no systematic evidence of a socioeconomic casualty gap.39 Reviewing this motley state of affairs, sociologist Thomas C. Wilson observed that the variance may be “due in large

After 9/11, THE HERITAGE FOUNDATION (Nov. 7, 2005), http://www.heritage.org/research/reports/2005/11/who-bears-the-burden- demographic-characteristics-of-us-military-recruits-before-and-after-9-11; Mili- tary Recruitment 2010, NATIONAL PRIORITIES PROJECT (June 30, 2011), https://www.nationalpriorities.org/analysis/2011/military-recruitment-2010. Indeed, since 1974 the federal government has mandated an annual Department of Defense report on social representation in the military. NESE F. DEBRUYNE & ANNE LELAND, CONG. RESEARCH SERV., RL32492, AMERICAN WAR AND MILITARY OPERATIONS CASUALTIES, LISTS AND STATISTICS (2015), https://www.fas.org/sgp/crs/natsec/RL32492.pdf. See generally JEANETTE KEITH, RICH MAN’S WAR, POOR MAN’S FIGHT: RACE, CLASS, AND POWER IN THE RURAL SOUTH DURING THE FIRST WORLD WAR (2004); EUGENE C. MURDOCK, PATRIOTISM LIMITED: 1862-1865 (1967); DAVID WILLIAMS, TERESA CRISP WILLIAMS & DAVID CARLSON, PLAIN FOLK IN A RICH MAN’S WAR: CLASS AND DISSENT IN CONFEDERATE GEORGIA (2002); DAVID WILLIAMS, RICH MAN’S WAR: CLASS, CASTE, AND CONFEDERATE DEFEAT IN THE LOWER CHATTAHOOCHEE VALLEY (1998); Tyler Anbinder, Which Poor Man’s Fight? Immigrants and the Federal Conscription of 1863, 52 CIV. WAR HIST. 344 (2006).

37. Emily Buzzell & Samuel Preston, Mortality of American Troops in Iraq, 33 POPULATION AND DEV. REV. 555, 562 (2007); Albert J. Mayer & Thomas Ford Hoult, Social Stratification and Combat Survival, 34 SOC. FORCES 155, 155 (1955); M. Zeitlin, K. G. Lutterman & J. W. Russell, Death in Vi- etnam: Class, Poverty, and the Risks of War, 3 POL. & SOC’Y 313, 313 (1973).

38. Gilbert Badillo & G. David Curry, The Social Incidence of Vietnam Casualties, 2 ARMED FORCES & SOC’Y 397, 401 (1976); Arnold Barnett et al.,America’s Vietnam Casualties: Victims of a Class War?, 40 OPERATIONS RES. 856, 857 (1992); John Willis, Variations in State Casualty Rates in World War II and the Vietnam War, 22 SOC. PROBLEMS 558, 558 (1975).

39. CHARLES C. MOSKOS & JOHN S. BUTLER, ALL THAT WE CAN BE: BLACK LEADERSHIP AND RACIAL INTEGRATION THE ARMY WAY (1996); Brian Gifford, Combat Casualties and Race: What Can We Learn from the 2003–2004 Iraq Conflict?, 31 ARMED FORCES AND SOC’Y 201, 201 (2005); Janet Schaefer & Marjorie Allen, Class and Regional Selection in Fatal Casualties in the First 18–23 Months of World War II, 23 SOC. FORCES 165, 165 (1944).

2016 Invisible Inequality 559

part to the cumulative effect of methodological inconsistencies across studies and methodological flaws within them.”40

Amidst these competing findings, we launched a new re- search project in 2004. The project represents the most compre- hensive investigation to date of inequality and military casualties. We published some of these findings in a 2010 book, The Casualty Gap. Here, we extend our earlier analysis both to include the war in Afghanistan and to examine inequality in non-fatal casualties. Our analysis in this Article considers non-fatal casualty data through December 26, 2009, and fatal casualty data through July 4, 2011.41

The Department of Defense does not release data on the so- cioeconomic status of individual soldiers who have died or been wounded in America’s wars. As a result, we cannot directly ob- serve whether poorer Americans with fewer educational opportuni- ties are disproportionately dying in or returning home wounded from the nation’s wars. However, we can examine the communi- ties from which our nation’s wartime casualties hail. This allows us to examine whether communities at the bottom of the socioeco- nomic ladder have sustained higher casualty rates than communi- ties at the top. Such casualty gaps between rich and poor commu- nities are of great importance. First, as we will discuss in more detail shortly, soldiers returning home to socioeconomically disad- vantaged communities may enjoy fewer and weaker support struc- tures, which can exacerbate their reintegration into civilian life. Second, most Americans view and assess war through the lens of their local community’s experiences with it. Casualty inequality

40. Thomas C. Wilson, Vietnam-era Military Service: A Test of the Class-Bias Theory, 21 ARMED FORCES & SOC’Y 461, 464 (1995); John Modell & Timothy Hagerty, The Social Impact of War, 17 ANNUAL REV. OF SOC. 205, 219–20 (1991) (reaching a similar conclusion). Some previous studies analyze casualties from only a single state or region of the country. Other researchers focus more narrowly on a specific age cohort or restrict their analyses to short periods of time. Moreover, the measures used for socioeconomic status change from study to study, and many analyses examine only one potential explanation for inequalities in casualties, while failing to control for other possibilities. Fi- nally, only a handful of analyses examine more than one conflict at a time. Wil- son, supra, at 464.

41. We use the term “wounded” and “non-fatal casualty” to mean the same thing in this Article. See infra Section V.C. and note 193 for additional discussion of the challenges of precisely defining these terms.

560 The University of Memphis Law Review Vol. 46

between rich and poor communities insures that many Americans view the same war very differently; some see its human costs di- rectly, while others are largely insulated from such costs.

We examined the relationship between the socioeconomic status of a community and its share of war sacrifice by looking at the relationship between county-level (or where available place- level) data on socioeconomic variables (such as income and educa- tion) and county-level casualty rates.42 To gain historical perspec- tive, we examined World War II, Korea, Vietnam, Iraq, and Af- ghanistan, the five wars on which data is available.43

Figure 2 illustrates the basic finding: The data show that while sacrifice was shared equally in World War II, beginning with the war in Korea, significant income gaps emerged. In raw, infla- tion-adjusted dollar terms, this income casualty gap increased over time from a $5,500 gap in Korea, to an $8,200 gap in Vietnam, and now to more than an $11,000 gap in Iraq and Afghanistan. More robust statistical analysis, controlling for a host of possibly con- founding variables, confirms this basic finding.44

42. For details of the statistical analysis, see the Appendix as well as KRINER & SHEN, supra note 18, at 14. For a discussion of the ecological infer- ence problem and efforts to overcome it, see particularly id. at 40–47. For Iraq and Afghanistan, we were able to use census “place” level data, a geographical unit even smaller than the county. Place refers to “Census Designated Place.” The U.S. Census Bureau defines a “Census Designated Place” as a place “delin- eated to provide data for settled concentrations of population that are identifiable by name but are not legally incorporated under the laws of the state in which they are located.” UNITED STATES CENSUS BUREAU: GEOGRAPHIC TERMS AND CONCEPTS – PLACE, https://www.census.gov/geo/reference/gtc/gtc_place.html (last updated Dec. 6, 2012).

43. For wars prior to World War II the requisite data is not available. See KRINER & SHEN, supra note 18, at 14.

44. See discussion in the Appendix.

2016 Invisible Inequality 561

Figure 2. Two Americas of Military Sacrifice: Difference in Median Family Income Levels Between High-Casualty Com- munities and Low-Casualty Communities

 

What to Notice in Figure 2: Figure 2 illustrates that since World War II, communities with higher casualty rates have had lower incomes than com- munities with lower casualty rates. To generate Figure 2, we divided all of the communities for each war into two groups: the first includes all com- munities whose casualty rates place them in the top quarter of the casualty distribution; the second group comprises all other communities.45 From census data, we then calculated the average median family income for both groups. To provide a constant metric, we adjusted the income data from previous periods to reflect their value in year 2000 dollars.

45. For World War II, Korea, and Vietnam, this analysis is at the county level. For Iraq and Afghanistan, it is at the place level. Because the total num- ber of casualties in the wars in Iraq and Afghanistan is comparatively small, we used a slightly different coding scheme to identify high and low casualty com- munities in Iraq and Afghanistan. High casualty communities include the 700 census places that have suffered casualty rates of higher than 9.31 fatal casual- ties per 10,000 male residents. This represents the top twenty-five percent of all communities that suffered at least one casualty in the Iraq War. The low casual- ty communities in Iraq and Afghanistan are the census places that had not yet suffered a casualty in either war—more than eighty-five percent of all census places.

562 The University of Memphis Law Review Vol. 46

The discussion and data presented thus far pertain to fatal casualties. But what of those soldiers who are wounded? Do they, too, hail disproportionately from poorer parts of the country? This question is more salient than ever given that more than seven Americans were wounded in Iraq and Afghanistan for every ser- vice member killed, a ratio much greater than that observed in ear- lier wars (see Figure 4 below).

To investigate, we made a Freedom of Information Act (“FOIA”) request to the Department of Defense (“DoD”) for the number of wounded soldiers for each county in the United States.46 Indicative of the challenge of studying wounded-in-action, one of our requests to the DoD was (we thought) a straightforward defini- tional query. We requested “the definitions used by the DoD to determine whether a soldier is considered ‘wounded’.” After all, how can one interpret the data on number of wounded if we don’t know what counts as “wounded”?

In response, the Department of Defense, in conjunction with the Defense Manpower Data Center (“DMDC”) wrote that we “cannot provide any definitions used by DoD to determine whether a soldier is considered wounded because this is a medical judg- ment.”47 We appealed but were not provided additional clarifying information. Thus, we proceeded with the analysis with a best- guess, but no clear certainty, on how the DoD actually determines if a soldier is considered wounded.48

46. The request, FOIA 10-F-0284, was initially made in writing on No- vember 19, 2009. We requested “1) The number of wounded soldiers from Operation Iraqi Freedom, by month, by branch, and by county; 2) The number of wounded soldiers from Operation Enduring Freedom, by month, by branch, and by county; and 3) The definitions used by the DOD to determine whether a soldier is considered ‘wounded.’” See Letter from Paul J. Jacobsmeyer, Chief, Dep’t of Def. Freedom of Info. Office, to author (May 18, 2010) (on file with authors). We received a partially responsive reply with a data file on May 18, 2010. Id.

47. Letter from Paul J. Jacobsmeyer, Chief, Dep’t of Def. Freedom of Info. Office, to author (May 18, 2010) (on file with authors).

48. In our appeal letter we asked:
Is it accurate to conclude then that the Department of Defense Manpower Data Center is wholly unaware of how its data on wounded soldiers is defined? For instance, the DMDC does not know whether its statistics include soldiers diagnosed with

2016 Invisible Inequality 563

We ran a similar analysis to that described above to see if the rate of wounded soldiers was correlated with the county’s soci- oeconomic indicators. We found that once again there was an un- equal relationship: Counties with lower education and income lev- els had higher percentages of their residents wounded in Iraq and Afghanistan.49

This relationship is seen vividly in Figure 3, which plots how non-fatal casualties are distributed across the country. We divided communities into deciles based on their median family income. Thus, the 10% of Americans living in the poorest com- munities are in the first income decile, and so on. If military sacri- fice was evenly shared, then each decile would account for 10% of the soldiers wounded in action in Iraq and Afghanistan. On the left hand side, the dark gray shaded bars above the 10% line indicate that communities in the lower deciles generally shouldered more of the burden.50 On the right hand side, by contrast, the light gray shaded bars are all below the 10% line suggest that those commu- nities in the higher income brackets have not experienced as many non-fatal casualties.51 Put slightly differently, the nation’s poorest communities (those in the lowest three income deciles) have suf- fered fifty percent more non-fatal casualties than the nation’s wealthiest communities (those in the top three income deciles).

49.

sprained wrists and twisted ankles? And the DMDC doesn’t know whether its data includes soldiers who are diagnosed with depression? Based on the FOIA response . . . [our] con- clusion is that DMDC does not know the answers to these questions because they are “medical judgments.”

Comparable data was not available in previous conflicts.

50.
data suggesting that the lowest income and lowest education communities do not have as many residents who meet the military’s requisite qualifications. See an extended discussion in DOUGLAS L, KRINER & FRANCIS X. SHEN, THE CASUALTY GAP: THE CAUSES AND CONSEQUENCES OF AMERICAN WARTIME INEQUALITIES online app. B (2010), http://www.casualtygap.com/KrinerShen_ TheCasualtyGap_OnlineAppendixB.pdf.

51. By matching the home of record information for each wounded sol- dier provided by the DOD with information on community median income lev- els from the U.S. Census, we found that communities in the bottom three income deciles suffered 4,573 casualties, while those in the top three deciles suffered only 2,995. See KRINER & SHEN, supra note 18.

The exception in the lowest decile is consistent with military service

564 The University of Memphis Law Review Vol. 46

Figure 3. Two Americas of Military Sacrifice: Distribution of Non-Fatal Casualties in Iraq and Afghanistan by Income Decile, Above and Below Equal Distribution (ten percent)

What to Notice in Figure 3: If one divides the nation into ten deciles by in- come, an equal distribution of casualties would produce ten percent casual- ties in each decile. But Figure 3 shows that there is inequality in the distri- bution of casualties: the five richest deciles (the light gray bars on the right) are all below-average, while the poorer deciles (the dark gray bars on the left) tend to take on above-average casualties. See text for discussion of da- ta analysis that produced the Figure.

B. The Causes of the Casualty Gap

The evidence presented above makes clear that there are Two Americas of military sacrifice. Working class America is sacrificing at a higher rate than affluent America. Why is this the case?

There are two mechanisms in play, both of which have ex- planatory power: differential selection into the armed forces (“the selection mechanism”) and then differential occupational assign- ment within the military (“the sorting mechanism”). We have

2016 Invisible Inequality 565

shown in previous work that both the selection and sorting mecha- nisms affect the unequal outcomes.52

One of the most straightforward explanations for inequality in wartime death is inequality in who serves in the military. For more than fifty years, an extensive literature at the crossroads of sociology, history, economics, and political science has investigat- ed military manpower policies and changes in them over time.53 Today, a small percentage of Americans serve in the military.54 This has led to a civil-military gap along a number of dimen- sions.55

Men and women join the military for many reasons; for many, patriotism and a desire to serve are undoubtedly key factors. Yet an extensive literature also documents the critical importance of economic incentives in spurring enlistments throughout Ameri- can history.56 At the aggregate level, a number of studies have

52. KRINER & SHEN, supra note 18, at ch. 3. Although we can only es- tablish a casualty gap between rich and poor communities, the most likely ex- planation for this gap is that a parallel inequality exists at the individual level. The selection and sorting mechanisms described here provide a logic for why individuals from socioeconomically disadvantaged backgrounds are more likely to find themselves on the front lines of America’s wars.

53. See, e.g., BERRYMAN, supra note 36; JOHN CHAMBERS, supra note 36; GEORGE Q. FLYNN, THE DRAFT: 1940–1973 (1993); PETER KINDSVATTER, AMERICAN SOLDIERS: GROUND COMBAT IN THE WORLD WARS, KOREA AND VIETNAM (2003); MORRIS JANOWITZ, THE PROFESSIONAL SOLDIER, A SOCIAL AND POLITICAL PORTRAIT (1960); CHARLES C. MOSKOS, THE AMERICAN ENLISTED MAN: THE RANK AND FILE IN TODAY’S MILITARY (1970); THE ALL- VOLUNTEER FORCE: THIRTY YEARS OF SERVICE (Barbara A. Bicksler, Curtis L. Gilroy & John T. Warner eds., 2004); NAT’L RESEARCH COUNCIL, COMM. ON THE YOUTH POP. AND MILITARY RECRUIT., ATTITUDES, APTITUDES ANDASPIRATIONS OF AMERICAN YOUTH: IMPLICATIONS FOR MILITARY RECRUITMENT 219 (Paul Sackett & Anne Mavor, eds. 2003); Stuart Altman & Alan Fechter, The Supply of Military Personnel in the Absence of a Draft, 57 AM. ECON. REV. 19 (1967); Peter Karsten, Consent and the American Soldier: Theory Versus Reality, 12 PARAMETERS 42 (1982).

54. Ivey, supra note 27, at 557 (“[O]nly one-half of one percent of Amer- icans served in the military at any given time during the past decade.”).

55. PEW RESEARCH CTR., THE MILITARY-CIVILIAN GAP: WAR AND SACRIFICE IN THE POST-9/11 ERA 2 (2011), http://www.pewsocial trends.org/files/2011/10/veterans-report.pdf.

56. For an example of the nuance that exists within this rich literature, see a recent study, Todd Woodruff, Ryan Kelty & David R. Segal, Propensity to

566 The University of Memphis Law Review Vol. 46

demonstrated strong correlations between the health of the econo- my and patterns in military enlistments. For example, in a 1994 RAND study of the factors correlated with the successful recruit- ment of high quality enlistments from 1978 to 1993, two of the factors with the greatest influence on the number of high quality recruits obtained by the Army were the youth unemployment rate and the rate of military pay growth relative to the civilian sector.57

This linkage continues to the present day. Reflecting on the surge in military enlistments during the economic troubles of 2008, which followed immediately on the heels of two of the most difficult recruiting years in recent memory in 2006 and 2007, Un- dersecretary for Personnel and Readiness David S.C. Chu readily acknowledged the faltering economy’s role in boosting volunteer- ing: “We do benefit when things look less positive in civil socie- ty.”58 Other analyses of enlistment decisions at the individual level demonstrate, logically, that the young men and women most likely to volunteer are those for whom the occupational and educational benefits that the military affords are most appealing compared to their options in the civilian labor market.59 As summarized by mil-

Serve and Motivation to Enlist among American Combat Soldiers, 32 ARMED FORCES & SOC’Y 353 (2006), suggesting institutional incentives are particularly important for the thirty percent new recruits who are “high-propensity” youth, i.e. those that had long planned on joining the military. Id. at 358. By contrast, among the seventy percent of recruits who had not thought seriously about en- listing while in high school, occupational and economic incentives were particu- larly important. Id. at 363.

57. The size of the military recruiting budget also had a strong impact on recruiting trends. BETH ASCH & BRUCE ORVIS, RECENT RECRUITING TRENDS AND THEIR IMPLICATIONS: PRELIMINARY ANALYSIS AND RECOMMENDATIONS 21 (1994), http://www.rand.org/content/dam/rand/pubs/monograph_reports/2005/ MR549.pdf; see also Altman & Fechter, supra note 53, at 19–20; Charles Brown, Military Enlistments: What Can We Learn from Geographic Variation, 75 AM. ECON. REV. 228 (1985); John Warner & Beth Asch, The Record and Prospects of the All-Volunteer Military in the United States, 15 J. ECON. PERSP. 169 (2001).

58. William H. McMichael, Economic Bust Creates Recruiting Boom, ARMY TIMES, Dec. 30, 2008.

59. For example, in their analysis of military volunteerism from 1973 to 1978, sociologists Morris Janowitz and Charles Moskos found that college- educated men, who enjoyed great advantages in the civilian labor market, were significantly under-represented in the armed forces. While almost thirty percent of the military-aged male population had some college education in 1977, only

2016 Invisible Inequality 567

itary historian Peter Karsten, “Most volunteers, today and for the past 200 years, joined the service in order to gain economic re- wards, social mobility, or skills needed later in civilian life.”60

Recognizing these economic incentives, it is not surprising that Army recruits have come disproportionately from parts of the country that are lower on the socioeconomic scale.61 The military has struggled in some years to meet its enlistment quotas, and as a result it has drawn on recruits with lower qualifications.62 During

five percent of new Army enlistees did. In 1964, more than seventeen percent of young men drafted into the service had some college education. Morris Jan- owitz & Charles C. Moskos Jr., Five Years of the All-Volunteer Force: 1973– 1978, 5 ARMED FORCES & SOC’Y 171, 194–95 (1979).

60. Karsten, supra note 53, at 43. This is not to say that the relative im- portance of economic incentives has not changed over time. For example, sur- veying the history of 20th century manpower policy, Charles Moskos identifies three eras—the modern (1900–1945), the late modern (1945–1990), and the postmodern (1990–)—and he argues that across these periods the military has become increasingly viewed more through an “occupational” and less through an “institutional” lens. If correct, this trend could also contribute to the emer- gence of the socio-economic casualty gaps we observed in the Korean and Vi- etnam wars and the widening of these gaps in the Iraq War. CHARLES C. MOSKOS, JOHN ALLEN WILLIAMS & DAVID R. SEGAL, THE POSTMODERN MILITARY 14 (Charles C. Moskos, John Allen Williams & David R. Segal eds., 2000).

61. For example, Kriner and Shen's ZIP-code level analysis of Army recruiting data shows that the high income communities were significantly un- der-represented in Army recruiting. KRINER & SHEN, supra note 18, at 65. For an analysis of more recent Army recruiting data, see 2011 DOD POPULATION REPRESENTATION IN THE MILITARY SERVICES REPORT, app. tbl. B-41, http://prhome.defense.gov/portals/52/Documents/POPREP/poprep2011/appendi xb/appendixb.pdf. An individual-level analysis of military recruits from the 1990s found that young people from high income families were significantly less likely to enlist in the military, all else being equal, than their peers from lower socioeconomic backgrounds. Amy Lutz, Who Joins the Military?: A Look at Race, Class, and Immigration Status, 36 J. POL. & MIL. SOC. 167 (2008). However, a recent analysis of individual-level data from the post-9/11 era finds little evidence of socioeconomic differences. Andrea Asoni et al., Rich Man's War, Poor Man's Fight? Technological Change, Tactical Developments and the Demographic Composition of the American Military, (Feb. 6, 2016), http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2728702.

62. Ivey, supra note 27, at 550 (“[P]revious minimum academic and moral standards for enlistment were now being waived in order to make up for the recruiting shortfall.”).

568 The University of Memphis Law Review Vol. 46

some of the relevant periods of recruitment for the wars in Iraq and Afghanistan, the Army lowered its recruitment standards63 and offered larger financial incentives.64 Such selection mechanisms have the potential to create casualty gaps. But this is only part of the story.

The vast majority of those who serve do not die in combat, and those who do die are not a random sample of the military pop- ulation as a whole. As Colonel Samuel Hays wrote in Army Mag- azine in 1967, “In many ways the differences in sacrifice between those who are called to the service and those who are excused are less drastic than the differences which result from different as- signments in the Services . . . no one could find much equity be- tween pounding a typewriter in the Pentagon and carrying the M16 rifle in the jungles of Vietnam.”65

Occupational assignment is far from random. Through a series of tests, the military assesses each new soldier’s aptitudes and pre-existing skill sets and, on the basis of this information and additional evaluations, it assigns each soldier to the tasks thought to be best-suited to his or her personal skills and to the military’s needs.66 If soldiers assigned to positions with high risks of combat exposure differ systematically from soldiers assigned to occupa- tions with lower levels of combat risk, occupational assignment, too, has the potential to generate a casualty gap.67

When one examines the difference between enlisted and of- ficer casualty rates, we find strong evidence that occupational sort- ing leads to casualty gaps.68 Casualty rates for the infantry and the enlisted ranks are more inversely related to community wealth and education than are non-infantry and officer casualty rates.69

Because lower-skilled recruits are more likely to come from less advantaged communities, and because they are subse-

63. Id.

64. Id.

65. Samuel H. Hays, Military Conscription in a Democratic Society,

ARMY MAG., Feb. 1967, at 31, reprinted in CHAMBERS, supra note 36.
66. The process of occupational assignment varies across service branch- es and varies across individuals as well. For instance, some recruits are given

the option to select an occupational field.

67. See KRINER & SHEN, supra note 18, at 67–72.

68. Id.

69. Id.

2016 Invisible Inequality 569

quently more likely to be assigned to occupations with greater combat risks than are recruits with higher skills, the occupational assignment mechanism may produce a casualty gap, even if the military as a whole were representative of the civilian population.70 Similarly, because the enlisted ranks come disproportionately from lower income/education communities, and because enlisted sol- diers are more likely, on average, to see front line combat than are officers, assignment by rank also explains why a casualty gap can develop even if the military’s overall demographics may roughly mirror society.

In sum: We believe there is extremely strong evidence that poorer parts of America are bearing a greater share of the human costs of war. In the next two Parts we explore some of the social and political consequences of these Two Americas of military sac- rifice.

IV. INEQUALITY AND THE VETERAN’S BRAIN

A. The Wounds of War

Historical comparisons plainly illustrate the increasing prominence of combat wounds in recent conflicts. For example, the ratio of soldiers killed versus soldiers wounded in Iraq is strik- ing in comparison to earlier conflicts. While the wounded/killed ratio was 1.65 in World War II, 1.9 in Korea, and 2.6 in Vietnam, in Iraq the ratio through March 2014 was 7.2, and in Afghanistan the ratio was 7.6.71 Thus, when compared with Vietnam and Ko- rea, the ratio of wounded to killed soldiers in Iraq/Afghanistan is

70. Moreover, there are differences in the service branches. The Army, for instance, which accounts for a majority of the casualties, is not as representa- tive of the population as a whole. OFFICE OF THE ASSISTANT SECRETARY OF DEFENSE FOR PERSONNEL AND READINESS, POPULATION REPRESENTATION IN THE MILITARY SERVICES (2011), http://prhome.defense.gov/portals/52/ Documents/POPREP/poprep2011/appendixb/b_41.html.

71. Ratios for World War II, Korea, and Vietnam were calculated using data from the Department of Defense. NESE F. DEBRUYNE & ANNE LELAND, CONG. RESEARCH SERVICE, RL32492, AMERICAN WAR AND MILITARY OPERATIONS CASUALTIES, LISTS AND STATISTICS (2015), https://www.fas.org/sgp/crs/natsec/RL32492.pdf. The Korean War ratio utilized the figure of 54,246 for worldwide military deaths. Id. at 9. Ratios for Iraq and Afghanistan were calculated using data from IRAQ COALITION CASUALTY COUNT, http://icasualties.org (last visited Mar. 10, 2016).

570 The University of Memphis Law Review Vol. 46

more than two and a half times larger. When compared to World War II, the ratio in Iraq/Afghanistan is more than four times as large.

The ratio of wounded to killed reflects advances in military medicine. For instance, the Army now utilizes Forward Surgical Teams (“FSTs”),72 and they have proven effective at reducing cas- ualties because of their rapid response.73 But saving lives means that more soldiers are surviving with catastrophic injuries. As one nurse working in Baghdad remarked, “We’re saving severely in- jured people, legs, eyes, parts of brains. These injuries are horrif- ic.”74 And as one of the medical surgeons remarked about the re- covery these soldiers can expect, “[w]e can save you, [but] [y]ou might not be what you were.”75

Veterans are often returning with a variety of symptoms. This is so much the case that caretakers now use the term “poly- trauma” to describe veterans with “multiple and complex physical

72. Timothy C. Counihan & Paul D. Danielson, The 912th Forward Sur- gical Team In Operation New Dawn: Employment Of The Forward Surgical Team During Troop Withdrawal Under Combat Conditions, 177 MIL. MED. 1267, 1269 (2012) (“FST have been used widely since the onset of the Global War on Terror in both Iraq and Afghanistan.”).

73. GLASSER, supra note 21, at 41 (“The efficiency of the new [FST] system, as well as the resulting survival rates, are quite extraordinary . . . .”).

74.

Id. at 47.
Th[e] newest type of casualty coming out of this our newest war involves severe and devastating multiple traumas: severe head injuries, vision and hearing loss, nerve damage, bone fractures, contaminated wounds, severed limbs, transected spinal cords along with emotional and behavioral problems. And the numbers of patients with these multiple awful wounds increase every month of the war.

Id. at 143.
75. Robert Carroll, quoted in Karl Vick, The Lasting Wounds of War:

Roadside Bombs Have Devastated Troops and Doctors Who Treat Them, WASH. POST (Apr. 27, 2004), http://www.washingtonpost.com/wp- dyn/articles/A44839-2004Apr26_2.html; see also Ann M. Hendricks & Jomana H. Amara, Current Veteran Demographics and Implications for Veterans’ Health Care, in RETURNING WARS’ WOUNDED, INJURED, AND ILL: A REFERENCE HANDBOOK 17 (Nathan D. Ainspan & Walter E. Penk eds., 2008) (“Battlefield medicine, evacuation procedures, and battlefield medical support services have evolved tremendously leading to greater survival rates for troops.”).

2016 Invisible Inequality 571

and or psychological injuries.”76 The most common trio of symp- toms are Traumatic Brain Injury (“TBI”), PTSD, and pain.77 Many veterans also have substance abuse challenges.78 Moreover, these substance use problems are comorbid with other psychiatric ill- nesses.79

Figure 4. Ratio of Killed in Action to Wounded in Action, Revolutionary War through Afghanistan

What to Notice in Figure 4: The graph illustrates how the United States’ proportion of Killed in Action (“KIA”) to Wounded in Action (“WIA”) sol- diers has increased substantially in the wars in Iraq and Afghanistan. This is due in large part to major advances in medical technology on the battle- field, which now allows many soldiers to avoid death from injuries that in earlier wars would have been fatal.

76. John Linck & Jared Benge, The Psychological Assessment of Veter- ans with History of Polytrauma, in PSYCHOLOGICAL ASSESSMENT OF VETERANS 404 (Shane S. Bush ed., 2014).

77. Id. at 409.

78. Dominick Dephilippis et al., Psychological Assessment of Veterans

with Substance Use Disorders, in PSYCHOLOGICAL ASSESSMENT OF VETERANS, supra note 76, at 177.

79. Id. at 185.

572 The University of Memphis Law Review Vol. 46

In addition to these types of brain injuries, the RAND Cor- poration’s 2008 study of the psychological consequences of Opera- tion Enduring Freedom (“OEF”) and Operation Iraqi Freedom (“OIF”) makes clear that our current wars have taken an immense toll on returning soldiers’ mental health.80 The data plotted in Fig- ure 4 undercount the actual number of soldiers wounded in action because the data (from the DOD) used to generate Figure 4 do not include “mental” injuries as wounds.81

80. Traumatic Brain Injury has been center stage since the start of the wars in Afghanistan and Iraq. Mild TBI is diagnosed when a person has:

a traumatically-induced physiological disruption of brain func- tion, as manifested by at least one of the following: (1) any period of loss of consciousness; (2) any loss of memory for events immediately before or after the event; (3) any alteration in mental state at the time of the accident (eg, feeling dazed, disoriented, or confused); and (4) focal neurological deficit or deficits that may or may not have been transient; but where the severity of the injury does not exceed the following: loss of consciousness of approximately 30 minutes or less; after 30 minutes, an initial Glasgow Coma Scale (GCS) of 13–15; and posttraumatic amnesia (PTA) not greater than 24 hours.

AM. CONG. OF REHABILITATION MED., DEFINITION OF MILD TRAUMATIC BRAIN INJURY (1993), https://www.acrm.org/wp-content/uploads/pdf/TBIDef_English_ 10-10.pdf. The Veterans Administration has made a number of changes in its services for veterans experiencing brain trauma, such as the creation of more robust rehab units. Kurt Samson, VA Reinforces Stateside Rehab Units for Iraq Blast Injuries, NEUROLOGY TODAY, Apr. 2006, at 18. The Government Ac- countability Office (“GAO”) found in 2008 that the VA has improved its screen- ing for Mild TBI, though it also suggested a number of policy reforms designed to make assessment more effective. U.S. GOV’T ACCOUNTABILITY OFF., GAO- 08-276, VA HEALTH CARE: MILD TRAUMATIC BRAIN INJURY SCREENING AND EV ALUA TION IMPLEMENTED FOR OEF/OIF VETERANS, BUT CHALLENGESREMAIN 5 (2008), http://www.gao.gov/assets/280/271988.pdf. We do not sug- gest, however, that it is only in recent wars that mental injuries have been preva- lent. It has been observed well before that “[t]he power of the battlefield to break men can never be overstated.” David Marlowe, The Human Dimension of Battle and Combat Breakdown, in MILITARY PSYCHIATRY: A COMPARATIVE PERSPECTIVE 7 (Richard A. Gabriel ed., 1986).

81. Moreover, it also excludes civilian casualties. An MIT project on the human costs of war tracks civilian casualties and a natural extension of our ar- gument would be that civilian casualty counts should consider distributions across the socioeconomic spectrum. See Iraq: the Human Cost, MIT CTR. FOR INT’L STUDIES, http://web.mit.edu/humancostiraq/ (last visited Mar. 10, 2016).

2016 Invisible Inequality 573

Before moving on to discuss inequality and non-fatal casu- alties, we should be clear that we are not arguing that soldiers are somehow treated differently when coming off the battlefield with injuries. When a soldier is injured in Afghanistan or Iraq, he or she is typically transported to Landstuhl Regional Medical Center in Germany.82 At Landstuhl, soldiers are treated for a variety of injuries and are screened for traumatic brain injury.83 Soldiers are also evacuated to Landstuhl for psychiatric evaluations.84 We have seen no studies suggesting differential treatment at these stages.

While in the Department of Defense’s hospital system care may be state-of-the-art,85 upon leaving active duty, the burden of care falls upon the medical services provided by the U.S. Depart- ment of Veterans Affairs.86

82. Harold L. Timboe & Richard R. Timboe, America’s Wounded Warri- ors, GPSOLO, Jan./Feb. 2005, at 26, 27, http://www.americanbar.org/newsletter/ publications/gp_solo_magazine_home/gp_solo_magazine_index/amer wounded.html (“The Landstuhl Regional Medical Center in Germany receives casualties from Europe, Africa, and the Middle East.”); see also Joachim J. Te- nuta, From The Battlefields to the States: The Road To Recovery. The Role of Landstuhl Regional Medical Center in US Military Casualty Care, 14 J. AM. ACAD. ORTHOPAEDIC SURGEONS S45, S45–S47 (2006); Raymond Fang et al., Critical Care at Landstuhl Regional Medical Center, 36 CRITICAL CARE MED. S383 (2008); Brent A. Johnson, Operation Iraqi Freedom: The Landstuhl Re- gional Medical Center Experience, 44 J. FOOT & ANKLE SURGERY 177 (2005).

83. Kenneth E. Dempsey et al., Landstuhl Regional Medical Center: Traumatic Brain Injury Screening Program, 16 J. TRAUMA NURSING 6 (2009).

84. James R. Rundell, Demographics of and Diagnoses in Operation Enduring Freedom and Operation Iraqi Freedom Personnel Who Were Psychi- atrically Evacuated From The Theater Of Operations, 28 GEN. HOSP. PSYCHIATRY 352, 352 (2006) (“Between the beginning of Operation Enduring Freedom (OEF; US military operations in Afghanistan) and Operation Iraqi Freedom (OIF; US military operations in Iraq) and July 2004, 12,480 medical, surgical and psychiatric evacuees from the theaters of operation were sent to the Landstuhl Regional Medical Center (LRMC) in Germany. The LRMC received virtually all evacuees leaving OEF and OIF during the reference period. One thousand two hundred sixty-four of those patients (10.1%) were sent to be man- aged primarily by psychiatry.”).

85. GLASSER, supra note 21, at 48.

86. Id. at 49.

574 The University of Memphis Law Review Vol. 46 B. Social Determinants of Veterans’ Brain Health

Recognizing that the number of wounded soldiers is large, what can be said about the relationship between the health of these soldiers and the socioeconomic inequality identified in Part II? To start, we note that veterans have experienced unemployment, hous- ing difficulties, and mental health afflictions, including a high sui- cide rate.87 But not all veterans have experienced this equally.88 So the question becomes: What factors differentiate those veterans who experience mental injury from those who do not?

Decades of research on the social determinants of health have made clear this conclusion: “Life chances differ greatly de- pending on where people are born and raised.”89 Inequality in so-

87. McCarl, supra note 15, at 398; Randi Jensen, Military Suicidality and Principles to Consider in Prevention, in WAR TRAUMA AND ITS WAKE: EXPANDING THE CIRCLE OF HEALING 156 (Raymond M. Scurfield & Katherine T. Platoni eds., 2013). Some of this may have been exacerbated by the use of the stop-loss policy, which “permits the retention of enlisted service members past the end of active obligated service (EAOS) as initially agreed upon in their enlistment contracts,” and was used extensively in the wars in Iraq. Ivey, supra note 27, at 548. Court challenges to the military use of the stop-loss policy were unsuccessful, even for the National Guard. Stop-loss was halted in 2011. Id. (“Although the Stop-Loss policy is not purely an invention of recent conflicts, the last decade marks the first time the military has used the policy so broad- ly.”).

88. Not explored in this Article are the differential experiences, and re- sulting treatment, of females as compared to males. We do not have a sufficient knowledge base yet. See Shirley M. Glynn, Impact on Family and Friends, in RETURNING WARS’ WOUNDED, INJURED, AND ILL, supra note 75, at 175 (“Little information exists now on the special needs of female warfighters and their ca- reers, and the research field is in its infancy.”). However, there is some evi- dence that female soldiers have a greater incidence of psychiatric disorder. See, e.g., Stephanie Booth-Kewley et al., Predictors of Psychiatric Disorders in Combat Veterans, 13 BMC PSYCHIATRY 130 (2013); Olympia Duhart, PTSD and Women Warriors: Causes, Controls and a Congressional Cure, 18 CARDOZO J.L. & GENDER 327 (2012); Rundell, supra note 84; Michelle Wilmot, Women Warriors: From Making Milestones in the Military to Community Rein- tegration, in WAR TRAUMA AND ITS WAKE, supra note 87, at 83–85.

89. Michael Marmot et al., Closing the Gap in a Generation: Health Equity Through Action On The Social Determinants Of Health, 372 LANCET1661, 1661 (2008).

2016 Invisible Inequality 575

cioeconomic resources is related to inequality in health out- comes.90

This insight, while often used in conversations about cross- national heath policy, also has implications for U.S. social policy on veterans care. In short: Inequality in pre-service opportunity is likely to lead to inequality in post-service options, and thus to an unequal distribution of health outcomes. We posit that one (though certainly not the only) reason we see some soldiers devel- op mental health issues, while others are able to resume life more seamlessly, is a lack of social supports.91

We do not have direct evidence for this claim, and indeed we are not aware of a publicly available dataset that would allow us to answer it. But even without direct evidence, the circumstan- tial case seems to us very strong.92

To start, Naval Health Research Center researcher Stepha- nie Booth-Kewley conducted a longitudinal study of mental health outcomes in 1,113 Marines who served in Iraq in OIF or Afghani- stan in OEF.93 Eighteen percent of the Marines in the study re- ceived a psychiatric diagnosis during the observation period.94 Common diagnoses were anxiety disorders, mood disorders, sub- stance abuse disorders, adjustment disorders, and PTSD.95 Includ- ed in the analysis was the Marine’s education level, and the re- searchers found that even when controlling for combat exposure, more education was associated with a lower incidence of a psychi-

90. We also know, from emerging research, that socioeconomic status affects brain development. Daniel A. Hackman & Martha J. Farah, Socioeco- nomic Status and the Developing Brain, 13 TRENDS COGNITIVE SCI. 65 (2009).

91. Sociologist Alair MacLean has recognized that “veterans may have worse health than non-veterans not because they served in the military, but be- cause they came from socioeonomically disadvantaged backgrounds and have fewer years of schooling.” MacLean, supra note 30, at 207.

92. Our observations here are necessarily preliminary, as “[r]esearchers are only just beginning to thoroughly explore the long-term consequences of physical and psychological wounds for service members’ family relationships.” Elaine Willerton et al., Introduction: Military Families under Stress: What We Know and What We Need to Know, in RISK AND RESILIENCE IN U.S. MILITARY FAMILIES 13 (Shelley MacDermid Wadsworth & David S. Riggs eds., 2010).

93. Booth-Kewley et al., supra note 88, at 130.

94. Id. at 135.

95. Id.

576 The University of Memphis Law Review Vol. 46

atric disorder.96 Less education was associated with a greater inci- dence of PTSD,97 anxiety disorders, and adjustment disorders.98

There is also evidence that, on average, the demographics of those treated for psychiatric disorders differs from those of the general fighting force. Psychiatrist James Rundell’s study of sol- diers treated for psychiatric disorders at Landstuhl found that en- listed soldiers were significantly more likely to be treated for psy- chiatric disorders than were officers.99 Because enlisted soldiers are more likely to be from the lower rungs of the socioeconomic ladder, this suggests an uneven burden.100

Moreover, there is also evidence that PTSD varies by rank in the military: Officers are significantly less likely than enlisted personnel to develop PTSD.101 Researchers suggest that this may

96. Id. at 136 (“[S]ignificant predictors of mental disorder diagnosis in- cluded education level (more education was protective), marital status (being divorced was associated with the highest risk), total number of career combat deployments (multiple deployments was associated with the highest risk), com- bat exposure (moderate exposure was associated with the highest risk), and posi- tive deployment experiences (a moderate level was the most protective).”).

97. Id. at 138 (“Four additional variables had marginally significant (p < .10) associations with PTSD: education, unit cohesion, positive deployment experiences, and total number of career combat deployments.”).

98. Id. (“Other predictors of anxiety disorders (p < .05) included female gender, education, number of combat deployments, and deployment stressors.”).

99. Rundell, supra note 84, at 354. (“When compared with all returned OEF and OIF veterans (N=213,150), psychiatric evacuees were more likely to be . . . enlisted (96% vs. 86%; P<.001) . . . .”).

100. Rundell also found that National Guard soldiers were more likely to be evacuated for psychiatric disorders than active-duty military. Id. (“When compared with all returned OEF and OIF veterans (N=213,150), psychiatric evacuees were more likely to be . . . National Guard/Reserve, as opposed to active-duty military (34% vs. 26%; P<.001).”).

101. Jessica Wolfe et al., Course and Predictors of Posttraumatic Stress Disorder Among Gulf War Veterans: A Prospective Analysis, 67 J. OFCONSULTING & CLINICAL PSYCHOL. 520, 526 (“[O]fficer rank could serve a protective function. Officers in our study showed negligible levels of PTSD, suggesting that nonofficer rank was influential in the exacerbation of PTSD over time. This protective effect could relate to any number of factors, including differences in entrance-level characteristics, differences in training and prepara- tion, or variations in actual wartime exposure. Although we cannot know for certain, it is possible that these vulnerabilities do not appear until certain contex- tual resources (e.g., the support of the military environment) are withdrawn.”); see also David T. Holmes et al., Preliminary Evidence of Psychological Distress

2016 Invisible Inequality 577

be both because the enlisted soldiers have different entry-level characteristics and because they have fewer supports upon the end of their service commitment.102

In addition, a number of other studies have identified social class as a risk factor for a range of veteran health outcomes:

Soldiers who return to strong support environ- ments may fare better in terms of mental health than peers who lack such supports.103

Social support can help to prevent the onset of PTSD.104

Veterans with stronger social networks are less likely to have PTSD.105

The incidence of depression in veterans is corre- lated with education level and rank.106

Substance abuse may be exacerbated by low- income status.107

Among Reservists in the Persian Gulf War, 186 J. NERVOUS & MENTAL DISEASE 166 (1998).

102. Wolfe et al., supra note 101, at 526.

103. Bradley E. Belsher et al., The Social Context of Post-Trauma Adjust-

ment in Veterans, in THE PRAEGER HANDBOOK OF VETERANS’ HEALTH, supra note 11, at 200.

104.

105. It is not clear, however, if this correlation is causation. It could be that those veterans who develop PTSD cause their friends/families to distance themselves. MacLean, supra note 30, at 217.

106. Anne M. Gadermann et al., Prevalence of DSM-IV Major Depression Among U.S. Military Personnel: Meta-Analysis and Simulation, 177 MIL. MED. 47, 57 (2012) (“Current prevalence among military personnel was estimated to be higher for women than men, young than old, the unmarried than the married, and those with lower than higher rank and education. These correlates are broadly consistent with those found in general population surveys.”).

Chris R. Brewin, Bernice Andrews & John D. Valentine, Meta-

analysis of Risk Factors for Posttraumatic Stress Disorder In Trauma-Exposed

Adults, 68 J. OF CONSULTING & CLINICAL PSYCHOL. 748 (2000); Emily J. Ozer,

Suzanne R. Best, Tami L. Lipsey & Daniel S. Weiss, Predictors of Posttraumat-

ic Stress Disorder and Symptoms in Adults: A Meta-Analysis, 129 PSYCHOL.

BULL. 52 (2003).

578 The University of Memphis Law Review Vol. 46 Beyond the veteran her/himself, there are ripple effects on

family members.

108  If a veteran returns from deployment injured, family members must often pick up the slack, and this compounds the economic crunch. Family members “may be forced to take unpaid leave . . . [or] relinquish jobs or sources of income.”109 As one wife of an injured soldier said, “We

are nobody . . . we don’t have a lot of money.”110

111

In these many ways, the Two Americas of military sacrifice extend well beyond the battlefield.

V. THE POLITICAL COSTS OF CASUALTY INEQUALITY

Inequality in military casualties most directly affects in- jured service members themselves and the families and communi- ties that care for them when they return home. However, the polit- ical ramifications of casualty inequality are also considerable.

In this Part, we examine how greater public awareness of wartime sacrifice, including its significant inequality dimension, may have profound consequences for military policymaking in America. We show that Americans view inequality in military sacrifice as qualitatively different from and more troubling than inequality in other spheres of American life (Section A), informing Americans of inequality changes their support for war (Section B),

107. Dephilippis et al., supra note 78, at 187 (“Substance use can cause and/or be a consequence of psychosocial problems such as low income . . . .”).

108. Shelley A. Riggs & David S. Riggs, Risk and Resilience in Military Families Experiencing Deployment: The Role of the Family Attachment Net- work, 25 J. FAM. PSYCHOL. 675, 681 (2011) (“[I]mportant contextual layers are intergenerational processes, the military unit, and the family’s social and eco- nomic resources.”).

109. Lee Lawrence, Physically Wounded and Injured Warriors and Their Families: The Long Journey Home, in WAR TRAUMA AND ITS WAKE, supra note 87, at 145.

110. Id. at 146.

111. Deborah A. Gibbs et al., Child Maltreatment Within Military Fami-

lies, in RISK AND RESILIENCE IN U.S. MILITARY FAMILIES, supra note 92, at 123.

A family’s economic standing affects how well

they adjust during the soldier’s deployment.

There can even

be ripple effects in terms of child maltreatment, as a stable income (as well as two-parent families and low drug use) reduces the like- lihood of child maltreatment in the face of deployment, while

family stress during deployment can do the opposite.

2016 Invisible Inequality 579

and that non-fatal casualties are less politically salient than fatal casualties (Section C). The combination of these effects suggests that the invisibility of casualty inequality artificially inflates public support for war and the leaders who wage it.

A. Is Inequality in Military Sacrifice Different From Other Forms of Inequality?

Increasing levels of socioeconomic inequality affect virtu- ally every aspect of contemporary American life, including educa- tional opportunity, health outcomes, and exposure to crime. No doubt inequality in casualties is related to these other types of ine- quality. And this raises the question: should we pay special atten- tion to inequality in military sacrifice?

We believe the answer is yes. Americans find inequality in military sacrifice to be particularly troubling because it violates a long-cherished norm of shared martial sacrifice. Indeed, George Washington labeled shared service obligations as a core responsi- bility of democratic citizenship: “Every citizen who enjoys the protection of a free government, owes not only a portion of his property, but even of his personal service to the defense of it.”112 Risking and laying down one’s life for the defense of country is the greatest sacrifice the state can ask of its citizens. As a result, there are strong reasons to believe that Americans will view inequality in military sacrifice as qualitatively different from inequality arising in other realms.

To explore how the public views military service relative to other high risk occupations, we included the following question on an internet-based survey: “Many jobs and careers require sacrific- es of various types. Compared to other jobs and careers that in- volve high risk, do you think that military service is a unique type of career?” More than ninety percent of respondents answered that a job in the military is, indeed, different from other high-risk jobs.113

The vast majority of Americans may agree that military service is different from other forms of high-risk occupations. But is inequality in military sacrifice more normatively troubling than other forms of inequality that are pervasive in contemporary Amer-

112. See KRINER & SHEN, supra note 18, at 4 n.7.

113. Additional details on this experiment are provided in the Appendix.

580 The University of Memphis Law Review Vol. 46

ican society? To explore this question, we conducted a follow-up internet-based survey. First, all participants in the survey were told about inequality in military sacrifice.114 Each participant was then asked: “Do you think it is important to address inequality in mili- tary sacrifice?” Eighty-two percent of the sample replied “Yes.” We then followed up with those who answered yes and asked, “Do you think inequality in military sacrifice is more important to ad- dress than other types of inequality in American life?” Seventy- one percent said yes, it is more important.

The survey data is consistent with the common-sense prac- tices evident in so many aspects of American life. We provide uniformed soldiers with upgraded seats on plane flights; we salute them at sporting events; and we annually celebrate their sacrifices on Memorial Day and Veterans Day. While other occupations also involve risk to physical health, Americans agree that there is some- thing unique about sacrifice as part of the U.S. military.

B. How Americans React to Information About Inequality

One potential mechanism to ameliorate inequality in mili- tary sacrifice is to reduce overall sacrifice: that is, to be more hesi- tant before sending troops into combat. Political scientists have established that “casualty aversion” affects policymaking.115

The theory owes its origin to political theorist Immanuel Kant.116 The crux of his logic focused on how democratic publics would hold their leaders accountable for costly wars. The public must pay both the financial costs of waging war as well as the toll it exacts in blood. As a result, only in the most exigent of circum- stances will democratic citizens support going to war. And by ex-

114. The text provided was: “There is evidence that the American soldiers who are dying in combat and those who are returning home wounded come disproportionately from parts of the country that are lower on the socioeconomic scale. That said, many jobs require sacrifices, and there is socioeconomic ine- quality in many aspects of American society.”

115. See Douglas L. Kriner & Francis X. Shen, Reassessing American Casualty Sensitivity: The Mediating Influence of Inequality, 58 J. CONFLICTRESOL. 1174 (2013); Douglas L. Kriner & Francis X. Shen, How Citizens Re- spond to Combat Casualties: The Differential Impact of Local Casualties on Support for the War in Afghanistan, 76 PUB. OPINION Q. 761 (2012).

116. IMMANUEL KANT, PERPETUAL PEACE AND OTHER ESSAYS 113 (Ted Humphrey trans., 1983) (1795).

2016 Invisible Inequality 581

tension, they will punish at the ballot box leaders who plunge their countries into costly foreign wars. This basic logic is the founda- tion of many arguments for the influential “democratic peace” the- ory.117

Recent history, however, fails to comport with Kant’s com- pelling logic. Repeatedly, the American public has supported the use of military force to achieve a wide array of foreign policy ob- jectives.118 Moreover, while public support for recent wars in Iraq and Afghanistan has fallen as their costs mounted, both wars and the leaders who waged them long enjoyed significant public sup- port, despite costs and casualty figures that far exceeded those promised by politicians in Washington.119 The democratic brake on costly military policies was much weaker than posited.120

117. See generally BRUCE BUENO DE MESQUITA & DAVID LALMAN, WAR AND REASON: DOMESTIC AND INTERNATIONAL IMPERATIVES (1992); JAMES LEE RAY, DEMOCRACY AND INTERNATIONAL CONFLICT: AN EVALUATION OF THE DEMOCRATIC PEACE PROPOSITION (1995); DAN REITER & ALLAN STAM, DEMOCRACIES AT WAR (1998); BRUCE RUSSETT, CONTROLLING THE SWORD: THE DEMOCRATIC GOVERNANCE OF NATIONAL SECURITY (1990); Bruce Bueno de Mesquita, James Morrow, Randolph Siverson & Alastair Smith, An Institu- tional Explanation of the Democratic Peace, 93 AM. POL. SCI. REV. 791 (2003); Zeev Maoz & Bruce Russett, Normative and Structural Causes of Democratic Peace, 1946–1986, 87 AM. POL. SCI. REV. 624 (1993); Clifton T. Morgan & Sally Campbell, Domestic Structure, Decisional Constraints, and War, 35 J. CONFLICT RESOL. 187 (1991).

118. Richard Eichenberg, Victory Has Many Friends: U.S. Public Opinion and the Use of Military Force, 1981-2005, 30 INT’L SECURITY 140, 140–77 (2005).

119. For example, the Congressional Budget Office estimated the Iraq War itself would cost $14 billion and then $8 to $10 billion a month for an unspeci- fied period of time. The Bush administration estimated the war would cost ap- proximately $50 billion, and it fired Larry Lindsay for speculating that the war might cost as much as $200 billion. Seth Cline, The Underestimated Costs, and Price Tag, of the Iraq War, U.S. NEWS AND WORLD REPORT (March 20, 2013), http://www.usnews.com/news/blogs/press-past/2013/03/20/the-underestimated- costs-and-price-tag-of-the-iraq-war; James Fallows, Paying the Costs of Iraq for Decades to Come, THE ATLANTIC (Mar. 29, 2013), http://www.theatlantic. com/politics/archive/2013/03/paying-the-costs-of-iraq-for-decades-to-come/ 274477/. Most contemporary estimates of the Iraq War’s costs are in the tril- lions. JOSEPH STIGLITZ & LINDA BILMES, THE THREE TRILLION DOLLAR WAR: THE TRUE COST OF THE IRAQ CONFLICT x (2008).

120. On casualties and public support for the Iraq War, see Christopher Gelpi, Peter D. Feaver & Jason Reifler, Success Matters: Casualty Sensitivity

582 The University of Memphis Law Review Vol. 46

Would citizens be more reticent to support ongoing wars and to engage in new ones if they were informed of the significant socioeconomic inequality sacrifice that has characterized recent American wars? To answer this question, in previous research we conducted a series of experiments embedded on nationally repre- sentative public opinion surveys.

In the first experiment, conducted in September of 2007, we explored the influence of information about inequality in sacri- fice on popular evaluations of the Iraq War.121 Subjects assigned to our control group were told nothing about inequality in military sacrifice. Subjects in our main treatment group were told that many of America’s more than 3,700 casualties to date in the Iraq War hailed from socioeconomically disadvantaged casualties.122 Four and a half years after the commencement of the Iraq War, most Americans had firmly made up their minds either to support or oppose the conflict. However, we found that even this modest treatment significantly raised opposition to the Iraq War. In our treatment group, 62% of respondents judged the Iraq War a mis- take versus only 56% in the control group, a modest but statistical- ly significant difference.123 If questions of inequality in sacrifice had received sustained attention and national debate, the adverse consequences on support for the Iraq War likely would have been far greater.

In 2009 we conducted a similar experiment on a second na- tionally representative survey to examine the influence of infor- mation about the Two Americas of military sacrifice on Ameri- cans’ willingness to support the use of force in future endeavors. All subjects were told of the number of American service members

and the War in Iraq, 30 INT’L SECURITY 7 (2005/2006); Douglas L. Kriner & Francis X. Shen, Iraq Casualties and the 2006 Senate Elections, 32 LEGIS. STUD. Q. 507, 516–23 (2007); Erik Voeten & Paul Brewer, Public Opinion, the War in Iraq, and Presidential Accountability, 50 J. CONFLICT RESOL. 809 (2006). On the costs of the Iraq War, see STIGLITZ & BILMES, supra note 119.

121. See KRINER & SHEN, supra note 18, at 96–97.

122. In both this and the experiment that follows we included a second

experiment treatment claiming that military sacrifice is shared equally. This treatment produced results substantively similar to those observed in the control. This suggests that most Americans implicitly assume shared sacrifice, unless provided with information explicitly contradicting it.

123. See KRINER & SHEN, supra note 18, at 94–97.

2016 Invisible Inequality 583

who died in World War II, Korea, Vietnam, and Iraq. Those in the control group received no further information. Those in the ine- quality treatment group were told that in most of these wars poor communities have suffered significantly higher casualty rates than rich communities.

Following an established literature in political science that measures casualty sensitivity, we then asked all respondents how many casualties they would be willing to accept for the United States to achieve a range of foreign policy goals: stabilizing a democratic government in Liberia; stopping ethnic cleansing in Darfur; eliminating Iran’s nuclear program; and killing or captur- ing al Qaeda operatives in Somalia.

In each case except the humanitarian intervention (Darfur), we found that Americans informed of casualty inequality in previ- ous wars were significantly less willing to sustain casualties in fu- ture military missions. Moreover, these effects were even stronger among residents of communities that had experienced inequality in military sacrifice firsthand in the form of disproportionately high casualty rates in the Iraq War.124

C. The Invisible Politics of Non-Fatal Casualties

Although the constraint exercised by public opinion on costly military policies is perhaps not as strong as theory suggests, a mass of empirical scholarship confirms that American support for war sours as war costs mount.125 A robust literature has exam- ined the effects of fatal combat casualties on presidential approv- al,126 support for the military campaign,127 and presidential and congressional election results.128

124. See Kriner & Shen, supra note 115, at 1186–89.

125. Following John Mueller’s lead, most studies on the effects of war

casualties have defined casualties as battle deaths. JOHN MUELLER, WAR, PRESIDENTS, AND PUBLIC OPINION (1973).

126. Richard Eichenberg, Richard Stoll & Matthew Lebo, War President: The Approval Ratings of George W. Bush, 50 J. CONFLICT RESOL. 783, 784–89 (2006); Michael Nickelsburg & Helmut Norpoth, Commander-in-Chief or Chief Economist? The President in the Eye of the Public, 19 ELECTORAL STUD. 313 (2000).

127. PETER FEAVER & CHRISTOPHER GELPI, CHOOSING YOUR BATTLES: AMERICAN CIVIL-MILITARY RELATIONS AND THE USE OF FORCE 102–05 (2004); ERIC V. LARSON, CASUALTIES AND CONSENSUS: THE HISTORICAL ROLE OF CASUALTIES IN DOMESTIC SUPPORT FOR U.S. MILITARY OPERATIONS 5–49

584 The University of Memphis Law Review Vol. 46

Similarly, scholarship confirms that presidents and mem- bers of Congress who support costly wars pay a price at the polls.129 Particularly in the smaller scale wars that characterize American military actions since World War II, casualties have been the primary way in which most Americans see a war’s costs.130 However, the literature is almost completely silent on

(1996); Adam J. Berinksy & James N. Druckman, Public Opinion Research And Support For The Iraq War, 71 PUB. OPINION Q. 126, 129–31 (2007); William A. Boettcher III & Michael D. Cobb, Echoes of Vietnam?: Casualty Framing and Public Perceptions of Success and Failure in Iraq, 50 J. CONFLICT RESOL. 831, 848–49 (2006); Scott Gartner & Gary Segura, War, Casualties, and Public Opinion, 42 J. CONFLICT RESOL. 278, 279–81 (1998); Gelpi, Feaver & Reifler, supra note 120.

128. Timothy Cotton, War and American Democracy: Electoral Costs of the Last Five Wars, 30 J. CONFLICT RESOL. 616, 618–25 (1986); Christopher Gelpi, Jason Reifler & Peter Feaver, Iraq the Vote: Retrospective and Prospec- tive Foreign Policy Judgments on Candidate Choice and Casualty Tolerance, 29 POL. BEHAV. 151, 160–66 (2007); Christian Grose & Bruce Oppenheimer, The Iraq War, Partisanship, and Candidate Attributes: Explaining Variation in Par- tisan Swing in the 2006 U.S. House Elections, 32 LEGIS. STUD. Q. 531, 533–36 (2007); David Karol & Edward Miguel, The Electoral Cost of War: Iraq Casu- alties and the 2004 U.S. Presidential Election, 69 J. POL. 633 (2007); Kriner & Shen, supra note 120, at 509–13.

129. Jamie Carson et al., The Impact of National Tides and District-Level Effects on Electoral Outcomes: The U.S. Congressional Elections of 1862-63, 42 AM. J. POL. SCI. 887, 894–98 (2001); Scott Sigmund Gartner & Gary M. Se- gura, All Politics Are Still Local: The Iraq War and the 2006 Midterm Elections, 41 POL. SCI. & POL. 95, 96–98 (2008); Scott Sigmund Gartner, Gary M. Segura & Bethany A. Barratt, War Casualties, Policy Positions, and the Fate of Legis- lators, 53 POL. RES. Q. 467, 469–70 (2004); Christian Grose & Bruce Oppen- heimer, supra note 128, at 533–36; Karol & Miguel, supra note 128, at 633–36; Kriner & Shen, supra note 120, at 509–13; Douglas L. Kriner & Andrew Reeves, The Influence of Federal Spending on Presidential Elections, 106 AM. POL. SCI. REV. 348, 350 (2012).

130. John Aldrich et al., Foreign Policy and the Electoral Connection, 9 ANN. REV. POL. SCI. 477, 481 (2006) (“Combat casualties are important because the willingness to pay the costs of war is one of the central mechanism by which public opinion might affect foreign policy choices.”); Scott S. Gartner, Second- ary Casualty Information: Casualty Uncertainty, Female Casualties, and War- time Support, 25 CONFLICT MGMT. & PEACE SCI. 98, 99–101 (2008).

2016 Invisible Inequality 585

whether increases in non-fatal casualties produce similar dynam- ics.131

We argue that there are strong reasons to expect non-fatal casualties—despite their large numbers and the significant socio- economic inequality they create—to be less politically salient than fatal casualties. First, Americans may simply discount wounds versus deaths as they seek to measure the human costs of war. Second, non-fatal casualties may be less visible than fatal casual- ties.

As the existing political science literature recognizes, the return of a wounded soldier often does not generate the same community response as the return of a deceased soldier. The death of a soldier is typically followed by a well-attended funeral and considerable local media attention.132 The return of a wounded soldier does not usually trigger the same sort of coverage in local media outlets; however, this is a claim subject to further examina- tion since stories in the popular press have appeared in major newspapers and magazines.133 For instance, in 2004 the New York Times Magazine ran a cover story on returning soldiers which gar-

131. An important, if sometimes overlooked, exception is Jeffrey Mil- stein’s study of Vietnam. JEFFREY S. MILSTEIN, DYNAMICS OF THE VIETNAMWAR: A QUANTITATIVE ANALYSIS AND PREDICTIVE COMPUTER SIMULATION 20, 55 (1973), https://kb.osu.edu/dspace/bitstream/handle/1811/24664/1/DYNA MICS_OF_THE_VIETNAM_WAR.pdf. Milstein’s definition included non- fatal casualties: “U.S. casualties are measured by ten times the number of U.S. troops killed in action, plus the number of wounded requiring hospitalization, plus half the number wounded not requiring hospitalization.” Id. at 20. Thus, Milstein was able to conclude from his analysis that “[t]he most significant costs to the American people were the number of American ‘boys’ killed and wound- ed in Vietnam.” Id. at 55. This notable exception aside, however, the field has relied on battle deaths as their measure of casualties. See Karol & Miguel, supra note 128 (examining the localized electoral effects of Iraq War casualties in the 2004 Presidential election).

132. Gartner & Segura, supra note 129, at 95.

133. See, e.g., Solotaroff, supra note 13. Systematic analysis of media

coverage does not yet exist, and indeed such systematic analysis of battle deaths is just now emerging. Scott L. Althaus et al., Uplifting Manhood to Wonderful Heights? Newspaper Reporting of American Combat Deaths from World War One to Gulf War Two, Presentation at Midwest Political Science Association (April 3–6, 2008), http://faculty.las.illinois.edu/salthaus/Publications/uplifting %20manhood%20paper.pdf.

586 The University of Memphis Law Review Vol. 46

nered significant attention.134 Investigative journalism by Wash- ington Post reporters spurred reforms at Walter Reed Army Medi- cal Center.135 And journalist Mark Benjamin won awards for his continued investigative journalism on the return of wounded sol- diers, and the military’s sub-standard treatment of them.136

Moreover, if the returning soldier’s wounds are not physi- cally visible, community members or even family and friends may not know the true extent of the soldier’s hardships. This lower visibility could theoretically dampen the likelihood of individual event response, the transmission of elite cues concerning wartime costs, and sustained coverage of the full consequences of war in media outlets. To the extent that the costs paid by wounded sol- diers are more removed from the public eye, the behavior of the public and public officials should not be altered.137

In an empirical analysis, detailed in the Appendix, we find that, at least in the 2006 midterms—an election in which the Iraq War was perhaps the most salient issue—non-fatal casualties did not have the same resonance with voters as fatal casualties. This does not mean that non-fatal casualties never have electoral ramifi- cations.138 However, if they did not in this context it is quite likely

134. Sara Corbett, The Permanent Scars of Iraq, N.Y. TIMES (Feb. 15, 2004), http://www.nytimes.com/2004/02/15/magazine/the-permanent-scars-of- iraq.html.

135. Kathy A. DeBarr, To Hell and Back: Wounded Warriors Return Home to Fight Yet Another Battle, 5 CAL. J. HEALTH PROMOTION 58, 61–63 (2007).

136. See UPI’s Benjamin Honored for Military Reporting, UPI.COM, Feb. 4, 2004, http://www.upi.com/Business_News/Security-Industry/2004/02/04/ UPIs-Benjamin-honored-for-Army-reporting/29781075950000/.

137. Of course, the significantly greater number of non-fatal casualties may compensate for the lower visibility of any one non-fatal casualty. Moreo- ver, wounded soldiers have the ability to directly engage in the political process. These countervailing forces suggest an alternate hypothesis that non-fatal casu- alties may have just as significant if not even more so political ramifications as fatal casualties. We test between these competing hypotheses in the analyses that follow.

138. In the only other analysis of the electoral effects of non-fatal casual- ties on electoral outcomes, Karol and Miguel find modest evidence (p < .10) that a state’s wounded in action rate depressed support for President George W. Bush in 2004 after controlling for the killed in action casualty rate. Karol & Miguel, supra note 128, at 633. However, the relationship between KIA rates and Bush’s electoral fortunes was statistically stronger. Id.

2016 Invisible Inequality 587

that non-fatal casualties also fail to encourage voters to punish the incumbents for costly military policies at the ballot box in many other conflict environments. If voters punished pro-war incum- bents for fatal and non-fatal casualties to the same degree, the democratic constraint on costly military policies would be consid- erably stronger.

D. Non-fatal Casualties and Public Support for the War in Afghanistan

To further assess the relative influence of information about fatal and non-fatal casualties on public support for war we em- ployed a survey experimental approach.139

Subjects were randomly assigned to one of five experi- mental groups. In the first treatment group, we told subjects that 2,312 American service members had been killed to date in Af- ghanistan. In treatments two and three, we instead told subjects the number of non-fatal casualties sustained in Afghanistan. In the second treatment we informed subjects that 17,674 Americans had been wounded in action. The third treatment was identical to the second; however, this treatment reported a much larger figure, 217,674, which represents the estimated non-fatal casualty count when expanding the definition to include non-physical wounds, such as PTSD and other brain injuries, the estimated numbers are orders of magnitude higher. The fourth treatment also used the larger figure of non-fatal casualties, but it informed subjects that of these 17,674 were physical wounds while the rest were “invisible” wounds of war, such as depression and PTSD. Our final treatment was identical to the fourth, but it also informed subjects of the number of fatal casualties sustained in Afghanistan. Complete wording for each treatment is provided in the Appendix.

From March 1 to March 3, 2014, we recruited an online convenience sample of 337 subjects. Demographics, and addition- al details on the experimental method are reported in the Appen- dix.

139. Because fatal and non-fatal casualty rates are highly correlated, it is difficult to assess their relative influence on public opinion by examining aggre- gate time series opinion data alone. In an experimental approach, we can direct- ly manipulate the information that subjects receive about casualties sustained in a conflict and examine how support for war varies across informational cues.

588 The University of Memphis Law Review Vol. 46

After reading a screen with some basic background infor- mation concerning the study, subjects were randomly assigned to one of the five treatment groups described above. All subjects were then asked the same question taken from previously pub- lished polls conducted by NBC News/The Wall Street Journal: “Do you think the war in Afghanistan against the Taliban and Al Qaeda has been very successful, somewhat successful, somewhat unsuccessful, or very unsuccessful?”

Because subjects were randomly assigned to one of the treatment groups, the resulting differences in means across treat- ments are unbiased. Figure 5 presents the percentage of respond- ents answering that the Afghan War has been very or somewhat successful across the five treatment groups.140

Consistent with the hypothesis that non-fatal casualties may not have the same resonance with the American public as fatal casualties, we observe a large and statistically significant (p = .05) difference in war support between the KIA and WIA treatments. Whereas only 44% of respondents who were only told of the 2,312 deaths judged the war very or somewhat successful, that number increased to 59% among the group told that 17,674 American sol- diers had been wounded.141 A difference this large is very unlikely to have emerged by random chance alone. Instead, the data strong- ly suggests that information about fatal casualties sustained in war can significantly lower support more than information about non- fatal casualties, even when the latter total is many times larger than the former.

Subjects in our third treatment group were told about the 200,000+ Americans wounded in war, physically or otherwise. In this treatment, we provided no additional context, but simply re- ported the estimated total of 217,674 wounded Americans. De-

140. The control group baseline is omitted here for ease of presentation. The mean in this control (50%) lies between that in the KIA and WIA treat- ments.

141. For comparative reference, the last time that the NBC/Wall Street Journal asked this question in January 12–15, 2013, fifty-five percent of Ameri- cans said that the war was either very or somewhat successful. See Mark Mur- ray, NBC/WSJ Poll: Public Lowers Expectations Heading into Obama’s 2nd Term, NBC NEWS (Jan. 17, 2013 3:30 PM), http://firstread.nbcnews.com/_news/ 2013/01/17/16570498-nbcwsj-poll-public-lowers-expectations-heading-into- obamas-2nd-term (citing polling data).

2016 Invisible Inequality 589

spite the staggeringly high total, the percentage of respondents in this treatment who judged the war a success was virtually identical to that in the WIA treatment, 58% versus 59%. Moreover, this figure is also significantly higher than that observed in the KIA treatment. It is striking that even when the number of non-fatal casualties reported is orders of magnitude larger, we find that pub- lic opinion is more opposed to the war when fatal casualties are discussed.

Our fourth treatment allows us to examine whether provid- ing additional context for the much larger estimated figure of 217,674 American soldiers wounded in Afghanistan influences public support for the war. Subjects in this treatment were in- formed that only 17,674 of these non-fatal casualties involved physical wounds; the remainder suffered more “invisible” wounds, such as PTSD. The additional information had no effect on eval- uations of the war, with virtually the same percentage judging the Afghan War a success as in the previous two non-fatal casualties treatments.

Finally, our fifth treatment was identical to the fourth, but it also informed subjects of the number of American soldiers who had died to date in Afghanistan. As shown in the final bar of Fig- ure 5, the small additional prompt about the number of fatal casual- ties significantly decreased support for the war in Afghanistan with only 44% in this treatment judging the war a success. This differ- ence in means across treatments four and five is statistically signif- icant (p < .10). Finally, we can compare treatment five with our initial KIA treatment. Both informed subjects of the number of American soldiers who had died in the war; however, the final treatment added information about the very large number of non- fatal casualties the war has also produced. Did learning about non- fatal casualties further depress evaluations of the Afghan War? Our data suggests that it did not. The percentage judging the war a success was exactly the same in our first (KIA only) and last (KIA plus non-fatal casualty information) treatments. The differences in means presented in Figure 5 are supported by more robust ordered logit analyses reported in the Appendix.

The data tell a compelling story. They plainly suggest that the true driver of popular assessments of the Afghan War was in- formation on fatal casualties; information about the much larger numbers of non-fatal casualties failed to lower popular beliefs about the war’s success.

590 The University of Memphis Law Review Vol. 46

This has immediate and tangible ramifications for politics and policy. To an extent unparalleled in American history, the wars in Iraq and Afghanistan have produced many more non-fatal than fatal casualties. And yet, non-fatal casualties fail to rally pub- lic sentiment against costly wars to the same extent as fatal casual- ties. This invisibility allows policymakers to wage war relatively free from the traditional democratic constraints on their actions.

60%

50%

40%

30%

20%

10%

0%

59%

Figure 5: Beliefs about Afghan War’s Success by Experimental Treatment

44%

58%

57%

44%

       

KIA WIA

Total wounded

Total wounded + note

KIA + total wounded + note

What to Notice in Figure 5: The experimental data presented in Figure 5 il- lustrate that evaluations of the War in Afghanistan are most affected by fa- tal, and not non-fatal casualties, suggesting that the wounded in action re- main more politically invisible. The mean in each of the three wounded casualty information treatments is significantly higher than in either of the two treatments reporting fatal casualty information, p < .10.

VI. DISCUSSION

We have shown to this point that there are indeed Two Americas with respect to military sacrifice; however, this reality is not routinely acknowledged. Moreover, we have shown that non- fatal casualties are rising vis-à-vis fatal casualties, yet those non- fatal casualties do not register politically in the same way.

% Believing War a Success

2016 Invisible Inequality 591

In this final Part we discuss the implications of these find- ings. Why don’t scholars and policymakers acknowledge this in- visible inequality? Do current law and policy adequately account for the challenges posed by inequality in military sacrifice? Is there a legal avenue for reform? If not, what type of intervention is warranted? We consider in turn:

(A) Why the inequality of military service remains invisible;

(B) Why the inequality is worth addressing;

(C) Why current legislative attempts to improve care for veterans are not sufficient to address the in- equality;

(D) Why courts are unlikely to intervene; and

(E)What would happen if the American public learned of the inequality?

A. Why Don’t We Want To Talk About the Inequality of Military Service?

It is not always easy to talk about class and military sacri- fice in America.142 But why? What explains the invisibility of inequality in military sacrifice in policy debates? To explore this question, we replicated the analysis reported in Part II on subjects recruited to take an online survey.143

142. This has been noted in the context of military inequality. GLASSER, supra note 21, at 128 (“Social and economic class in America has never been a comfortable thing to talk about in private, much less to discuss in public.”). Bunting observed that there is a “national uneasiness about the profoundly une- qual sharing of the military burden in the early years of the twenty-first centu- ry.” Bunting, supra note 24. He also wrote, “The issue of military conscription is deeply controversial, of course; and it is one of a family of public policy ques- tions, recurrent and vexed, upon whose difficulties people advance, make nerv- ous reconnaissances, and then withdraw, unwilling to engage them fully.” Id.

143. From February 14–15, 2015, we recruited 314 subjects via Mechani- cal Turk to take an online survey on which we asked whether they believed there was a casualty gap. Additional details are provided in the Appendix.

592 The University of Memphis Law Review Vol. 46

We asked an internet convenience sample of 314 Ameri- cans: “Thinking about the American soldiers who have died fighting in Iraq and Afghanistan, what parts of the United States do you think they are coming from?” As in the survey described above, respondents had three choices: (i) More casualties are com- ing from poorer, less educated parts of the country; (ii) More casu- alties are coming from richer, more educated parts of the country; or (iii) There is not a significant difference in the share of casual- ties coming from rich/high education and poor/low education parts of the country.

But this time, we added another layer to our analysis by asking these same respondents to answer the question, “What do you think is the primary reason that motivates young men and women to join the United States Armed Forces?” Subjects were able to type in a response, and we then coded the responses into three variables: whether the respondent cited only patriotic moti- vations; only economic motivations; or a mix of the two. Twenty- five percent of respondents listed only patriotic motivations. Fifty- three percent of the respondents listed only socioeconomic motiva- tions.

Differential beliefs in motivations for enlisting were corre- lated with being mistaken about inequality. People who believe in shared sacrifice also tend to believe that individuals join the mili- tary for purely patriotic, rather than economic, reasons. Whereas 65% of respondents who believe in a casualty gap cited socioeco- nomics as the only main motivation for enlisting, only 30% of those who rejected the existence of a casualty gap did so.144

The survey data suggests to us that part of the refusal to face up to inequality in military sacrifice is due to the belief that the unequal results are solely the result of freely-made individual choices.

The allure of choice is well documented in decades of re- search by social psychologists. Much of this work has built on Melvin Lerner’s (1980) “just world” hypothesis that humans prefer to believe that individual choice, rather than the surrounding situa- tion, is responsible for outcomes. The phenomenon is at work in law, and has been well documented by legal scholars such as Jon

144. This difference is statistically significant, p < .001.

2016 Invisible Inequality 593

Hanson and colleagues.145 There is, in particular, a large body of literature in psychology discussing how the desire to believe that one is living in a just world can affect attitudes toward inequality and redistributive policies.146

For instance, experimental data show that we prefer to point to a rape victim’s “poor choices” to explain the victim’s as- sault, and like to give ourselves credit for individual hard work instead of fully appreciating the situational context.147 The conclu- sion from this large body of research is clear: our situations de- termine our actions more than we would like to admit. The world is not “just,” but we go to great lengths to make it so in our heads.

145. See Jon Hanson & David Yosifon, The Situational Character: A Crit- ical Realist Perspective on the Human Animal, 93 GEO. L. J. 1 (2004); Jon Han- son & Adam Benforado, The Costs of Dispositionism: The Premature Demise of Situationist Law and Economics, 64 MD. L. REV. 24 (2005); Ronald Chen & Jon Hanson, Categorically Biased: The Influence of Knowledge Structures on Law and Legal Theory, 77 S. CAL. L. REV. 1106 (2004); Ronald Chen & Jon Hanson, The Illusion of Law: The Legitimating Schemas of Modern Policy and Corpo- rate Law, 103 MICH. L. REV. 1 (2004); Jon Hanson & David Yosifon, The Situa- tion: An Introduction to the Situational Character, Critical Realism, Power Economics, and Deep Capture, 152 U. PA. L. REV. 129, 132 (2003); Jon Hanson & Kathleen Hanson, The Blame Frame: Justifying Racial Oppression in Ameri- ca, 41 HARV. C.R.-C.L. L. REV. 413 (2006).

146. Lauren D. Appelbaum, Mary Clare Lennon & J. Lawrence Aber,

When Effort Is Threatening: The Influence of the Belief in a Just World on Americans’ Attitudes Toward Antipoverty Policy, 27 POL. PSYCHOL. 387, 390 (2006). See generally MELVIN J. LERNER, THE BELIEF IN A JUST WORLD: A FUNDAMENTAL DELUSION (1980); RESPONSES TO VICTIMIZATIONS AND BELIEF IN A JUST WORLD (Leo Montada & Melvin J. Lerner eds., 1998); Roland Bé- nabou & Jean Tirole, Belief in a Just World and Redistributive Politics, 121 Q. J. ECON. 699 (2006). These attitudes, of course, are not held uniformly by Ameri- cans and vary according to how the issue is framed or worded. Christopher Faricy & Christopher Ellis, Public Attitudes Toward Social Spending in the United States: The Differences Between Direct Spending And Tax Expenditures, 36 POL. BEHAV. 53, 58 (2014) (“[T]he way a social program is presented and framed to the public will have a substantial impact on citizens’ support for it”); see also Max Rose & Frank R. Baumgartner, Framing the Poor: Media Cover- age and U.S. Poverty Policy, 1960–2008, 41 POL’Y STUD. J. 22 (2013); Gregory A. Huber & Celia Paris, Assessing the Programmatic Equivalence Assumption in Question Wording Experiments Understanding Why Americans Like Assis- tance to the Poor More Than Welfare, 77 PUB. OP. Q. 385 (2013).

147. Francis X. Shen, How We Still Fail Rape Victims: Reflecting on Re- sponsibility and Legal Reform, 22 COLUM. J. GENDER & L. 1, 1–2 (2011).

594 The University of Memphis Law Review Vol. 46

Reframing issues to downplay fundamental tensions is common in the realm of inequality in military service. In their book Tragic Choices, former Dean of Yale Law School and now Federal Judge Guido Calabresi, with law professor Philip Bobbit, addressed the issue head on. Recognizing that military manpower policies are not givens, but are the result of political choices, Cala- bresi and Bobbit observe that “[b]y making the result seem neces- sary, unavoidable, rather than chosen, it attempts to convert what is tragically chosen into what is merely a fatal misfortune.”148

Like Calabresi and Bobbit, we believe that “[h]onesty is the most influential brace in the tragic equilibrium.”149 As the authors argue, “[t]he failure to make society aware of its implicit choices will diminish, with each averting of the eyes, the values of open- ness and honesty.”150

The empirical reality is that the “choice” to join the military is contingent on a number of factors. To be sure, many who serve cite non-economic reasons, chief amongst them patriotic duty.151 Some acknowledge that the economic benefits are also a factor.152 Others point out that they didn’t join for the money at all.153 Some look to the military after deciding their lives are not what they want them to be.154 They cite discipline, structure, and honor as

148. GUIDO CALABRESI & PHILLIP BOBBITT, TRAGIC CHOICES 21 (1978).

149. See id. at 26.

150. Id. at 48.

151. See Adam Silow, Why They Chose the Military, THE PRESS

DEMOCRAT: TEEN LIFE (Aug. 29, 2012), http://teenlife.blogs.pressdemocrat. com/12004/why-they-chose-the-military/ (quoting many teens in a high school recruiting program as joining for patriotic reasons).

152. See Stacy Bare, Why I Joined the Army, HUFFPOST (Jan. 30, 2012), http://www.huffingtonpost.com/stacy-bare/army-experience_b_1240598.html (acknowledging that he received a free college education).

153. Matt, Comment to Why Did You Join?, RANGERUP (Aug. 18, 2010, 11:01 AM), http://rhinoden.rangerup.com/why-did-you-join/; Jeremy Leo, Comment to Why Young People Join the Military, WORDPRESS (Mar. 16, 2010, 9:49 PM), http://counterrecruiter.wordpress.com/2007/08/03/why-young- people-join-the-military/#comment-5688 (“[T]oo many young kids are joining the military of late just for money and all the wrong reasons.”).

154. See, e.g., The Girls Guide to the AIR FORCE: The Reasons Why We Join, HUBPAGES (Mar. 28, 2012), http://hotpinkcombtboots.hubpages.com/hub/ the-girls-guide-to-surviving-air-force-basic-military-training-the-embracing-the- military-lifestyle.

2016 Invisible Inequality 595

important reasons they join.155 Some join because they think it is their moral responsibility.156 Others join to support or give back to the country.157 Many cite the desire to protect loved ones.158 Those who currently serve may cite the attack of September 11th as an inspiration.159 Many join because of family members who have done the same.160

In sum, the reasons for military service are plentiful and di- verse. It is clear that many serve for non-economic reasons; it is equally clear that many decide in part based on economic consid- erations.161

Despite these complexities about whether the choice to serve is truly a voluntary one, government officials have regularly invoked the voluntary nature of today’s military. Former Defense Secretary Donald Rumsfeld offered an illustrative response in 2003. When asked about the possibility of a draft, Secretary

155. Victoria Swingler, Comment to Why We Joined, NAVYGIRL.ORG (May 31, 2006), http://www.navygirl.org/whywejoined.htm (Responders Yatsu, Mace, and Gonzalez also expressed this sentiment); Steve Sybert, Comment to Why Did You Join?, supra note 153 (saying he needed direction and didn’t have work ethic).

156. Mark Daily, Why I Joined, L.A. TIMES http://www.latimes.com/local/ la-me-daily16feb16_essay-htmlstory.html (saying he joined because he thought it was the duty of a humanist; killed in explosion in Iraq).

157. Drew Z., Comment to Why Did You Join?, supra note 153 (giving a sentiment similar to responder Clifford Fargason, among others).

158. Rye MacCallan, Comment to Why Did You Join?, supra note 153.

159. Alex Kingsbury, The Pros and Cons of Military Service, USNEWS.COM (Oct. 21, 2010, 9:12 AM), http://www.usnews.com/news/articles/ 2010/10/21/the-pros-and-cons-of-military-service?page=3 (citing an increase in a public service incentive since 9/11); twenty-two other responders on message threads mentioned 9/11. See Why We Joined, supra note 155; Why Did You Join?, supra note 153.

160. Thirty-seven separate commenters and sources mention following in a family tradition or being inspired by a family member. See Why Did You Join?, supra note 153.

161. As political theorist Michael Sandel has pointed out, “[t]he term ‘vol- unteer’ is something of a misnomer. Soldiers do not volunteer in the way that people volunteer to work in the local soup kitchen on Thanksgiving – that is, to serve without pay. The volunteer army is a professional army, in which soldiers work for pay.” MICHAEL J. SANDEL, WHAT MONEY CAN’T BUY: THE MORAL LIMITS OF MARKETS, THE TANNER LECTURES ON HUMAN VALUES 110 (1998), http://tannerlectures.utah.edu/_documents/a-to-z/s/sandel00.pdf.

596 The University of Memphis Law Review Vol. 46

Rumsfeld replied, “We have people serving today – God bless ‘em – because they volunteered. They want to be doing what it is they’re doing. . . . Today . . . every single person there is there be- cause they stuck their hand up, said “I’d like to do that.”162 Reac- tions such as Rumsfeld’s prevent an honest accounting of the issue of military sacrifice and economic inequality.

B. Should We Care About Inequality in Military Sacrifice?

For some, the Two Americas of military sacrifice are nei- ther surprising nor cause for concern. Consider, for the purposes of comparison, inequality and another American institution, McDonald’s. It would not be surprising to learn that those work- ing on the front lines of McDonald’s are disproportionately from lower-income neighborhoods.163 Nor would it be shocking to find that those in upper management at McDonalds are more likely to have had better educational opportunities. These market forces— that put low-education, low-skilled workers on the fry griddles and high-education, high-skilled managers into upper level corporate offices—is what shareholders want because it maximizes efficien- cy of operations. Not only are we less inclined to see a moral problem with McDonald’s operations, an argument can be made that McDonald’s provides its entry-level workers with important opportunities for career advancement that they would not obtain otherwise. If this market logic holds in the military service context as well, then Americans should expect a casualty gap, and the gap should not affect their support for war efforts.

Indeed, many have pointed out that positive benefits can flow from the military’s reaching out to individuals of lower socio-

162. Pentagon Briefing, TRANSCRIPTS, CNN.COM (Jan. 7, 2003, 11:02), http://transcripts.cnn.com/TRANSCRIPTS/0301/07/se.02.html.

163. This is generally the case in the fast-food industry. See SYLVIA ALLEGRETTO ET AL., FAST FOOD, POVERTY WAGES: THE PUBLIC COST OF LOW- WAGE JOBS IN THE FAST-FOOD INDUSTRY 6 (2013), http://laborcenter.berkeley.edu/pdf/2013/fast_food_poverty_wages.pdf; KATHERINE S. NEWMAN, NO SHAME IN MY GAME: THE WORKING POOR IN THE INNER CITY 4–7 (1999); Orley Ashenfelter & Stepan Jurajda, Cross-country Comparisons of Wage Rates: The Big Mac Index 8–10, 12–14 (Oct. 2001), http://crei.cat/conferences/Unemployment_in_Transition_Economies_Developm ents,_Challenges_and_Lessons_from_the_EU_and_the_US_/activities/sc_confe rences/12/ashenfe.pdf.

2016 Invisible Inequality 597

economic status.164 Military service can be an important mecha- nism for improving economic attainment. Sociologists Pamela Bennett and Katrina Bell McDonald have reviewed the evidence and conclude that, at least for some, the military can be a “turning point” for disadvantaged youth.165 In terms of Black social mobili- ty, Colin Powell says, “let the rest of American society open its doors to African Americans and give them the opportunities they now enjoy in the armed forces.”166 There may be society-wide benefits too, if one agrees with Judge Richard Posner’s argument that “the true consequence of the demographics of the armed forc- es—a consequence that communitarians should applaud—is that the nation’s admiration for these scions of the lower middle class helps to bind the different income classes together.”167

While we don’t doubt the value of social mobility provided to some soldiers in the military, we take the view that the U.S. Armed Forces is not just another employer. As our data discussed earlier showed, this assessment is shared by many if not most Americans. This view starts with Defense Department Form 4, the form an American soldier signs when they enlist or re-enlist in the armed forces. On the second page of the form, individuals are in- structed: “My enlistment/reenlistment agreement is more than an employment agreement.”168 The language of the form codifies what our civics class teaches us: military service is more than just a job. It is service to the nation that may place one in harm’s way.169

164. See Alair MacLean & Glen H. Elder Jr., Military Service in the Life Course, 33 SOCIOLOGY 175, 184–85 (2007); Robert J. Sampson & John H. Laub, Socioeconomic Achievement in the Life Course of Disadvantaged Men: Military Service as a Turning Point, Circa 1940-1965. 61 AM. SOC. REV. 347, 347 (1996), http://scholar.harvard.edu/sampson/files/1996_asr_laub.pdf.

165. Pamela R. Bennett & Katrina Bell McDonald, Military Service as a Pathway to Early Socioeconomic Achievement for Disadvantaged Groups, inLIFE-COURSE PERSPECTIVES ON MILITARY SERVICE, supra note 18, at 120.

166. COLIN POWELL & JOSEPH E. PERSICO, MY AMERICAN JOURNEY 501 (1995).

167. Richard A. Posner, An Army of the Willing, THE NEW REPUBLIC (May 18, 2003), http://www.newrepublic.com/article/army-the-willing.

168. ARMED FORCES OF THE U.S. DEF. TECH. INFO. CTR., FORM DD 4 2 (Oct. 2007), http://www.dtic.mil/whs/directives/forms/eforms/dd0004.pdf.

169. As Robert Osgood has observed,
[T]he nation’s ability to sustain a defense program is not only a matter of the gross national product, per capita income, and

598 The University of Memphis Law Review Vol. 46

When he took office on January 20, 2009, President Barack Obama concluded his inaugural speech by recalling the words of Thomas Paine: “Let it be told to the future world . . . that in the depth of winter, when nothing but hope and virtue could survive . . . that the city and the country, alarmed at one common danger, came forth to meet [it].”170 What President Obama didn’t quote were the sentences immediately before and after this passage. If he had, he would have also told the nation:

I call not upon a few, but upon all: not on this state or that state, but on every state . . . It matters not where you live, or what rank of life you hold, the evil or the blessing will reach you all. The far and the near, the home counties and the back, the rich and the poor, will suffer or rejoice alike.171

the other objective criteria of economic strength but, just as much, a reflection of what the citizenry, its political represent- atives, and government officials are willing to sacrifice in terms of competing values for the sake of a particular national strategy.

ROBERT OSGOOD, LIMITED WAR: THE CHALLENGE TO AMERICAN STRATEGY 275 (1957).

170. Barack Obama Inaugural Address (Jan. 20, 2009) (alteration in origi- nal), http://www.whitehouse.gov/blog/inaugural-address/.

171. The quote is from Thomas Paine’s Common Sense, The Crisis. It appeared in the Philadelphia Journal in December 1776, and George Washing- ton had it read to his troops at Valley Forge to boost their morale. The full quote reads:

Quitting this class of men, I turn with the warm ardor of a friend to those who have nobly stood, and are yet determined to stand the matter out: I call not upon a few, but upon all: not on this state or that state, but on every state: up and help us; lay your shoulders to the wheel; better have too much force than too little, when so great an object is at stake. Let it be told to the future world, that in the depth of winter, when noth- ing but hope and virtue could survive, that the city and the country, alarmed at one common danger, came forth to meet and to repulse it. Say not that thousands are gone, turn out your tens of thousands; throw not the burden of the day upon Providence, but “show your faith by your works,” that God may bless you. It matters not where you live, or what rank of life you hold, the evil or the blessing will reach you all. The

2016 Invisible Inequality 599

In our view, inequality in military service runs counter to the American ethos of shared sacrifice.172 That said, we do not advocate for a return to the draft.173 What we should do, however, is recognize inequality as another cost of war—a cost that should be addressed through even better resources for wounded veterans and for the families of the fallen.

We agree with Matthew Ivey, who writes:

Any moral society would demand that the basic needs of veterans be met in exchange for the sacri- fices that their country has asked of them. Support- ing our troops and our veterans must go beyond bumper stickers and political bluster. The true measure of our society will be defined by how we treat our returning veterans.174

far and the near, the home counties and the back, the rich and
the poor, will suffer or rejoice alike. The heart that feels not
now is dead; the blood of his children will curse his coward-
ice, who shrinks back at a time when a little might have saved
the whole, and made them happy. I love the man that can
smile in trouble, that can gather strength from distress, and
grow brave by reflection. ‘Tis the business of little minds to
shrink; but he whose heart is firm, and whose conscience ap-
proves his conduct, will pursue his principles unto death.
THOMAS PAINE, THE CRISIS (1776), http://www.ushistory.org/paine/crisis/c-

01.htm.
172. We agree with law professor Florence Wagman Roisman, who, writ-

ing in the context of veterans homelessness, concludes that our current system “is hardly a model of gratitude for the wealthiest, most powerful nation on earth; it can and should be corrected.” Florence Wagman Roisman, National Ingrati- tude: The Egregious Deficiencies of the United States’ Housing Programs for Veterans and the “Public Scandal” of Veterans’ Homelessness, 38 IND. L. REV. 103, 176 (2005).

173. We agree with Ivey: “Going forward, it is important to acknowledge the inefficiency and possible immorality of the draft. It is equally important to acknowledge that the last decade of U.S. military involvement overseas has expressed immorality and inefficiency in the all-volunteer force.” Ivey, supra note 27, at 561.

174. Id.

600 The University of Memphis Law Review Vol. 46

And we believe that a step in that direction involves an honest assessment of inequalities in who has made these sacrific- es.175

C. Are Veterans Already Receiving the Care They Need?

Our concerns about inequality would be significantly miti- gated if veterans—rich and poor alike—received compensatory resources in recognition of their sacrifice. But, as many commen- tators have pointed out, today’s health care for veterans remains sub-par.

The VA has, almost since its inception, been criticized.176 In 2013 a scholar summarized the sentiment in this way: “Most

175. Although we find them, at least at present, to be a political non- starter, we encourage efforts to expand national service. Ivey, for instance, has suggested that we might make national service a prerequisite for some benefits (such as serving in certain civilian leadership positions). Id. Military sociolo- gist Charles Moskos also proposed a national service solution. See generally CHARLES C. MOSKOS, A CALL TO CIVIC SERVICE (1988). National service, however, has not gained traction in Congress. See Congressional Commission on Civic Service Act, H.R. 1444, 111th Cong. (2009). Perhaps this will change in the future due to the advocacy of groups such as Service Nation. SERVICE NATION, http://www.servicenation.org/ (last visited Mar. 15, 2016).

176. Historically there are many instances of the country’s failure to ade- quately provide health care for returning soldiers. In the wake of World War I, for instance, the federal Veteran’s Bureau was created amidst “widespread frus- tration among veterans and veterans’ groups, legislators, and the popular press with problems ranging from excessive red tape and the slow processing of claims to an appalling lack of services and unfair determination of eligibility.” Rosemary A. Stevens, The Invention, Stumbling, and Reinvention of the Modern U.S. Veterans Health Care System, 1918-1924, in VETERANS’ POLICIES, VETERANS’ POLITICS: NEW PERSPECTIVES ON VETERANS IN THE MODERN UNITED STATES 38 (Stephen R. Ortiz ed., 2012); see also ROBERT KLEIN, WOUNDED MEN, BROKEN PROMISES 22 (1981) (“Disgust with the VA is na- tionwide, and its expression is often visceral.”). As one World War I quip went:

God and the military veteran we adore ... In times of danger, not before;
The Danger pass’d and all things righted God is forgotten and the veteran slighted.

KLEIN, supra, at 20; see also Lawrence Ingraham & Frederick Manning, Ameri- can Military Psychiatry, in MILITARY PSYCHIATRY: A COMPARATIVE PERSPECTIVE 25 (Richard A. Gabriel ed., 1986) (“American attempts to under- stand and respond to battle stress casualties have ranged from the positively brilliant to the positively pathetic.”).

2016 Invisible Inequality 601

agree that the system has vastly improved in recent years, but it remains for many a challenging bureaucratic maze whose efficien- cy depends on the responsiveness and attentiveness of individuals who are often overwhelmed.”177

To be sure, the Veterans Administration has taken many positive steps toward improving care. For instance, the VA’s Health Services Research and Development Service (“HSR&D”) has funded Centers of Excellence and Centers of Innovation.178 Treatment of schizophrenia has advanced in the VA system in par- allel with relevant advances in the understanding of and treatments for schizophrenia.179 The military also provides a large number of support services designed to help veterans reintegrate into the workforce.180 Yet much work remains to be done.181

The administration of veterans’ disability claims continues to be a contested space. Historically, there was great concern with fraud. During the United States Civil War “all symptoms or claimed disabilities that were not accompanied by verifiable physi- cal injury were considered malingering unless proven other-

wise.”182

177.

And, to be sure, there remain valid concerns today about

Lawrence, supra note 109, at 142.

178.
http://www.hsrd.research.va.gov/centers/ (last visited Mar. 15, 2016). At least one commentator suggests that investments such as these will “generate innova- tive programs of healthcare that will provide a leading direction for healthcare in the United States for the twenty-first century.” Thomas W. Miller, Centers of Excellence in the Department of Veterans Affairs, in THE PRAEGER HANDBOOK OF VETERANS’ HEALTH, supra note 11, at 20.

179. Daniel N. Allen & Gerald Goldstein, Schizophrenia Spectrum and Other Psychotic Disorders, in PSYCHOLOGICAL ASSESSMENT OF VETERANS,supra note 76, at 233.

180. Nathan D. Ainspan, Finding Employment as a Veteran With a Disa- bility, in RETURNING WARS’ WOUNDED, INJURED, AND ILL, supra note 75, at 106–09.

181. See, e.g., Patricia E. Roberts, Post-9/11 Veterans: Welcoming Them Home As Colleagues and Clients, 45 U. MEM. L. REV. 771 (2015); Jayme M. Cassidy, Suddenly Discharged the Combat Continues: Eliminating the Legal Services Gap to Ensure Veterans’ Success After Leaving Military Service, 45 U. MEM. L. REV. 837 (2015).

182. Shane S. Bush, Assessment of Symptom and Performance Validity in Veterans, in PSYCHOLOGICAL ASSESSMENT OF VETERANS, supra note 76, at 436.

Health Services Research and Development, VA HSR&D CENTERS,

602 The University of Memphis Law Review Vol. 46

malingering amongst veterans.183 But to its credit, the VA now uses a “benefit of the doubt” approach when assessing veterans’ disability claims.184 When experts are unsure about the presence of a disability, the scales should tip in the veteran’s favor.

Yet despite improvements, today’s VA is still underper- forming. As recently as July 21, 2015, President Obama an- nounced that he was “still not satisfied” with the VA.185 There is a great backlog for veterans seeking services.186 Of particular con- cern for the argument of this Article is the VA’s treatment for brain injuries. Even when receiving treatment, a large percentage of veterans report being unsatisfied.187

183. Id. at 437; Thomas Freeman, Melissa Powell & Tim Kimbrell, Meas- uring Symptom Exaggeration In Veterans With Chronic Posttraumatic Stress Disorder, 158 PSYCHIATRY RESOL. 374, 376 (2008).

184. A unique standard of proof applies in decisions on claims for veterans benefits. Unlike other claimants and litigants, pursuant to 38 U.S.C. § 3007(b), a veteran is entitled to the “benefit of the doubt” when there is an “approximate balance of positive and negative evidence.” Gilbert v. Derwinski, 1 Vet. App. 49, 53 (Vet. App. 1990); see also Rory E. Riley, The Importance of Preserving the Pro-Claimant Policy Underlying the Veterans’ Benefits Scheme: A Compar- ative Analysis of the Administrative Structure of the Department of Veterans Affairs Disability Benefits System, 2 VETERANS L. REV. 77, 115 (2010).

185. Press Release, Office of the Press Sec’y, Remarks by the President to the VFW National Convention, THE WHITE HOUSE (July 21, 2015), https://www.whitehouse.gov/the-press-office/2015/07/21/remarks-president- vfw-national-convention.

186. DAVID GODFREY, VETERANS APPEALS GUIDEBOOK: REPRESENTING VETERANS IN THE U.S. COURT OF APPEALS FOR VETERANS CLAIMS 41 (Ronald L. Smith ed., 2013) (“The Department of Veterans Affairs receives about 900,000 claims for benefits each year and has a current backlog of about 600,000 claims.”).

187. VA Mental Health Care: Hearing on Access to VA’s Mental Health Care Before the H. Comm. on Veterans Affairs, 113th Cong. (July 10, 2014) (statement of Warren Goldstein, Assistant Director for TBI and PTSD Programs, National Veteran Affairs and Rehabilitation Commission The American Legion) (“Two troubling numbers stood out in a recent survey conducted by The Ameri- can Legion to evaluate the effectiveness of treatments provided by VA when treating veterans suffering from Posttraumatic Stress Disorder (PTSD) and Traumatic Brain Injury (TBI)—fifty-nine percent and thirty percent. Fifty-nine percent of veterans surveyed reported ‘no improvement’ or that they were ‘feel- ing worse’ after having undergone TBI and PTSD treatment. Nearly a third of veterans, 30 percent, stated they had terminated their treatment plan before it reached conclusion.”).

2016 Invisible Inequality 603

Law professor Olympia Duhart is one of many who have argued that “the entire Veterans Affairs (“VA”) regulatory scheme reflects an outmoded cultural refusal to acknowledge the mental and emotional strains of war.”188 Looking at the history of veter- ans’ health care in the Bush and Clinton administrations, a differ- ent set of commentators observed that “[m]ental and behavioral healthcare was seen as fundamentally separate from physical healthcare.”189 As one of us (Shen) has argued elsewhere, the ma- terial dualism distinction between “physical” and “mental” is prob- lematic in light of neuroscientific insights that all mental life is instantiated in the physical brain.190

Recognizing the neurobiological underpinnings of mental life would likely lead to more resources for the treatment of mental injuries. This is in part because it would lead to a change in how the military conceptualizes “wounded in action.”

The official Department of Defense definition of “Wound- ed in Action” reads this way:

A casualty category applicable to a hostile casualty, other than the victim of a terrorist activity, who has incurred an injury due to an external agent or cause. The term encompasses all kinds of wounds and oth- er injuries incurred in action, whether there is a piercing of the body, as in a penetration or perforat- ed wound, or none, as in the contused wound. These include fractures, burns, blast concussions, all effects of biological and chemical warfare agents, and the effects of an exposure to ionizing

188.
Anti-Subordination Analysis, 44 CREIGHTON L. REV. 883, 900 (2011). Duhart has similarly criticized the failure of the military to allow the Purple Heart to be awarded to those whose only injury is PTSD. Id. at 902 (“[T]he government’s stance on the debate makes clear that in its assessment, PTSD struggles earned in battle do not merit recognition generally associated with sacrifice and val- or.”).

189. Patrick H. Deleon & Paul C. Lewis, Foreword, in PSYCHOLOGICAL ASSESSMENT OF VETERAN, supra note 76, at xii.

190. See Francis X. Shen, Sentencing Enhancement and the Crime Vic- tim’s Brain, 46 LOY. U. CHI. L.J. 405, 406–07 (2014); Francis X. Shen, Mind, Body, and the Criminal Law, 97 MINN. L. REV. 2036 (2013).

Olympia Duhart, Soldier Suicides and Outcrit Jurisprudence: An

604

The University of Memphis Law Review Vol. 46

radiation or any other destructive weapon or agent. The hostile casualty’s status may be categorized as SI [“seriously ill or injured”], VSI [“very seriously ill or injured”], or NSI [“not seriously injured”].191

The WIA category covers many of the injuries soldiers suf- fer, but the operative phrase “due to an external agent or cause” excludes a number of injuries historically thought to be “internal” or “mental,” but which are today readily recognized by many in the neuroscientific community as being a physical injury just as con- crete as a broken bone. PTSD is not considered a “brain injury” because a brain injury is considered a disruption in brain function from an external source.192 We think this accounting should change, and returning soldiers who develop Post-Traumatic Stress Disorder should count as “wounded” soldiers.193

We recognize that expansion of this sort would require careful consideration of the causal relationship between military service and the mental disorder.194 The notion of a service- connected disability remains at the heart of Veterans Administra-

191. DEP’T OF DEF., DEPARTMENT OF DEFENSE INSTRUCTION NUMBER 1300.18 37 (2008), http://www.dtic.mil/whs/directives/corres/pdf/130018p.pdf.

192. John D. Otis et al., The Psychological Assessment of Veterans with Pain and Pain-Related Disorders, in PSYCHOLOGICAL ASSESSMENT OFVETERAN, supra note 76, at 393.

193. A related issue came up in 2008 and 2009, when the Department of Defense considered whether soldiers suffering from PTSD should be eligible to receive the Purple Heart. Explaining the DOD’s reasoning for answering no, Defense Department spokeswoman Eileen Lainez said, “PTSD is an anxiety disorder caused by witnessing or experiencing a traumatic event; it is not a wound intentionally caused by the enemy from an ‘outside force or agent,’ but is a secondary effect caused by witnessing or experiencing a traumatic event.” Jeff Schogol, Pentagon: No Purple Heart for PTSD, STARS AND STRIPES (Jan. 6, 2009), http://www.stripes.com/articleprint.asp?section=104&article=59810. For coverage in the blogosphere, see Ilona Meagher, Reaction to DoD Decision Against Awarding Purple Heart to Veterans with Combat PTSD, PTSD COMBAT BLOG (Jan. 11, 2009), http://ptsdcombat.blogspot.com/2009/01/reaction-to-dod- decision-against-purple.html.

194. Shen, Mind, Body, and the Criminal Law, supra note 190, at 2103 (“Chief amongst the concerns . . . [is] the issue of causation. The evidentiary concern is: how could a court verify, beyond a reasonable doubt, that the victim experienced a mental injury?”).

2016 Invisible Inequality 605

tion disability claims today.195 To gain disability benefits, the vet- eran must not only demonstrate evidence of the disability, but evi- dence that the disability is sufficiently connected to the veteran’s military service.196 Although scientific advances may allow us in the future to make more precise connections between combat ex- posure and precise brain injuries, at present the causal links be- tween exposure to combat, and in particular explosions, and Trau- matic Brain Injury (“TBI”) remain uncertain.197

Of course, in the future this may change. Today, the as- sessment of substance use disorders is only beginning to use bi- omarkers.198 But some “predict that it won’t be long before a sub- stantial amount of testing will be done under a magnet during a functional MRI procedure.”199 To that end, the National Center for PTSD has been actively studying the neurobiology of PTSD, in- cluding the use of new psychotherapies.200

Aware of the many deficiencies of the VA system, on May 24, 2014, President Obama told the nation in his Memorial Day radio address that “taking care of our veterans and their families is a sacred obligation.”201 We agree. We also agree with Obama’s

195. Jonathan Krisch, Judge, Jury, and the Gatekeeper: Admitting and Weighing Expert Testimony in Veterans’ Claims Adjudication and the Federal Courts, 4 VETERANS L. REV. 41, 57 (2012) (“Veterans alleging that a current disability is related to their service in the United States Armed Forces may apply to VA for compensation. If their disability was incurred in or aggravated by service (“service-connected”), they are awarded various levels of benefits de- pending on the severity of their disability.”).

196. Nema Milaninia, The Crisis at Home Following the Crisis Abroad: Health Care Deficiencies for US Veterans of the Iraq and Afghanistan Wars, 11 DEPAUL J. HEALTH CARE L. 327, 335 (2008) (“The requirement that the injury be ‘service connected’ is often the basis of legal dispute by veterans in need of expansive medical coverage or pension plans, particularly when the degree of the disability has a significant economic impact.”).

197. MILITARY NEUROPSYCHOLOGY (Carrie Kennedy & Jeffrey Moore eds., 2010); Katherine H. Taber et al., Blast-Related Traumatic Brain Injury: What Is Known?, 18 J. NEUROPSYCHIATRY 141 (2006).

198. Dephilippis et al., supra note 78, at 190–91.

199. Allen & Goldstein, supra note 179, at 234.

200. Matthew J. Friedman, The National Center for PTSD, in 4 THE

PRAEGER HANDBOOK OF VETERANS’ HEALTH, supra note 11, at 107–10.
201. Office of the Press Sec’y, Weekly Address: Paying Tribute to Our Fallen Heroes this Memorial Day, THE WHITE HOUSE (May 24, 2014),

https://www.whitehouse.gov/the-press-office/2014/05/24/weekly-address-

606 The University of Memphis Law Review Vol. 46

observation that these veterans have “done their duty,” and it is time that “this country does ours – now and for decades to come.”202 Yet for all of the rhetoric, the reality of the Obama Ad- ministration’s and Congress’s treatment of veterans is that it does not acknowledge—anymore than the courts do—the systemic ine- quality of military sacrifice.

A 2014 scandal at the VA Hospital in Phoenix—in which widespread delays for veterans made headlines—resulted in close scrutiny from Congress and the resignation of U.S. Veterans Af- fairs Secretary Eric Shinseki.203 An August 2014 report from the VA Office of Inspector General “identified serious conditions . . . that resulted in delays, some significant, in veterans’ access to health care services.”204

In response, the Veterans Access to Care through Choice, and Accountability, and Transparency Act of 2014 (“the Act”) was signed into law on August 7, 2014.205 The Act provided ten billion dollars to immediately aid veterans who had gone without care206 as well as five billion dollars for the VA to use for internal im- provement.207 Additionally, the Act contained provisions for au- thorizing leases on twenty-seven medical facilities,208 established

paying-tribute-our-fallen-heroes-memorial-day. The term “sacred obligation” had been used before, e.g. the title of a May 3, 2011 Congressional hearing on veterans affairs was called, “Sacred Obligation: Restoring Veteran Trust and Patient Safety.” Sacred Obligation: Restoring Veteran Trust and Patient Safety: Hearing Before the H.R. Comm. on Veterans’ Affairs, 112th Cong. (2011), https://veterans.house.gov/hearing-transcript/sacred-obligation-restoring- veteran-trust-and-patient-safety.

202. Office of the Press Sec’y, supra note 201.

203. Obama Accepts Veterans Affairs Chief Resignation with ‘Regret’,

WESTLAW J. MED. MAPRAC., June 5, 2014, at 1 (“U.S. Veterans Affairs Secre- tary Eric Shinseki resigned May 30 after a political firestorm over widespread delays in veterans’ medical care . . . .”).

204. VA OFFICE OF INSPECTOR GEN., REVIEW OF ALLEGED PATIENT DEATHS, PATIENT WAIT TIMES, AND SCHEDULING PRACTICES AT THE PHOENIX VA HEALTH CARE SYSTEM 34 (2014), http://www.va.gov/oig/pubs/VAOIG-14- 02603-267.pdf.

205. Pub. L. No. 113–146, 128 Stat. 1754 (2014).

206. Id. § 802(d).

207. Id. § 801(a).

208. Id. § 601(a).

2016 Invisible Inequality 607

new wait time goals,209 and directly addressed many of the issues perceived to have led to the recent improprieties.210 Finally, the Act called for accountability through investigations, reports, and third party audits, which promise to plague the VA for the foresee- able future.

There was some recognition in the Act that not all veterans had the same baseline access to health care services. Of particular note is section 38 C.F.R. section 64, which provides grants specifi- cally designed to help extend care to “underserved veterans.”211 Underserved communities are areas that meet one or more of the following criteria:

(1) Have a high proportion of minority group repre- sentation;

(2) Have a high proportion of individuals who have limited access to health care; or

(3) Have no active duty military installation that is reasonably accessible to the community.212

Yet these small bits of recognition fail to come close to recognizing the structural inequality—that is, the underlying soci- oeconomic makeup of the wounded warrior population—that per- vades the provision of veterans’ health care services more general- ly.

209. Id. § 101(s)(1).

210. See id. § 209 (explicitly disallows forgery of data).

211. In addition, in many places throughout the Code it is explicitly stated

that special consideration should be given to veterans living in rural areas. See 38 U.S.C. § 1720G(b)(5) (West 2015) (“The outreach shall include an emphasis on covered veterans and caregivers . . . living in rural areas.”). Also, the Secre- tary is instructed in section 1703(a) to contract with other facilities to provide care to veterans who may have limited access. Id. § 1703(a).

212. 38 C.F.R. § 64.2. Limited access to healthcare, in turn, is defined by the Health Resources and Services Administration of the Department of Health and Human Services. 38 C.F.R. § 64.2; see U.S. DEP’T OF HEALTH & HUMAN SERVS.: Medically Underserved Areas/Populations: Guidelines for MUA and MUP Designation, HEALTH RES. & SERVS. ADMIN., http://www.hrsa.gov/shortage/mua/index.html (last updated June 1995).

608 The University of Memphis Law Review Vol. 46

Without addressing these structural issues, it is no wonder that even after passage of the landmark Act, we continue to see headlines such as these (all from calendar year 2015):

VA to Iraq War Vet: ‘We’re not accepting any new patients’ (describing how an Iraq War veteran was turned away from the VA when he requested an ap- pointment to assess possible PSTD)213

V eterans Affairs Whistle-Blowers Blast New Agency Watchdog (describing the disappointment of many former VA workers with the new VA Deputy Inspector)214

Veterans Still Waiting For Care at VA Hospitals (describing long wait times and poor service for veterans in Arkansas)215

A centerpiece of the Act—the “Veterans Choice Card” sys- tem—allows veterans who have waited longer than thirty days for an appointment, or who live more than forty miles from a VA fa- cility, to seek care from a third party. This would seem to at least partially address the needs of those with long wait times. Yet without addressing, or even acknowledging, the structural econom- ic inequality, we remain skeptical that the Act will truly change the system.

213. Patricia Kime, VA to Iraq War Vet: ‘We’re Not Accepting Any New Patients,’ USA TODAY (July 1, 2015, 2:53 PM), http://www.usatoday.com/ story/news/nation-now/2015/06/30/iraq-war-veteran-veterans-affairs-no-new- patients/29546453/.

214. Donovan Slack, Veterans Affairs Whistle-Blowers Blast New Agency Watchdog, USA TODAY (July 30, 2015, 5:51 PM), http://www.usatoday.com/ story/news/politics/2015/07/30/veterans-affairs-whistle-blowers-blast-new- agency-watchdog/30890527/.

215. John Boyle & David B. Caruso, Veterans Still Waiting for Care at VA Hospitals, CITIZEN-TIMES (April 9, 2015, 9:47 AM), http://www.citizen- times.com/story/news/local/2015/04/09/veterans-still-waiting-care-va- hospitals/25480601/.

2016 Invisible Inequality 609 D. Are Courts Likely to Intervene?

In the face of inadequate policymaking from the executive and legislative branches, can we expect courts to successfully in- tervene? While we hesitate to predict too far into the future, recent case law suggests that this is unlikely.

Consider, by way of illustration, the lawsuit by Veterans for Common Sense (“VCS”) against the Department of Veterans Affairs seeking injunctive and declaratory relief to remedy exten- sive delays in the provision of mental health care and disability compensation claims by the VA.216

While the district judge found for the VA, a three-judge panel on the Ninth Circuit Court of Appeals found that these de- lays violated veterans’ Constitutional due process rights.217 The ruling was described at the time as “an enormous legal victory.”218

The victory, however, was short-lived. Just months later the Ninth Circuit, sitting en banc, ruled 10-1 that the court lacked jurisdiction to reprimand the VA in this way.219 The court con- cluded that the “complaint sounds a plaintive cry for help, but it has been misdirected to us.”220 Congress was the culprit:

216. Veterans for Common Sense v. Peake, 563 F. Supp. 2d 1049 (N.D. Cal. 2008), aff’d in part, rev’d in part and remanded, 678 F.3d 1013 (9th Cir. 2012).

217. Veterans for Common Sense v. Shinseki, 644 F.3d 845, 850 (9th Cir. 2011) opinion vacated on reh’g en banc, 678 F.3d 1013 (9th Cir. 2012) (“Veter- ans ask us to decide whether these delays violate veterans’ due process rights to receive the care and benefits they are guaranteed by statute for harms and inju- ries sustained while serving our country. We conclude that they do.”). Writing for the two-judge majority, Judge Stephen Reinhardt wrote that “The VA’s un- checked incompetence has gone on long enough; no more veterans should be compelled to agonize or perish while the government fails to perform its obliga- tions.” Id. at 851.

218. Paul Sullivan, Executive Director of Veterans for Common Sense, quoted in Paul Muschick, Veterans Deserve Better Treatment, THE MORNINGCALL (May 25, 2011), http://articles.mcall.com/2011-05-25/news/mc-watchdog- veterans-mental-health-be20110525_1_veterans-for-common-sense-ninth- circuit-court-district-court.

219. Shinseki, 678 F.3d at 1016, cert. denied, 133 S. Ct. 840 (2013) (“As much as we as citizens are concerned with the plight of veterans seeking the prompt provision of the health care and benefits to which they are entitled by law, as judges we may not exceed our jurisdiction.”).

220. Id. at 1036.

610

The University of Memphis Law Review Vol. 46

We would have preferred Congress or the President to have remedied the VA’s egregious problems without our intervention when evidence of the De- partment’s harmful shortcomings and its failure to properly address the needs of our veterans first came to light years ago. . . . We willingly acknowledge that, in theory, the political branches of our government are better positioned than are the courts to design the procedures necessary to save veterans’ lives and to fulfill our country’s obligation to care for those who have protected us. But that is only so if those governmental institutions are will- ing to do their job.221

The en banc decision echoed discussion earlier in the case history about separation of powers. Judge Alex Kozinski in dis- sent below had argued that “[m]uch as the VA’s failure to meet the needs of veterans with PTSD might shock and outrage us, we may not step in and boss it around.”222

The bottom line of Veterans for Common Sense v. Shinseki, and thus the bottom line for those hundreds of thousands of veter- ans awaiting VA responses to their mental health disability claims, is that Congress is the cause of the problem, and Congress holds the keys to real reform.

We are thus left with a circle of blame. The Courts blame Congress. Congress blames the VA. And the VA promises it will improve, but with a track record of broken promises. We suggest that this cycle will not be broken until we recognize that rampant economic inequality pervades the system. Inequality in military sacrifice, borne most especially by those who seek medical care upon returning from combat, is not just an administrative problem for Congress or the VA to “fix.” It is a deep, systemic problem that requires courts—and us—to act.

But if courts are unlikely to act, then that leaves the aca- demic community and the public as levers for change.

221. Shinseki, 644 F.3d at 850–51.

222. Id. at 890–91 (“Congress erected a big ‘keep out’ sign for us in the

Veterans’ Judicial Review Act (VJRA) . . . .”).

2016 Invisible Inequality 611 E. Fostering a National Debate on Inequality in Sacrifice

So what can be done? We believe that encouraging a na- tional dialogue on the Two Americas of military sacrifice is both realistic and would be effective in shifting public opinion.

For example, in the first survey experiment detailed in Part IV we found that telling individuals just one additional bit of in- formation about the presence or absence of a casualty gap had a considerable impact on respondents’ beliefs of whether the war in Iraq was a mistake. This 6% point increase is of note in its own right; the magnitude of the effect is even more striking when we remember the modest nature of the inequality cue and the amount of information that most Americans already possessed on Iraq with which this new cue had to compete. If a simple cue about casualty inequality can change popular attitudes on a military venture to this extent in the Iraq context, it is quite possible that the acknowl- edgement of a casualty gap could have even larger effects in other environments in which popular attitudes are more malleable and not so polarized along partisan lines.

Indeed, our second experiment—exploring how infor- mation about a casualty gap in previous conflicts affected the pub- lic’s willingness to use force in a range of future scenarios— showed evidence of much greater effects. In three of our four hy- pothetical scenarios, receiving the inequality information increased the percentage of Americans willing to sustain fewer than fifty casualties to achieve the stated objective by roughly 10% from the control group baseline.223

It is quite possible that an even more thorough accounting of the casualty gap, complete with vivid descriptions of individual soldiers and the effects of their deaths on poor communities could produce an even larger effect. Similarly, while our analyses of election data and follow-up experiment suggest that non-fatal cas- ualties are less influential on public opinion, greater awareness and discussion of the lasting ramifications of non-fatal casualties for socioeconomically disadvantaged communities might heighten their political salience. Greater public awareness of non-fatal cas- ualties and inequality of sacrifice would foster a more nuanced

223. Kriner & Shen, supra note 115, at 1183.

612 The University of Memphis Law Review Vol. 46 accounting of the human costs of war among both citizens and pol-

icymakers alike.

VII. CONCLUSION

The substantial empirical evidence amassed in this Article leads to the inescapable conclusion that there is socioeconomic inequality in military sacrifice. Americans wounded in war, as well as those who die, are disproportionately coming from poorer parts of the country. Moreover, wounded soldiers are more likely to return home to fewer community resources, which may nega- tively affect mental health. Yet courts have failed to recognize this reality, and policymakers have strong incentives to do the same because it reduces criticism of their deploying and keeping combat forces abroad. We have argued in this Article that such ignorance, whether willful or unintentional, is inexcusable in the face of the empirical evidence. Although it is politically convenient to over- look the Two Americas of military sacrifice, continuing to ignore the invisible inequality of America’s modern warfare will not make it go away.

APPENDIX

In this Appendix we discuss additional details of various statistical analyses discussed in the main text.

I. DETAILS OF THE PUBLIC OPINION SURVEYS

In the main text, we discuss results from seven original public opinion surveys that we conducted between 2007 and 2015. We used two different types of survey instruments. For three of the surveys we used a truly nationally representative telephone survey, conducted by a professional polling organization. For the remaining four surveys we recruited subjects nationally through Amazon’s Mechanical Turk service. In this section we discuss the details of each approach. Summary demographics for each of these samples are provided in Table A1.

Our questions were embedded on three separate CARAVAN omnibus surveys conducted by Opinion Research Corporation. CARAVAN is a twice-weekly telephone survey that employs a random-digit dialing (RDD) methodology to ensure a nationally representative sample of 1,000 adult Americans. Re- sults from the questions embedded on CARAVAN surveys are presented in Part II and Section V.B. of the Article.

In Section V.A., Section V.C., and Section VI.A., we report results from original web-based surveys hosted on the web site Qualtrics. Research using Qualtrics-based experiments has been published in a number of academic fields, suggesting that it meets scholarly expectations for quality online web-based experi- ments.224

All subjects were recruited via modest payments made available through Amazon Mechanical Turk’s payment service. No personally identifying information was collected. Studies as- sessing the quality of Turk subjects have found them to be engaged

224. Studies relying on Qualtrics experiments include Matthew R. Ginther et al., The Language of Mens Rea, 67 VAND. L. REV. 1327 (2014); David L. Schwartz, Christopher B. Seaman, Standards of Proof in Civil Litigation: An Experiment from Patent Law, 26 HARV. J.L. & TECH. 429 (2013); and Francis X. Shen, et al., Sorting Guilty Minds, 86 N.Y.U. L. REV. 1306 (2011).

614 The University of Memphis Law Review Vol. 46

in the online experimental stimuli, and to be significantly more representative than the convenience samples that would otherwise be used.225

Samples recruited via Mechanical Turk are particularly well-suited for survey experimental research. Indeed, recent re- search by political scientist Adam Berinsky and his colleagues demonstrates that replicating experiments on samples recruited in this way yields very similar results to previously published studies with nationally representative samples.226

225. Multiple studies have validated results using Amazon’s Mechanical Turk on a variety of assessments, especially when compared to samples of con- venience. See, e.g.,

Michael D. Buhrmester, Tracy Kwang & Samuel D. Gosling, Amazon’s Mechanical Turk: A New Source of Inexpensive, Yet High-Quality, Data?, 6 PERSP. ON PSYCH. SCI. 3

(2011); Joseph K. Goodman et al.,

Jon Sprouse, A Validation of Amazon Mechanical Turk for the Collection of Acceptability Judgments in Linguistic Theory, 43 BEHAV RES.

METHODS 155 (2011). 226.

Tara S. Behrend et al., The Viability of Crowdsourcing for

Survey Research, 43 BEHAV. RES. METHODS 800 (2011);

Data Collection in a Flat World: The

Strengths and Weaknesses of Mechanical Turk Samples, 26 J. BEHAV. DECISION

MAKING 213 (2012);

Adam J. Berinsky et al., Evaluating Online Labor Markets for Exper-

imental Research: Amazon.com’s Mechanical Turk, 20 POL. ANALYSIS 351, 366

(2012).

2016 Invisible Inequality Appendix 615

227. The national percentage of citizens with a college degree, percentage male, median age, and racial demographics was calculated based on data from the 2010 US Census. The source for the percentage of citizens in each political party was is a survey by NBC News, Wall Street Journal. Methodology: Con- ducted by Hart and McInturff Research Companies, November 1 - November 3, 2012 and based on 1,800 telephone interviews. Sample: National registered voters. The source for the median personal income in 2012 figures is the Bureau of Business & Economic Research. Per Capita Personal Income by State, BUREAU OF BUS. & ECON. RESEARCH (Apr. 2, 2013), https://bber.unm.edu/ econ/us-pci.htm.

   

% Republicans % Democrats % College degree

40% -- 43% --

22% 57%

36% 43% 48% 51%

23% 20% 41% 42% 42% 43%

Median income
% Male Median age % White

$35k- $20k- $40k 40k 50% 41% 55-59 32 82% 66%

--

$40k- $40k- $50k $50k 50% 50% 50-54 55-59 81% 82%

$20k- $20k- $42,693 $40k $40k
48% 55% 49%
30 33 36

CARAVAN Mturk 1 (2011) (6/16- 20/15)

MTurk 2 (7/23/15)

CARAVAN CARAVAN (2007) (2009)

MTurk3 MTurk4 (3/1- (2/14- 3/15) 15/15)

Nation as a whole1

38% 49%

51%

39% 40%

46% 53% 28%

Appendix Table A1: Demographics of Participants in the Surveys Reported in the Main Text.227

52% 32 76%

77% 78% 72%

616 The University of Memphis Law Review Vol. 46 II. DETAILS OF STATISTICAL ANALYSIS IN PART III

OF THE MAIN TEXT228

Part III in the main text discussed inequality in military sac- rifice. In this section of the Appendix we present the details of our analytic strategy to examine the relationship between a communi- ty’s socio-economic status and its share of American war casual- ties in World War II, Korea, Vietnam, and Iraq. We elaborate on the data and present the full results of the statistical models used to generate the figures and tables presented in the main text.

A. Data

To assess whether casualty gaps emerged in each of our na- tion’s last four major wars, we had to construct measures of com- munity casualty rates across the country. This involved first de- termining the total number of casualties—herein defined as sol- diers killed in wartime in combat zones—that came from each county or place for each war, and then dividing those casualties by the relevant county or place population to obtain a per-capita casu- alty rate.229

228. Additional details of related analyses are available in The Casualty Gap, including an additional round of analyses modeling Vietnam casualty rates at the place level as a robustness check on our county-level results. DOUGLAS L. KRINER & FRANCIS X. SHEN, THE CASUALTY GAP: THE CAUSES AND CONSEQUENCES OF AMERICAN WARTIME INEQUALITIES 136–60 (2010).

229. We also ran models using casualty counts as dependent variables and controlling for county or place population as an independent variable. While there were inconsistencies across model specifications, the patterns largely mir- rored those we discuss here. The inconsistencies across specification are likely due to the fact that population and casualty counts are very highly correlated and hence multicollinearity is a problem in the count models. Because complete individual-level data on wounded soldiers is unavailable for several conflicts, we limit ourselves to examining soldiers killed in action. To keep our focus on those soldiers killed in the theater of war, we limited our casualty counts for the Korean War to those soldiers who died between June 1950 (as North Korean forces invaded South Korea on 6/24/50) and July 1953 (as the Military Armi- stice Agreement was signed on 7/27/53). This only dropped 42 observations. For Vietnam, we limited our casualty counts to those soldiers who died between August 2, 1964 (when the U.S.S. Maddox was first attacked in the Gulf of Ton- kin) and March 29, 1973 (when the last U.S. soldiers left Vietnam). Again, this included almost all soldiers in the casualty files.

2016 Invisible Inequality Appendix 617

We obtained raw casualty data on individuals killed in World War II, Korea, and Vietnam from a series of casualty data- bases maintained by the United States National Archives.230 For Iraq and Afghanistan, we used data made publicly available by the Statistical Information Analysis Division (“SIAD”) of the Depart- ment of Defense.231 Our casualty data for Operation Iraqi Freedom includes all soldiers killed through December 31, 2008. Our casu- alty data for Operation Enduring Freedom includes all soldiers killed through July 4, 2011. These data files, for each war, provid- ed individual casualty records with information on the deceased soldier’s home of record prior to entering the armed forces.232 We

230. For World War II, we used the World War II Honor List of Dead and Missing Army and Army Air Forces Personnel. See World War II Honor List Dead and Missing Army and Army Air Forces Personnel, NAT’L ARCHIVES (June 1946), http://www.archives.gov/research/arc/ww2/army-casualties/. The vast majority of casualties come from this Army and Air Force datafile. Id. The data on Korea and Vietnam deaths and casualties comes from databases ar- chived by the United States National Archives as part of its Access to Archival Databases (AAD) System. All data was downloaded (first in summer 2005 and subsequently in early 2009 after minor file updates) from the AAD website. Access to Archival Databases, NAT’L ARCHIVES, http://www.archives.gov/aad/ (last visited Mar. 17, 2016). For Korea, we utilized the “Records of Military Personnel Who Died as a Result of Hostilities During the Korean War, ca. 1977 - 11/1979.” The database was created by the Department of Defense, Direc- torate for Information Operations and Reports, Manpower Management Infor- mation Division. For Vietnam, we used the “Records with Unit Information on Military Personnel Who Died During the Vietnam War, created ca. 1983 - 12/18/2005, documenting the period 6/8/1956 - 10/10/2003” (COFFELT file) and the “Records of Deceased, Wounded, Ill, or Injured Army Personnel, In- cluding Dependents and Civilian Employees, 1/1/1961 - 12/1981.” The first database is maintained by the Department of Defense, Washington Headquarters Services, Directorate for Information Operations and Reports, Statistical Infor- mation Analysis Division. The second database was created by the Adjutant General’s Office.

231. This data is now maintained and publicly available through the De- fense Manpower Data Center. Defense Casualty Analysis System, https://www.dmdc.osd.mil/dcas/pages/casualties.xhtml (last visited Mar. 23, 2016).

232. The COFFELT database tracking Vietnam casualties provides home state and city, not county (which is the lowest geographical unit for which com- plete 1970 census data is available), information for each casualty. Aggregating from the city to county level generally posed few problems, as we were able to assign counties based on cross-referenced census data. For some cities, addi-

618 The University of Memphis Law Review Vol. 46

then tallied these individual casualties by the smallest possible ge- ographic unit for which both casualty and complete census data was readily available. This was the county level for World War II, Korea, and Vietnam, though in Vietnam we also had a sub-national sample of place-level data to use for limited analyses. We were able to use national, comprehensive place-level data for our anal- yses of fatal casualties in Iraq and Afghanistan.233 For the Iraq and

tional steps were necessary. For single cities such as New York City, which span two or more counties in a single state, we followed two methods. The first method, which we used in all of the statistical analyses in this Article, evenly divided such casualties for each city among all of the counties it spanned. The second method assigned each casualty to each county spanned by the city under the premise that deaths from a city spanning multiple counties could affect resi- dents of all counties involved. The results across specifications for both our analyses are virtually identical regardless of which casualty rate operationaliza- tion is used. For towns such as Bethlehem, Pennsylvania, for which there is more than one city of the same name in a single state (less than five percent of the total), we also used two methods. First, we dropped all such casualties and ran our models. We then ran alternative models in which we randomly assigned each casualty to one of the towns. The two methods yielded nearly identical results. To construct the casualty rates presented we employed the first method.

233. The United States Census Bureau defines place as it refers to “Census Designated Place.” The U.S. Census Bureau defines a “Census Designated Place” as a place “delineated to provide data for settled concentrations of popu- lation that are identifiable by name but are not legally incorporated under the laws of the state in which they are located.” UNITED STATES CENSUS BUREAU: GEOGRAPHIC TERMS AND CONCEPTS – PLACE, https://www.census.gov/geo/ reference/gtc/gtc_place.html (last updated Dec. 6, 2012). The Census Bureau notes that, “An incorporated place usually is a city, town, village, or borough, but can have other legal descriptions. . . . exclude[ing] Boroughs in Alaska . . . Towns in the New England states, New York, and Wisconsin . . . [and] Bor- oughs in New York.” Id. The Census Bureau further distinguishes between “four major ‘groups’ that differentiate between populated places, other geopolit- ical and census units, institutional facilities, and terminated entries. Some sub- classes relate an entry to a class different from its own, which is useful because a number of entries serve in more than one capacity.” Appendix J: FIPS Class Code Definitions, http://webapp1.dlib.indiana.edu/virtual_disk_library/ index.cgi/4293203/FID1358/DOC/ASCII/APP_J.ASC (last visited Mar. 23, 2016). Because “some sub-classes identify entries in different classes that are coextensive,” we use as our unit of analysis the major group: Class-C, incorpo- rated places. Id.

2016 Invisible Inequality Appendix 619

Afghanistan wounded analysis, we used the number of wounded in action, by county.234

Once we determined the total number of casualties per lo- cale, we calculated the casualty rate by dividing through by a male population denominator to control for the significant variation in size across counties and places in the country. To make the casual- ty rate more accessible, we then multiplied the per-capita rate by 10,000.235

Using these measures, we are able to examine the relation- ships between a community’s local casualty rate and its demo- graphic characteristics, including its unemployment rate, median income, level of educational attainment, racial composition, rural farm population, median age, partisan composition, and geogra- phy. To operationalize these community demographics, we turned to various years of data publications by the United States Census Bureau.236 The data collected in the decennial censuses are well timed to capture the demographic characteristics of the counties from which the wartime casualties occurred. We were able to match 1940 census data with WWII casualty data, 1950 census data with Korean casualties, 1970 census data with Vietnam casu- alties, and 2000 census data with the present conflicts in Iraq and Afghanistan.237

To measure income, we use median family income in all but our World War II models. For World War II, where the meas-

234. Place-level data was not available for analysis of non-fatal casualties.

235. We also considered alternative models using other variables (e.g.

number of males age 18–34) as the denominator. The results were nearly identi- cal, as there were very high correlations between all of the alternative population denominator variables.

236. For the 1940, 1950, and 1970 census data, we utilized data files pre- pared by Michael Haines (2004) and published by the Inter-university Consorti- um for Political and Social Research (ICPSR). Michael R. Haines, Historical, Demographic, Economic, and Social Data: The United States, 1790-2002 (ICPSR 2896), DATA SHARING FOR DEMOGRAPHIC RESEARCH, https://www.icpsr.umich.edu/icpsrweb/DSDR/studies/2896 (last visited Mar. 17, 2016). For the 2000 census data, we downloaded raw summary file 3 (sf3) files from the Census Bureau website and built a place-level database for analysis. U.S. CENSUS BUREAU, SUMMARY FILE 3: 2000 CENSUS OF POPULATION AND HOUSING (2007), https://www.census.gov/prod/cen2000/doc/sf3.pdf.

237. For analysis of the Vietnam conflict, we used 1960 census data in- stead of 1970 census data yields virtually identical substantive results.

620 The University of Memphis Law Review Vol. 46

ure was unavailable in the Census dataset, we use the very similar measure of median rent per month.238 Education measures for all years were highly correlated with measures of income; as a result, we estimated separate income and education models. To measure partisanship, we included a measure of the percentage of county residents who voted for the Republican presidential candidate in the election immediately preceding each war: Wendell Willkie in 1940, Thomas Dewey in 1948, Barry Goldwater in 1964, and George W. Bush in 2000.239 Because we do not have this partisan- ship variable measured at the place level, in the place-level models we included the state percentage for Bush in 2000. The coeffi- cients and significance for the socioeconomic variables, however, were not sensitive to inclusion of this state level partisanship measure. To capture regional variation, we include a South re- gional dummy variable.240 Information on demographic variables is available in Kriner and Shen (2010).

238. Median rent is the same measure that Schaefer and Allen used. See Janet Schaefer & Marjorie Allen, Class and Regional Selection in Fatal Casual- ties in the First 18-23 Months of World War II, 23 SOCIAL FORCES 165–69 (1944). Median rent correlates with 1950 median income at .84 (with signifi- cance of p<.001). Alternative models were run using average and median value of owner-occupied dwellings. Because median and average value of owner- occupied dwellings correlate at .8 (p<.001) with median rent, the results were substantively the same. Finally, we also re-estimated the models using 1950 median income figures, which produced virtually identical results.

239. County-level returns for the 1940 and 1948 elections were obtained from the United States Historical Election Returns, 1824–1968 data file. United States Historical Election Returns, 1824-1968 (ICPSR 1), ICPSR (Apr. 26, 1999), http://www.icpsr.umich.edu/icpsrweb/ICPSR/studies/1. Returns for the 1964 election were obtained from the General Election Data for the United States, 1950-1990. General Election Data for the United States, 1950-1990 (ICPSR 13), ICPSR (Nov. 22, 2013), https://www.icpsr.umich.edu/ icpsrweb/ICPSR/studies/13.

240. The U.S. Census Bureau identifies four census regions: Northeast, Midwest, South, and West. Census Bureau Regions and Divisions with State FIPS Codes, CENSUS.GOV, http://www2.census.gov/geo/docs/maps- data/maps/reg_div.txt (last updated Mar. 17, 2016). The South region includes Delaware, Washington, D.C., Florida, Georgia, Maryland, North Carolina, South Carolina, Virginia, West Virginia, Alabama, Kentucky, Mississippi, Ten- nessee, Arkansas, Louisiana, Oklahoma, and Texas. Id.

2016 Invisible Inequality Appendix 621 B. Details of the Statistical Model

Having prepared our casualty and demographic databases, we developed county/place-level regression models. For each of the four conflicts, our dependent variable in all of the models is the casualty rate per 10,000 males. Our analysis is truly national, with virtually every county or place included for each war.241 Our inde- pendent variables are the eight demographic measures, including either income or education (but not both simultaneously) in each regression.242 Because our observations are clustered by state, we also cluster on the state and employ robust standard errors. The general form of our regression model is:

[1] CASUALTY_RATEi = β0 + β1UNEMPLOYMENTi + β2INCOMEi + β3AFR- AMERICANi + β4FARMi + β5AGEi + β6GOP_PREZi + β7SOUTHi + εi 243

Results for regression analyses of casualties suffered in World War II, Korea, and Vietnam are presented in Kriner and Shen (2010, 52). Here, we update our earlier analyses to include an assessment of casualty inequality in the war in Afghanistan. Regression results for the Iraq War, the war in Afghanistan, and the two combined, are reported in Table A2. Consistent with the alternative analyses presented in the article itself, this regression analysis shows a strong and significant negative relationship be- tween a community’s median family income and its casualty rate across both wars.

241. In models that included county-level partisanship measures, the coun- ties from Alaska were dropped because Alaska reports its election returns by election district and not county.

242. We avoid including both education and income measures in the same regression due to problems of multicollinearity. The two variables are very highly correlated at the county and place levels and, following standard practice in many economic analyses, we choose to run separate models.

243. εi is an error term, and the other variables in the model are defined as discussed above.

622 The University of Memphis Law Review Vol. 46

Appendix Table A2. Results of Ordinary Least Squares Regression Analysis of Casualty Rates, Iraq and Afghanistan

 

Iraq Afghanistan

UNEMPLOYMENT 0.421 -3.628 (2.308) (2.601)

INCOME -0.015*** -0.005** (0.003) (0.002)

RACE -1.000*** -1.011 (0.361) (0.627)

RURAL 3.271 -3.095*** (2.616) (0.985)

AGE -0.019 -0.021 (0.012) (0.015) PARTISANSHIP -1.847 -2.621* (1.125) (1.328)

SOUTH REGION 0.390** 0.713* (0.187) (0.423)

CONSTANT 3.397*** 2.845** (0.720) (1.356)

Observations 19,413 19,413 R-squared 0.001 0.001

Iraq +Afghanistan

-4.049 (3.329) -0.020*** (0.004) -2.011*** (0.730) 0.177 (2.453) -0.040** (0.020) -4.469* (1.696) 1.104** (0.442) 6.242*** (1.461)

19,413 0.001

  

Notes on Table A2: Robust
significance tests are two-tailed, and significance is indicated as follows: *** p<0.01, ** p<0.05, * p<0.10

standard errors are reported in parentheses. All

2016 Invisible Inequality Appendix 623 III. THE CHALLENGE OF ECOLOGICAL INFERENCE

The analyses presented in Part III of the main text and just described in further detail above show strong evidence of a socio- economic casualty gap between rich and poor communities. That is, in recent wars communities with lower median incomes and levels of education have sustained casualty rates that are systemat- ically higher than those experienced by communities with higher median incomes and levels of educational attainment.

This casualty gap at the community level is normatively troubling. In all of the experiments in which we exposed subjects to information about a casualty gap presented in the text, we were explicit that this was a gap between rich and poor communities.

The logical question raised by the strong evidence for a casualty gap between rich and poor communities is whether a par- allel gap arises at the individual level; that is, are individuals from lower socioeconomic backgrounds dying disproportionately in America’s wars?

The two mechanisms described in Part III suggest a gap at the individual level. Military manpower scholarship has long es- tablished that the economic benefits military service affords are more attractive to young men and women from socioeconomically disadvantaged backgrounds who lack greater opportunity in the civilian job market. Moreover, new recruits from socioeconomi- cally disadvantaged backgrounds are more likely to lack the skills that help one be assigned to a position within the military that is more insulated from combat risks.

The most logical explanation for the community-level cas- ualty gap described in the text is that it results from a parallel gap at the individual level. However, from the analysis of aggregate- level data alone we cannot conclusively prove the existence of ine- qualities in sacrifice between rich and poor individuals. In The Casualty Gap, we undertook a number of additional rounds of analysis to seek more insight into this question.244 For example, we were able to exploit variation in casualty rates within a city to show that casualties hail from neighborhoods that are significantly poorer than the city average. These additional analyses are con-

244. KRINER & SHEN, supra note 228, at 40–47.

624 The University of Memphis Law Review Vol. 46

sistent with the hypothesis that socioeconomically disadvantaged Americans are more likely to die in military service than their peers with greater socioeconomic opportunity. However, even these analyses cannot overcome the ecological inference barrier completely. Without data on individual soldiers’ socioeconomic backgrounds prior to entering the military, we cannot conclude definitively that a casualty gap exists between individuals from socioeconomically advantaged and disadvantaged backgrounds. However, multiple rounds of empirical analysis suggest that that this is the most likely explanation for the patterns we observe be- tween rich and poor communities.

IV. DETAILS OF THE 2006 SENATE ELECTIONS ANALYSIS IN SECTION V.C

In Section V.B we examined the effect of casualties on po- litical outcomes. To test whether non-fatal casualties are indeed less politically visible and influential than fatal casualties, we ex- amined the influence of state-level fatal and non-fatal casualty rates on the electoral fortunes of Republican candidates in the 2006 Senatorial elections. The Iraq War was extremely salient in 2006. According to a national exit poll of more than 13,000 Americans, 67% of voters answered that the Iraq War was either extremely or very important to their vote choice.245 In that election, Democrats nationwide ran on a “Six for ’06” plan. The plan’s final item promised a “significant transition” in Iraq, and indeed, in the very first months following the Democratic takeover of Congress the Democratic majority pushed through a Defense budget that en- deavored to mandate a timetable for withdrawal from Iraq. As such, the 2006 midterms present a critical case in which to look for the electoral ramifications of non-fatal casualties.

Data on soldiers killed in the Iraq War and their home state of residence is made publicly available by the Department of De- fense. Data on non-fatal casualties is not reliably released, which

245. See 2006 Election Exit Polls, CNN.COM, http://www.cnn.com/ ELECTION/2006/pages/results/states/US/H/00/epolls.0.html (last visited Mar. 17, 2016). By contrast, in 2008 only ten percent of Americans listed Iraq as the most important issue guiding their choice between Obama and McCain. 2008 Election Exit Polls, CNNPOLITICS.COM, http://www.cnn.com/ELECTION/2008/ results/polls/#val=USP00p6 (last visited Jan. 22, 2016).

2016 Invisible Inequality Appendix 625

is one of the main reasons that the bulk of empirical scholarship on casualties and their consequences has focused exclusively on fatal casualties. However, through a Freedom of Information Act re- quest, we obtained data on the number of non-fatal Iraq War casu- alties from each state in each month of the war.246 We then divid- ed each state’s fatal and non-fatal casualty tally by the state’s pop- ulation as obtained from the 2000 Census to create state-level fatal and non-fatal casualty rates.

To illustrate the relationship between a state’s casualty rate (fatal or non-fatal) and Republican electoral fortunes graphically, Figure A1 reports a pair of scatter plots.

Appendix Figure A1: Scatter Plots of State Iraq War Fatal and Non-Fatal Rates and Change in GOP Vote Share,
2000 to 2006 Senatorial Elections

Fatal Casualties

IN

MA HI WV

MD ME NV

CA NJUTMO VTAXMSMT MN MI WI FLWA NM RI PA OH ND WY NY TN NE DE

CT AZ

VT

5 10 15 20 25 30 KIA per million residents

    

246. The data made available by the Department of Defense also provided some information at the county level. However, a significant percentage (ap- proaching a majority in the early years of the war) of the non-fatal casualties were reported as hailing from an “unknown” county within a state. It is im- portant to emphasize that the home state information refers not to where the soldier was based before deploying to Iraq, but his or her home of record upon applying to the armed forces. For an extended discussion of this data, we refer interested readers to KRINER & SHEN, supra note 5, at 178–79.

Change in GOP vote share
-30 -20 -10 0 10 20

626 The University of Memphis Law Review Vol. 46 Figure A1, continued: Non-Fatal Casualties

IN

MA HI WV

MD

NV

CA
NJ MSMO

ME

WY

MT

 

UVTA TX MIMN

FL WI

NM WA

ND

 

NY DE

RI PA

NE TN OH

CT AZ

VT

   

0 50 100 150 WIA per million residents

On the y-axis is the change in the Republican candidate’s vote share from 2000 to 2006. Using the change in vote share pro- vides an important measure of control as it allows us to assess the GOP’s performance in a state in 2006 against an earlier baseline. On the x-axis is the state’s casualty rate per million residents. The top panel illustrates the relationship for fatal casualties. The bot- tom panel illustrates the relationship for non-fatal casualties. Con- sistent with past research, there is a strong inverse relationship be- tween a state’s fatal casualty rate and how the Republican Party’s candidate fared in the 2006 Senate elections. Firmly tied in the public conscience to President Bush and his costly war parties, the incumbent party in power suffered significant losses in 2006, par- ticularly in the states that had suffered the most fatal casualties. The two are correlated at r = -.41, which is statistically significant, p < .02.

By contrast, as illustrated in the bottom panel, we see little evidence of a relationship between non-fatal casualty rates and Republican electoral fortunes. The two are only weakly related at r = -.18, and the correlation is not statistically significant (p = .35).

Change in GOP vote share
-30 -20 -10 0 10 20

2016 Invisible Inequality Appendix 627

At first blush, non-fatal casualties do not seem to impose the same political costs on incumbents as fatal casualties.

To explore this relationship further, we replicated an earlier analysis of the 2006 Senate elections but this time include both fatal and non-fatal casualty rates in the statistical model.247 This approach allows us to assess the relationship between a state’s fatal and non-fatal casualty rates and GOP electoral fortunes after con- trolling for a number of alternative factors that could be driving variation in Republican vote share.

Part V, Section C reports the conclusion that non-fatal cas- ualties did not affect electoral outcomes in the same way that fatal casualties did. Here we present that regression analysis that led to that conclusion.

To examine the relationship between a state’s non-fatal casualty rate and electoral outcomes, we specified a regression model that included a series of important control variables. In ad- dition to casualties, an extensive literature has identified opponent quality and campaign spending as two of the most important pre- dictors of a candidate’s electoral fortunes.248 To account for

247. Douglas L. Kriner & Francis X. Shen, Iraq Casualties and the 2006 Senate Elections, 32 LEGIS. STUD. Q. 507 (2007). In this earlier article, we ex- amined the relationship between local casualty rates and GOP Electoral fortunes at both the state and county level. Here, however, we focus only on the state level because county home of record information is reported as “unknown” for a significant percentage of Iraq War non-fatal casualties in the DOD data.

248. For opponent quality, see GARY JACOBSON, POLITICS OF CONGRESSIONAL ELECTIONS (2004); Donald Philip Green & Jonathan S. Kras- no, Rebuttal to Jacobson’s “New Evidence for Old Arguments,” 34 AM. J. POL. SCI. 363 (1990); Donald Philip Green & Jonathan S. Krasno, Salvation for the Spendthrift Incumbent: Reestimating the Effects of Campaign Spending in House Elections, 32 AM. J. POL. SCI. 884 (1988); and Peverill Squire, Challeng- er Quality and Voting Behavior in Senate Elections, 17 LEGIS. STUD. Q. 247 (1992). For campaign spending, see Alan I. Abramowitz, Explaining Senate Election Outcomes, 82 AM. POL. SCI. REV. 385 (1988); Alan I. Abramowitz, Campaign Spending in U.S. Senate Elections, 14 LEGIS. STUD. Q. 487 (1989); Alan Gerber, Estimating the Effect of Campaign Spending on Senate Election Outcomes Using Instrumental Variables, 92 AM. POL. SCI. REV. 401 (1998); Gary C. Jacobson, The Effects of Campaign Spending in Congressional Elec- tions, 72 AM. POL. SCI. REV. 769 (1978); Gary C. Jacobson, The Effects of Cam- paign Spending in House Elections: New Evidence for Old Arguments, 34 AM. J. POL. SCI. 334 (1990); and Gary C. Jacobson, Money and Votes Reconsidered: Congressional Elections, 1972–1982, 47 PUB. CHOICE 7 (1985). An additional

628 The University of Memphis Law Review Vol. 46

changes in opponent quality, we coded each Republican’s oppo- nent according to the eight-point ordinal scale created by political scientists Don Green and Jonathan Krasno, and we calculated the change in this measure across the two electoral cycles. To control for the influence of campaign expenditures, we include the change in the percentage of total campaign expenditures by the Republican candidate from 2000 to 2006.249

political factor that may have influenced the change in GOP vote share is any change in the incumbency status of the Republican candidate from the 2000 to the 2006 campaign. All of the models were re-estimated with two dummy vari- ables that indicate whether the GOP candidate went from being a challenger (either facing an incumbent or vying for an open seat) to an incumbent from 2000 to 2006 or vice versa. All of our results remain virtually identical in this expanded specification. These augmented models show the expected negative relationship between a shift from incumbent to challenger status and GOP vote share at both the state and county levels. A complementary shift from challeng- er to incumbent status, however, had no effect at the state level and, contra ex- pectations, a negative correlation with the change in GOP vote share at the coun- ty level. The relationship is almost certainly spurious. Only three states in- volved a Republican challenger from 2000 (2002 for James Talent) running in 2006 as an incumbent—Virginia, Nevada, and Missouri. In the Virginia race, George Allen lost to James Webb; in Nevada, John Ensign handily beat Jack Carter, but not by the same margins as he trounced his Democratic opponent, who lacked a presidential name, in 2000; and the Missouri races were decided by razor-thin margins in 2000, 2002, and 2006. A confluence of national trends and idiosyncratic factors, not any change in incumbency status, determined the results of these three elections.

249. Green & Krasno, Salvtion for the Spendthrift Incumbent, supra note 248. Because Green and Krasno’s scale was designed to measure challenger quality, it required one minor modification. If the Republican candidate faced an incumbent senator, we coded the opponent quality score at its maximum value of 8. Prior studies have adopted varied operationalizations of relative campaign spending. To control for several outliers in Republican-opponent spending, in this model we took the log of both major candidates’ expenditures as reported to the Federal Elections Commission and calculated the percentage of this total spent by the Republican. All of our results are robust across other operationalizations, such as the change in the percentage of unlogged total ex- penditures spent by the Republican candidate and the change in the ratio of Re- publican to Democratic spending. Following Jacobson, Green and Krasno, and others, we recoded the handful of missing expenditure data points as $1,000. All of these were minor, dark-horse candidates with little in the way of a formal campaign apparatus.

2016 Invisible Inequality Appendix 629

In addition to factors specific to the Senate race at hand, scholars have long documented the connections between presiden- tial performance and the success of their co-partisans in presiden- tial elections, even in midterm contests.250 To account for this in the current context, we include a measure of President Bush’s share of the two-party vote in each state in the 2004 election. Ad- ditionally, a number of previous studies have debated the relative impact of economic conditions on congressional election out- comes.251 To control for economic factors, we include measures, obtained from the Bureau of Labor Statistics, of the change in the state unemployment rate during the year preceding the 2006 mid- term elections. Voters in areas of increasing unemployment may be more likely to punish Republican candidates in this era of uni- fied Republican control of Congress and the presidency.

Finally, the models also control for two important demo- graphic constituency characteristics that might be correlated with considerable change in Republican electoral fortunes from the peacetime election of 2000 to the wartime 2006 contest: the per- centage of residents aged 18 to 64 who were serving in the military and the percentage of all residents who were veterans of the armed forces.252 Conventional wisdom suggests that military communi- ties have largely rallied around the president and his policies; if correct, Republican candidates may have performed better relative

250. Alan I. Abramowitz & Jeffrey A. Segal, Determinants of the Out- comes of U.S. Senate Elections, 48 J. POL. 433 (1986); Lonna Rae Atkeson & Randall W. Partin, Economic and Referendum Voting: A Comparison of Guber- natorial and Senatorial Elections, 89 AM. POL. SCI. REV. 99 (1995); James E. Campbell, The Presidential Surge and its Midterm Decline, 1868-1988, 53 J. POL. 477 (1991); James E. Campbell & Joe A. Sumners, Presidential Coattails in Senate Elections, 84 AM. J. POL. SCI. 513 (1990); Thomas M. Carsey & Ger- ald C. Wright, State and National Factors in Gubernatorial and Senatorial Elections: A Rejoinder, 42 AM. J. POL. SCI. 1008 (1998).

251. See, e.g., GARY JACOBSON, GARY & SAMUEL KERNELL, STRATEGY AND CHOICE IN CONGRESSIONAL ELECTIONS (1981); MICHAEL S. LEWIS-BECK & TOM W. RICE, FORECASTING ELECTIONS (1992); Alberto Alesina & Howard Rosenthal, Partisan Cycles in Congressional Elections and the Macroeconomy, 83 AM. POL. SCI. REV. 373 (1989); Peverill Squire, Candidates, Money and Voters: Assessing the State of Congressional Elections Research, 48 POL. RES. Q. 891 (1995).

252. These demographic controls were constructed from data obtained from the U.S. Census Bureau’s summary files (sf3) for the 2000 Census.

630 The University of Memphis Law Review Vol. 46

to their 2000 baseline in these areas than in otherwise comparable communities. Additionally, an extensive literature at the elite level has examined the different perspectives that veterans bring to ques- tions of military policy; however, expectations for electoral behav- ior in states or counties with large veteran contingents at the mass level are less clear.253 It is possible that communities with large contingents of veterans, like those with high percentages of active- duty personnel and their families, rallied around the president and the Republicans in the 2006 midterms; alternatively, residents of such communities may have viewed the war and the administra- tion’s military policies through a distinctly different and more crit- ical lens and adjusted their voting behavior accordingly. The em- pirical affords insight into these competing hypotheses.

Table A3 presents the results of our linear regression mod- el.254 The model in column 1 includes only the fatal casualty rate variable, and the relevant coefficient is negative and statistically significant. Confirming the bivariate relationship illustrated in the top panel of Figure 1, Republican electoral fortunes declined as a state’s fatal casualty rate increased, even after controlling for a host of other factors long held to influence election outcomes. Model 2 estimates the same specification, but examines the rela- tionship between non-fatal casualty rates and the change in Repub- lican vote share. The relevant coefficient is negative, but substan- tively very small and it is not statistically significant. Finally, model 3 re-estimates our model but includes both fatal and non- fatal casualty rates simultaneously. The coefficient for fatal casu- alty rates remains strongly negative and highly statistically signifi-

253. On the civil-military gap, see, e.g., Samuel P Huntington, Conserva- tism as an Ideology, 51 AM. POL. SCI. REV. 454 (1957); RICHARD K. BETTS, SOLDIERS, STATESMEN, AND COLD WAR CRISES (1991); PETER FEAVER & RICHARD H. KOHN, SOLDIERS AND CIVILIANS: THE CIVIL-MILITARY GAP AND AMERICAN NATIONAL SECURITY (2001); and PETER D. FEAVER & CHRISTOPHER GELPI, CHOOSING YOUR BATTLES: AMERICAN CIVIL-MILITARY RELATIONS AND THE USE OF FORCE (2004).

254. In Kriner & Shen, supra note 247, at 514, 526 n.11, we drop Con- necticut and Vermont, as Joe Lieberman and Jim Jeffords switched partisan affiliation after 2000. Replicating the models in Table 1 excluding these two states yields virtually identical results. The coefficient for fatal casualty rates remains strongly negative and statistically significant in both models 1 and 3. The coefficient for non-fatal casualty rates is positive in this specification in both models 2 and 3.

2016 Invisible Inequality Appendix 631 cant. By contrast, the coefficient for a state’s non-fatal casualty

rate is small, positive, and not statistically significant.

Appendix Table A3: Results of Ordinary Least Squares Regression Analysis of State Iraq War Fatal and Non-Fatal Casualty Rates and 2006 GOP Senatorial Electoral Fortunes

(1) (2)

KIA per million residents -1.36***
(0.41) (0.50)

WIA per million residents -0.10
(0.10) (0.10)

% Bush 2004

Change in opponent quality

Change in GOP spending Change in unemployment % in Military
% Veterans

Constant

Observations R-squared

0.60** 0.48 (0.26) (0.30)

-0.03 -0.18

(0.71) (0.84) 0.25 0.24 (0.15) (0.18) 14.03** 9.72 (6.40) (7.39) 5.29* 4.73 (3.06) (3.58) 0.41 -0.11 (1.40) (1.78) -21.97 -20.18 (16.76) (20.11)

33 33 0.44 0.22

0.57** (0.26)

-0.08

(0.72) 0.25 (0.15) 14.98** (6.58) 5.41* (3.09) -0.03 (1.53) -19.16 (17.30)

33 0.45

(3)

-1.57***

0.08

Notes on Appendix Table A3: Robust standard errors are reported in paren- theses. All significance tests are two-tailed, and significance is indicated as follows: *** p<0.01, ** p<0.05, * p<0.10

V. DETAILS OF STATISTICAL ANALYSIS IN SECTION V.D

In the main text, Part V, Section D discusses results from a web-based experiment exploring the effect of casualties on support

632 The University of Memphis Law Review Vol. 46

for war. In this section we discuss additional methodological de- tails about the experiment.

Concerns about subjects’ compliance with task instructions are of special concern with online experiments because subjects cannot be monitored while engaged in the experimental tasks.255 To address this issue, experimental psychologists have developed “attention filters” designed to ascertain whether subjects are in fact following instructions and paying attention to the material being presented to them online. In our experiment reported here, we em- ployed a modified version of the filter developed by psychologist Daniel Oppenheimer and his colleagues.256 The design of the pri- mary attention filter question was such that users who did not read carefully would see, in large font, a headline reading “Background Questions on Sources for News” as well as another large, bold question: “From which of these sources have you received infor- mation in the past month?” A series of check-box options were provided (e.g., local newspaper, local TV news). Subjects reading carefully, however, were instructed not to check any of the boxes, but instead to type “123” into the text box provided. The results presented in this Article are based only on the “good” subjects, i.e. those subjects who were paying attention.

As mentioned in the main text, after receiving one of the experimental prompts chosen at random all subjects were then asked the same question taken as previously utilized in published polls conducted by NBC News/The Wall Street Journal: “Do you think the war in Afghanistan against the Taliban and Al Qaeda has been very successful, somewhat successful, somewhat unsuccess- ful, or very unsuccessful?”

We employed this measure of public support for the Af- ghan War for both theoretical and practical reasons. First, a large literature has argued that popular belief about whether or not a mil- itary operation is succeeding is the linchpin of public support for

255. A filter employed after data collection allowed for the experiment to exclude from the dataset subjects with duplicate IP addresses.

256. See Daniel M. Oppenheimer et al., Instructional Manipulation Checks: Detecting Satisficing to Increase Statistical Power, 45 J. EXPERIMENTAL SOC. PSYCHOL. 867, 867–68 (2009) (describing a filter in which subjects must carefully read instructions which, counter to the boldface headline above the instructions, tell subjects not to actually click on an answer to the question).

2016 Invisible Inequality Appendix 633

war.257 Second, this question wording, with its explicit reference to the Taliban and Al Qaeda, has consistently generated higher levels of public support for the war than alternative question word- ings in national polls. It is important to emphasize that, if any- thing, our experimental study is biased against finding any effects for casualty information on support for the war. After more than a dozen years of fighting, most Americans have made up their mind on the conflict and are unlikely to be swayed by a modest prompt of new information about the conflict and its costs. If we had con- ducted this experiment earlier in the war when public opinion was more malleable, we would expect stronger effects. As such, select- ing a question wording that produces the strongest levels of ex ante support for the war affords the best estimates of the potential influ- ence of fatal and non-fatal casualty information on war support.

The main text presents basic difference in means results. However, to insure the robustness of our results, we also estimated an ordered logit regression model to assess the influence of each of our experimental treatments on beliefs about the war’s success, controlling for each individual respondent’s demographic charac- teristics.

The ordered logit model allows us to use our question’s full four-point answer range (very successful; somewhat successful; somewhat unsuccessful; very unsuccessful) as the dependent vari- able, with higher values equaling a more positive assessment of the war and its progress. Our main independent variables of interest are indicators for assignment to each experimental treatment, with treatment 1 (KIA information only) being the omitted baseline cat- egory.258 This allows us to see examine whether the three non- fatal casualties treatments and the final treatment presenting both fatal and non-fatal casualty information raised or lowered war sup- port above the KIA treatment baseline. Finally, our ordered logit model includes a number of standard demographic controls includ-

257. See, e.g., CHRISTOPHER GELPI, PETER D. FEAVER & JASON REIFLER, PAYING THE HUMAN COSTS OF WAR: AMERICAN PUBLIC OPINION ANDCASUALTIES IN MILITARY CONFLICTS (2009).

258. As discussed in the text, the experiment also had a control group that received no information about the number of fatal or non-fatal Afghan War cas- ualties. The coefficient for the variable identifying assignment to the control group is positive but not statistically significant. The only significant differ- ences are between the KIA treatments and the WIA treatments.

634 The University of Memphis Law Review Vol. 46

ing each respondent’s gender, race, and age, partisan affiliation, educational attainment, and two measures of religious affilia- tion.259 The results are presented in Table A4.

Even after controlling for a host of individual-level factors that might affect Americans’ assessment of progress in the war in Afghanistan, we continue to find a significant gulf between those informed of the number of Americans who have died in Afghani- stan compared to those told only about the number of American soldiers wounded in the war. The coefficient for our WIA treat- ment (17,674 soldiers with physical wounds) is positive and statis- tically significant. Subjects in this treatment were significantly more likely to judge the war a success than those who learned the total number of American soldiers killed in Afghanistan. The co- efficients for the two additional wounded treatments providing either a much larger estimated number of non-fatal casualties (treatment 3) or this number with a brief explanation about the “in- visible” wounds of war (treatment 4) are also positive, and not sig- nificantly different from the WIA coefficient. Finally, the coeffi- cient for treatment 5, which informed subjects of both the number of fatal-casualties and gave them the full accounting of non-fatal casualties is significantly smaller than both the WIA coefficient and the coefficient for treatment 4, which provided subjects exactly the same information except for the number of KIAs in the war.260 Thus, the ordered logit regression analysis yields the same conclu- sion as the simple assessment of differences in means: fatal casu- alties have a significantly greater impact on popular assessments of the war in Afghanistan than non-fatal casualties.

259. Roughly three quarters of our sample identified as being Catholic, Protestant, or having no religion. Of the remaining quarter, the vast majority identified as “other,” with only four percent of the entire sample identifying as Jewish or Muslim. Catholics and atheists were the only two groups whose war assessments differed from others, on average.

260. Wald tests confirm that the coefficient for treatment 5 is significantly smaller than the coefficient for treatment 4, p = .06.

2016 Invisible Inequality Appendix 635

Appendix Table A4. Results from Ordered Logit Analysis of Beliefs about Afghan War Success

No casualty information

WIA (17,674)

Total wounded (217,674)

Total wounded + PTSD note

KIA (2,312) + Total wounded + PTSD note

Republican Democrat
Male
Age
College graduate White

Atheist Catholic

Observations

0.362 (0.348) 0.677** (0.338) 0.328 (0.327) 0.479 (0.378)

-0.182

(0.365) 0.370 (0.293) 0.107 (0.236) 0.297 (0.216) -0.007 (0.009) 0.102 (0.221) -0.469* (0.272) -0.773*** (0.259) 0.770*** (0.296)

337

Notes on Table A4: Robust standard errors are reported in parentheses. All significance tests are two-tailed, and significance is indicated as follows: *** p<0.01, ** p<0.05, * p<0.10