ONGOING RACIAL DISPARITIES

Disparities between Whites and Blacks continue to plague American society. Relative to Whites, Blacks fall on the negative side of a wide array of important social indicators, with infant morality rates 146% higher; life chances of imprisonment (state or federal facility) 447% higher; rate of death by homicide 521% higher; lack of health insurance coverage 42.3% more likely; median income rate 55.3% lower; poverty rates 173% higher; and proportion with a college degree or beyond 59.5% lower. Strikingly, the average White American will even live five and one-half years longer than the average Black American (seven years for males).³

White privilege has been likened to an “invisible knapsack” that Whites carry with them wherever they go (McIntosh 1988). The knapsack represents the hidden benefits that accrue to Whites as opposed to members of other races. For example, McIntosh argues that Whites can ask for the person in charge and be more likely to encounter someone of their own race. To take another example, Whites do not have to worry that the mere color of their skin will work against the appearance of financial responsibility. Whites have learned to ignore these privileges conferred by race (McIntosh 1988).

One of the most important racial disparities, and one that relates to each of those in the preceding paragraphs, is the racial wealth gap. An analysis of economic disparity has shown that the overall objective wealth disparity between White and Black heads of household is about $150,000 (Wolff 2001). The ratio between Black wealth and White wealth (about 0.18 overall) varies only trivially when controlling for a range of important factors such as marital status, age, and level of education of the household head. The wealth gap is consistent even across differences in income level between Blacks and Whites. In other words, even when focusing on Blacks in the top income quartile and Whites in their respective top income quartile, there is still a one-to-five wealth gap in favor of Whites. In addition, Black/White net worth ratios have remained relatively unchanged for more than twenty years.

It is not difficult to trace most racial disparities, especially the racial wealth gap, back to legalized slavery and the roughly hundred years of ensuing de jure and de facto discrimination, which still prevail in various forms to this day. Consistent with this claim, the wealth disparities identified by Wolff have been attributed primarily to starting off on an unequal footing; the disparities are negligibly attributable to other factors (such as differential savings rates). Wolff concludes that, at current rates, “decades would be required for the wealth gap to close” (Wolff 2001, p. 26; see also Oliver and Shapiro, 1995, p. 4).

REPARATIONS: BACKGROUND AND EXPLANATIONS FOR OPPOSITION

The history of the reparations movement is a long one (Coates 2004, pp. 845–847). Popular monographs have made strong and passionate cases for reparations (Bittker 1973; Robinson 2001). Currently, there are numerous lawsuits, either underway or being planned, aimed at companies that benefited from slavery (Kong 2002; Rigby 2002), as well as at the U.S. government, which not only benefited from slavery, but played an active role in shaping and maintaining it (Thornton and Yanochik, 2003). At the same time, there is also considerable resistance to reparations. For example, Rep. John Conyers (D-MI) has proposed a bill (H.R. 40) to Congress every year since 1989 that would provide a formal apology for slavery, and would study the ongoing effects of slavery and discrimination, as well as appropriate remedies. This bill has yet to make it out of committee, even though it would not actually provide reparations. Resistance to Conyer's legislation is symptomatic of broader opposition to reparations among the White majority.⁴

Legal attempts to pursue reparations have been discussed by class-action specialists; the proposed suits would seek reparations on behalf of slave descendants (Hitt et al., 2000). The discussion has tackled questions such as expiration (statutes of limitation), appropriate defendants, and the profiling of persuasive plaintiffs. Most importantly, the discussion has reflected a consensus that, in order to be effective, a reparations class-action suit must be supported in the court of public opinion, and, furthermore, the damages must be quantified (Hitt et al., 2000, pp. 44, 46). Hence, research quantifying restitution in the court of White public opinion would be central to seeking reparations in the real world.

Resistance to reparations is also relevant to basic research on the Principle-Implementation Gap (P-I Gap) (Schuman et al., 1985). The P-I Gap describes the phenomenon by which most people support egalitarian ideals such as equal opportunity, fairness, and diversity, while at the same time opposing many of the specific policies designed to deliver these ideals (e.g., affirmative action, slave-descendant reparations). Two rather obvious explanations for the P-I Gap with respect to reparations are racism (Sears and Jessor, 1996) and economic self- or group-interest (Tuch and Hughes, 1996). Certainly these factors still play an important role in opposition to programs like reparations or affirmative action; however, there are a number of additional explanations that may explain resistance to reparations above and beyond these two factors.

One of the most popular explanations for the P-I Gap is based on beliefs and attitudes relating to social stratification (Kluegel and Smith, 1983; 1986). For example, to the extent that one's place in society is seen as a reflection of work ethic (i.e., assuming meritocracy), many social disparities can be justified (Son Hing et al., 2002). Work on social dominance orientation (Pratto et al., 1994) has examined the belief that some groups are inherently superior to others and deserve a greater share of the good things in life. Finally, there is evidence that many attitudes and beliefs regarding social stratification are based on justifying the status quo, whatever that might be [i.e., system justification theory; see Jost et al., (2004)].

There are several additional reasons that a given individual may oppose reparations. Opposition to reparations may, for instance, be based on political ideology. Conservatives may oppose programs like reparations and affirmative action to the extent that they are construed as unnecessary governmental interventions (Sniderman and Tetlock, 1986). And even one who is in favor of reparations in the abstract may disagree with the specifics of a given reparations plan (Kravitz 1995). Finally, some have suggested that many people do not really endorse egalitarianism when confronted with the full implications of the concept (Jackman 1996).

We focus here on yet another explanation for the P-I Gap: many Whites may desire racial equality, but feel that sufficient progress has already been made on that front, thus rendering reparations unneeded. In other words, although a primary focus of reparations is to redress past injuries, support may depend on perceptions of present-day inequalities that are the persisting results of past injustices. Kluegel and Smith (1983, 1986) and others (Swim and Miller, 1999; Tuch and Hughes, 1996) have previously demonstrated that perceptions of ongoing discrimination and disparities are positively correlated with support for programs such as affirmative action. However, that scholarship has now become somewhat dated. Not only have national attitudes regarding race continued to evolve, but major events (e.g., the September 11, 2001, terrorist attacks) have transpired that may influence attitudes toward intergroup disparities and related policy.⁵

In addition to possible differences in attitudes toward progressive racial policy in general, none of the prior mainstream social psychological work on perceptions of disparities has dealt directly with the issue of reparations. In fact, the topic of reparations has been widely ignored by social psychologists. An online literature review of major publication outlets within the field over the past twenty-five years yielded zero hits for the keyword reparations, although affirmative action generated fifty-five hits.⁶

We examined the relationship between Whites' perceptions of disparity (Black cost vs. White privilege) and support for slave-descendant reparations using a somewhat novel method. Typically, perceptions of racial disparity are measured with simple Likert items, or occasionally short scales (Swim and Miller, 1999). Dawson and Popoff (2004) measured perceptions that equality had been achieved (response options: equality achieved, equality soon, not in a lifetime, equality never), but found no association between responses to this item and support for slave-descendant reparations. Likert items can tend to be impersonal, however, and they may not foster elaborated cost/benefit considerations. Contingent valuation, in contrast, is designed to elicit cost/benefit comparison and, as such, is well suited to assess Whites' perceptions of Black cost.

CONTINGENT VALUATION

The essence of contingent valuation is to ask respondents to ascribe a value to a set of circumstances that is counterfactual, i.e., which does not presently exist: “Should X happen, what would you be willing to pay?” (Cummings et al., 1986, p. xi). Our scenario was the same, except that the respondent became the payee, hence, “If X happened to you, what should you be paid?”

Human beings are observed in practice to be willing to pay money—directly or through their government—for the preservation of environmental resources that they do not intend to use or exploit personally (Mitchell and Carson, 1989). Examples are the use of tax funds to set aside or preserve wilderness areas, contributions to endangered species, preservation of national parks, and so forth. We examined here whether the White majority could construe reparations as a contribution to a deserving human resource: the slave-descendant minority.

A contingent valuation approach to estimating White perceptions of Black cost seems more likely to elicit cost/benefit consideration than would be Likert scales. When considering Black costs, Whites may try to imagine themselves as Black and imagine how they would respond to resultant challenges. Specifically, we intended to estimate the amount of money that White participants would require to change their racial status to Black in a way that would relate to support for reparations. This estimate, then, could be used as an indicator of perceptions of racial disparity and would be expected to correlate positively with support for reparations.

A contingent valuation approach to perceptions of racial disparity was previously reported by Hacker, who asked his students to respond to a race-change scenario by answering this question: “How much financial recompense would you request?” (Hacker 2003, p. 42). In the Hacker scenario, students were asked to consider that they were supposed to have been born Black, but a mistake had been made, and they were actually born White:

Hacker reported that “Most seemed to feel it would not be out of place to ask for $1 million for each future year they would be living as a Black American” (Hacker 2003, p. 42). We speculate that a sizeable portion of these huge race-change remuneration requests could be due to having to change (traumatically?) one's appearance beyond recognition by friends and acquaintances, as well as the enormous imputed wealth/power of a mysterious donor organization that could guarantee targeted persons “to live another sixty years.” In addition, the extent to which Hacker's scenario encouraged rumination about racial disparities is unclear, because the scenario held that the race change was a forced one. At any rate, although Hacker never counted how many students responded to his parable and never tabulated their replies,⁷

In the studies below, we sought to sidestep problems in the scenarios used by Hacker. We asked White participants to imagine that they were actually Black, but had always passed as White. We then asked them to choose an amount that they would require to be paid to formally and publicly change their racial status to Black. This task should encourage a nuanced computation of the costs and benefits of being White versus Black, because an actual choice needs to be made. The hypothesized race change, however, requires no physical transformations, merely a change in public status.

Analogues of our race-change scenario are not without precedent. For example, famous civil rights activist Walter White looked White, but actually had significant Black ancestry. His decision to publicly “come out” as Black resulted in threats on his life (Janken 2005). Furthermore, new DNA technology that can detect Black ancestry could offer Whites a choice similar to the one presented in our scenario (Harmon 2006). Indeed, a recent Public Broadcasting Service (PBS) documentary showed Whites being told that they had varying levels of Black ancestry (Gates et al., 2006).

In addition to the race-change scenario (described in detail in Study 1), we developed two control scenarios. The first was analogous to the race-change scenario, but instead asked participants to consider changing their statehood status. The second asked participants to forgo watching television for the rest of their lives. Would the race-change scenario elicit compensation requests that were more similar to no-television amounts or to statehood-change amounts? State-change requests were predicted to be low because the change was nominal, with no apparent costs. People were not being asked to move to another state, but merely to change their statehood status. No-TV requests, in contrast, were expected to be quite large. Most people highly value television and consider it an important component of their lives. Asking them to forgo this medium for life was expected to be met with sharp resistance, reflected in high monetary requests. Hence, we attempted to place race-change compensation requests on a continuum, from life changes that were largely superficial, to life changes that would be considered by most to be central and encompassing.

We predicted that typical race-change requests, in contrast to the huge sums reported by Hacker (2003), would be relatively low. McIntosh (1988) has speculated that many aspects of White privilege are largely invisible to Whites themselves. In addition, Whites may harbor perceptions that many social disparities (economic and otherwise) are much smaller than they used to be (Eibach and Ehrlinger, 2006), are continuing to get smaller, and/or are issues of class as opposed to race. Denial of widespread discrimination and economic disparities plays a key role in several theories relating to racial-policy attitudes such as modern racism (McConahay 1986), symbolic racism (Henry and Sears, 2002; Kinder and Sears, 1981), and color-blind racism (Bonilla-Silva 2003). The logic of these various models is the same: If discrimination and economic disparities are minimal, programs such as reparations and affirmative action are no longer warranted (Kluegel 1985).

Despite our general predictions of relatively low race-change requests, we did expect substantial variance in the level of requests (corresponding with variance in perceptions of racial disparity). Further, we predicted that this variance would relate positively to support for reparations. These predictions were examined across a series of separate data collections using a wide variety of participants. In Study 1, we examined general patterns of race-change compensation requests using participants who varied in terms of age, gender, geographic location, and student status. In Study 2, we varied the wording of our key race-change scenario to address a potential criticism of the measure used in Study 1. In Study 3, we examined the correlation of the new race-change measure with support for reparations. Study 4, a conceptual replication of Study 3, also included an attempt to alter perceptions of racial disparities in order to ascertain the effects on race-change requests. Study 5 presented forms of the race-change scenario that were designed to address alternative interpretations, as well as to further examine the impact of altering perceptions of racial disparity. Following Study 5, we examined participants' explanations for their race-change compensation requests. Finally, Study 6 attempted to show that participants were willing to claim restitution today for a long-ago injustice to a distant ancestor.

STUDY 1

Results and Discussion

The median race-change request was $1500. This was comparable to the $1000 median amount requested to change one's statehood, but vastly smaller than the $1,000,000 median amount requested to forgo TV. We report medians because extreme outliers (common in open-response contingent valuations) rendered means a poor descriptive measure of central tendency. The graph in Figure 1 conveys the variance in responses.

Distributions of dollars required for race change, state change, and no TV for Study 1: Initial Ohio State sample. N = 61

To analyze the race-change measures and other open-response items (e.g., required compensation for state change and for no TV), we routinely used log-transformed scores that reduced the influence of extreme outliers. Means reported with those analyses were always log transformed. To examine differences between race-change requests and the two comparison scenarios, we conducted separate within-participants analyses using the log-transformed data. The difference between the race-change requests and no-TV requests was statistically impressive [(M_{Race Change} = 7.84, SD = 4.24; M_{No TV} = 13.68, SD = 5.00), F(1,60) = 76.34, p < 0.001, η² = 0.56]. In contrast, although race-change requests were significantly larger than state-change requests, the differences were nowhere near as pronounced as the race/TV difference [(M_{Race Change} = 7.84, SD = 4.24; M_{State Change} = 6.67, SD = 3.53), F(1,60) = 8.55, p = 0.005, η² = 0.13]. In general, then, the race-change requests were much more similar to state-change requests than no-TV requests.

The effect of question order (race-change first vs. no-TV first) did not reach conventional levels of significance [(M_{Race First} = 6.88, SD = 4.64; M_{TV First} = 8.71, SD = 3.70), F(1,59) = 2.92, p = 0.093, η² = 0.05]. Median raw race-change requests corresponding to the log means were as follows: Race-change first = $1000, no-TV first = $7500.⁸

For all reported studies (Studies 1, 1A, 2, 3, 4, 5, and the studies referred to in endnotes 9 and 10), gender effects were examined. No significant effects or interactions were observed aside from a single main effect of gender on log-transformed race-change requests in the study reported in endnote 9, F(1220) = 8.49, p = 0.004, η² = 0.03. In this case, males (M = 6.21; SD = 5.48; corresponding raw median value = $500) made higher race-change requests than did females (M = 4.16; SD = 4.97; corresponding raw median value = $10).

In sum, the typical White participant in our sample requested very little money to change his or her racial status. However, our ability to generalize from this sample was limited by several factors. The vast majority of our participants were between the ages of eighteen and twenty, and all were college students. In addition, the geographic location of our participants was limited to a particular midwestern state. Hence, the next step was to administer our contingent valuation items to samples that varied in terms of age, student status, and geographic location.

STUDY 1A

Results and Discussion

The results across the three generalization samples and the initial sample were remarkably consistent (see the first four data rows of Table 1 for a summary of contingent valuation responses). The race-change values ranged from $100 to $1500; state-change values ranged from $100 to $1000; and the no-TV values ranged from $150,000 to $1,000,000.

Median responses to the contingent valuation items across samples varying in region, age, and student status

As before, for all analyses including open-ended responses, logarithmic transformations were conducted to reduce the influence of outliers on statistical analyses. The three samples of participants did not differ with respect to the race-change item, F(2,257) = 2.29, p = 0.103, η² = 0.02; hence, the data from these samples were collapsed for the subsequent analyses. As in Study 1, there was a large difference between the race-change and no-TV requests [(M_{Race Change} = 6.04, SD = 5.25; M_{No TV} = 11.36, SD = 5.97), F(1,259) = 160.04, p < 0.001, η² = 0.38]. Also as in Study 1, the difference between the race-change and state-change requests was significant, but unimpressive [(M_{Race Change} = 6.04, SD = 5.28; M_{State Change} = 5.25, SD = 4.34), F(1,259) = 7.45, p = 0.007, η² = 0.03].

In addition, the question-order difference in the Harvard sample was not significant [(M_{Race First} = 5.69, SD = 4.44; M_{TV First} = 5.01, SD = 4.36), F(1,123) = 0.74, p = 0.391, η² = 0.01]. Median raw race-change requests were as follows: Race-change first = $150, no-TV first = $100.

Within the sample of older adults, it was possible to examine the relationship between age and race-change requests. If race-change requests tapped perceptions of Black cost, older Whites might have been expected to list higher values, given that they might have witnessed the struggles of Blacks over the decades; in addition, older participants might be more sensitive than undergraduates to the distinction between income and wealth and therefore more likely to reflect wealth disparities (positive correlation expected). On the other hand, older participants might list lower values if they felt that fortunes had improved for Blacks since, say, the 1950s (negative correlation expected). Finally, if race-change requests are related directly to prejudice, older adults, who are more prone to expressing prejudice (von Hippel et al., 2000), may be more inclined to list high values (e.g., “I would never be Black”). In fact, a negative correlation obtained between race-change requests and age [r(77) = −0.31, p < 0.01], indicating that older respondents actually requested smaller amounts to change their racial status. This finding suggested, at the very least, that race-change request increases were not a proxy for increases in age-associated prejudice.

STUDY 2

One alternative account of the low race-change requests is that participants felt that they could continue to pass as White, and so they would not feel the full brunt of racism. To address this issue, we ran an experiment in which the original race-change measure was compared with a measure that was the same, except that participants were told that they would be identified as Black “in any encounters” after accepting the money.

STUDY 3

Studies 1 and 2 demonstrated that median race-change requests were relatively small. However, inspection of the race-change request frequencies in Figure 1 reveals that a fair percentage of participants requested relatively larger sums (in excess of $10,000 dollars). If the race-change measure were really a measure of the perceived costs of being Black, we would expect it to be positively correlated with support for reparations. As discussed above, those who perceive more extreme present-day inequalities (ostensibly due to the lingering effects of slavery) should be more open to remedial programs such as reparations. In addition, we explored whether degree of explicit prejudice was correlated with race-change requests. Low race-change requests might indicate relatively little prejudice (e.g., “I have no problem with being a Black person”), and high requests may indicate higher prejudice (e.g., “I would never be Black!”). Data from Study 1A provided evidence against a prejudice-tapping interpretation of the race-change requests (older individuals requested lower sums); however, no strong conclusions could be drawn from the simple correlation between age and race-change requests. In Study 3, we directly examined the relationship between reparations support and the race-change measure by asking people whether or not they would vote for a reparations bill, and how much reparations should be paid if such a bill were to pass. If the race-change measure tapped explicit prejudice, a negative correlation between the race-change requests and support for reparations would be expected (higher race-change requests equaling more prejudice, leading to lower support for reparations). We therefore included the Modern Racism Scale (McConahay 1986) and a “Black feeling thermometer” in Study 3 to more directly ascertain the relationship between race-change requests and prejudice.

Results and Discussion

The overall median values were as follows: race change = $75; state change = $100; no TV = $500,000. Hence, the pattern of medians across the three items was very similar to the previous samples. As in the previous studies, logarithmic transformations were used to examine the differences between the race-change scenario requests and the other two scenarios. As before, race-change requests diverged widely from no-TV requests [(M_{Race Change} = 4.58, SD = 4.86; M_{No TV} = 10.54, SD = 5.97), F(1,81) = 63.71, p < 0.001, η² = 0.44]. State-change requests, in contrast, did not differ significantly from race-change requests [(M_{Race Change} = 4.58, SD = 4.86; M_{State Change} = 4.70, SD = 3.90), F(1,81) = 0.07, p = 0.786, η² = 0.00].

In response to the support-for-reparations item, 14.6% said yes, and 85% said no, numbers similar to the poll reported above (Dawson and Popoff, 2004; Viles 2002). The median suggested-reparations payment was $5.50; of course, most of the people who said they would not support reparations listed $0 in response to this item.

A logarithmic transformation was performed on the open-ended reparations payment item for the purposes of correlational analyses. Relevant correlations are reported in Table 2 (column 1). Most importantly, positive correlations were found between requested race-change dollars and both support for reparations (r(82) = 0.22, p < 0.05) and suggested reparations payments (r(82) = 0.45, p < 0.001). That is, the greater that Black cost was perceived to be, the higher the amounts of recommended reparations were. Finally, the race-change measure was not correlated with the Black feeling-thermometer rating (r(82) = −0.13, p = 0.24) or the MRS (r(82) = 0.10, p = 0.39). Hence, the race-change measure was predictive of two indicators of support for reparations in the expected direction, and was not significantly associated with two different measures of explicit prejudice. These findings indicate that, to the extent that Black costs are perceived as negligible (i.e., people would assume the “burden” for very little compensation), the perceived need for reparations decreases. The low median race-change requests found in Studies 1 and 2 indicate that low perceptions of Black costs among Whites are widespread; hence, the weak support for reparations found in national polls (Dawson and Popoff, 2004; Viles 2002) may indicate a widespread ignorance of Black costs within the broader society.

Correlations of various measures with race-change requests across Studies 3–5

STUDY 4

In Study 3, higher race-change requests were associated with increased support for reparations and higher suggested reparations payments. However, the majority of race-change requests were very low, indicating that most participants were unmindful or unaware of Black costs, or corresponding White privilege. If that is the case, one way to increase support for reparations may be to increase awareness of White privilege (and correlative non-White costs). We sought to manipulate such awareness in Study 4, hypothesizing that awareness of White privilege would increase support for reparations. The same race-related measures from Study 3 were included in Study 4 to ensure that the correlations in Study 3 were stable, and to examine the effects of awareness of White privilege on those items as well.

We have argued that thought-provoking contingent valuation scenarios may have advantages not shared by simple Likert-type items. In this study, as a foil to our contingent valuation methodology, we simply asked people to report on a Likert rating scale whether they thought it was easier to be Black or White.

STUDY 5

In Study 5, participants received either the original status-change scenario, or one of three alternative scenarios. We desired to interpret the low median race-change requests found across the first four studies as evidence of a general ignorance of racial disparities, but alternative explanations for these low requests required consideration. As discussed in the previous section, there are a number of Black costs that the original status-change scenario potentially allows participants to avoid (growing up as a Black person, having unambiguously dark skin and “Black” features). However, it may be unfeasible to instruct participants to re-imagine their own childhood through a different lens. And, as with the Hacker scenario, fantastic scenarios involving a change of appearance may encourage artificially high requests. To avoid these problems, we chose a minimalist approach to counterfactual thinking. Participants were asked to imagine that they were about to be born as a random White person in America, but that they were being offered a cash gift to switch to being born as a random Black person (similar to John Rawls's 1971 “veil of ignorance” thought experiments). Presumably, race-change requests would be heavily influenced by perceptions of the total costs of the average Black person across the lifespan (versus perceptions of the total benefits of the average White person). Furthermore, participants were not asked to imagine a radical skin-change procedure—skin color would be initially assigned as a part of the supposed birthing process.

To further examine the effects of educating participants about Black costs, in one condition, this birth-change scenario was followed with a list of disparities suffered by African Americans. The third new condition was nearly identical to the new birth-change condition, except that instead of referring to America, the scenario referred to a fictional land called Atria. In addition, instead of referring to Black and White, the scenario referred only to a majority and a minority. This scenario was also followed by the list of costs (to the fictional minority group). The idea was to obtain a baseline for the valuation of the costs, and then to see if this valuation differed from when the groups in question were Blacks and Whites in contemporary America. The original status-change scenario, in which participants were asked to consider a switch in public racial status at the present time in their lives, was the fourth condition. As in prior studies, the status-change scenario did not include information relevant to racial disparities.

In addition to the new scenarios, several additional measures were included in Study 5. To further explore the relationship of the race-change measure to the other measures of perceptions of racial disparity, Swim and Miller's (1999) five-item White Privilege Scale (WPS) was administered, as well as a variant of the single-item measure of Black cost from Study 4. We also included items from the Crocker et al. (1994) racialized Collective Self-Esteem Scale (CSES). One important alternative explanation of the various contingent valuation scenarios is that race-change requests reflect the extent to which an individual has become personally invested in his or her own (White) racial identity—regardless of racial disparity perceptions [similar to the endowment effect; see Thaler (1980)]. Inclusion of a measure of collective self-esteem allowed for a simple check on this possibility.

Results and Discussion

Main Analyses

Contingent Valuation Scenarios.

There were no significant effects or interactions involving participant location, so we collapsed across this variable for the remainder of the analyses. The median requests in the status/no-costs condition ($1000) and the birth/no-costs condition ($5500) were relatively similar. Making participants aware of the costs of being Black (the birth/costs scenario) led to markedly higher median request ($500,000). However, removing the issue of race from the equation (Atria/costs scenario) led to a still higher median request ($1,000,000). As in previous samples, the median no-TV requests were quite large ($1,000,000), despite the deletion of the text relating to being monitored for life, whereas the median state-change requests were quite small ($1000).

Log-transformed requests were entered into an ANOVA (Analysis of Variance) with scenario type as the independent variable. A significant main effect was obtained [(M_Atria/Costs = 12.42, M_Birth/Costs = 10.85, M_{Birth/No Costs} = 7.75, M_{Status/No Costs} = 7.15), F(3,184) = 6.47, p < 0.001, η² = 0.10]. LSD (Least Significant Difference) posthoc tests were conducted to probe for significant differences between conditions. The Atria/costs condition was found to be significantly different from both the birth/no-costs condition (p = 0.001) and the status/no-costs condition (p < 0.001), but was not appreciably different from the birth/costs condition (p = 0.262). The birth/costs condition was significantly different from the birth/no-costs condition (p = 0.026) and the status/no-costs condition (p = 0.007). Finally, and of particular importance, the difference in requests between the status and birth (no-costs) race-change conditions was not significant (p = 0.666).

Responses in each of the four conditions were then compared to the two baseline measures: state change and no TV (using the log-transformed means—see Table 3). The within-participants comparisons with the state-change measure were as follows: Atria/costs, F(1,43) = 30.76, p < 0.001, η² = 0.42; birth/costs, F(1,48) = 8.86, p = 0.005, η² = 0.16; birth/no costs, F(1,45) = 3.38, p = 0.073, η² = 0.07; status/no costs, F(1,48) = 4.47, p = 0.040, η² = 0.09. The within-participants comparisons with the no-TV measure were as follows: Atria/costs, F(1,43) = 0.16, p = 0.693, η² = 0.00; birth/costs, F(1,48) = 1.45, p = 0.234, η² = 0.03; birth/no costs, F(1,45) = 19.68, p < 0.000, η² = 0.30; status/no costs, F(1,48) = 9.26, p = 0.004, η² = 0.16. Generally, responses in the status/no-costs and birth/no-costs conditions were similar to the state-change scenario, whereas responses to birth/costs and the Atria/costs scenarios were more similar to the no-TV scenario.

Median and log-transformed responses to the various scenarios presented in Study 5

In summary, despite the encompassing scope of the birth-change scenarios (which might have been expected to increase the race-change compensation requests), when no mention of Black costs was made, the status-change and birth-change conditions yielded similarly low requests ($1000 vs. $5500, respectively). When costs were described, the median request jumped to $500,000. Perhaps most strikingly, when only considering the costs of being a minority, without referring to race in America (Atria/Costs condition), compensation requests soared. Hence, there seem to be some factor or factors that can mitigate Whites' perceptions of the seriousness of certain racial disparities when thinking in terms of Black versus White (discussed below).

Similar to the benefits induction (Appendix A) in Study 4, scenario type did not significantly influence support for reparations [(M_Atria/Costs = 3.15, SD = 2.12; M_Birth/Costs = 3.03, SD = 2.42; M_{Birth/No Costs} = 2.61, SD = 2.17; M_{Status/No Costs} = 3.81, SD = 2.58), F(3,184) = 2.15, p = 0.096, η² = 0.03].

Correlational Analyses

Although we report (Table 2, columns 3–6) the correlation between race-change requests and other measured variables (e.g., reparations support) within each of the four Study 5 scenarios, our purpose was to clarify the original status-change measure. Comparison with the birth/no-costs results (column 4) was key to the clarification and the focus of our present use of Table 2 results for Study 5.

As in previous studies, we predicted that, to the extent that any of the scenarios tapped into perceptions of racial disparity (Black cost/White privilege), corresponding requests should be positively associated with support for reparations. However, within none of the four scenario-type conditions were race-change requests significantly correlated with support for reparations. The correlation within the status/no-costs condition was in the expected direction (positive), but smaller than in Studies 3 and 4. It is possible that this attenuation was the result of the specificity of the reparations questions used in Study 5. In Studies 3 and 4, participants were asked whether they supported reparations in general. Consistent with work on the P-I Gap (Schuman et al., 1985), the more specific reparations items (specifying either the government or corporations as payees) may have aroused more resistance.

Next, we examined the patterns of correlations within the status/no-costs condition and the comparison birth/no-costs condition. Beginning with the status/no-costs scenario, in contrast to Studies 3 and 4, race-change requests were significantly correlated with both MRS [r(49) = 0.32, p = 0.024], and the Black feeling-thermometer scale [r(49) = −0.34, p = 0.015]. Race-change requests were not significantly correlated with either the single-item measure of Black costs [r(49) = −0.12, p = 0.406] or Swim and Miller's (1999) WPS [r(49) = −0.14, p = 0.349]. Finally, race-change requests in the status/no-costs condition did not relate significantly to private CSE [r(49) = −0.02, p = 0.890], public CSE [r(49) = −0.10, p = 0.499], or CSE identity importance [r(49) = 0.09, p = 0.539].

The pattern of correlations with race-change requests for the birth/no-costs condition (Table 2, column 5) was similar to status/no costs (column 4), except that there was a marginal association with private CSE [r(46) = 0.27, p = 0.065]; the association with Black warmth was nonsignificant; and the relationship with reparations support tended to be negative.

Finally, we examined the relationships between support for reparations and the two measures of perceptions of White privilege/Black costs within both the status/no-costs and birth/no-costs conditions. The results for the status/no-costs condition were as follows: WPS [r(49) = 0.36, p = 0.012]; single-item measure of Black costs [r(49) = 0.30, p = 0.036]. In contrast, these two correlations were nonsignificant in the birth/no-costs condition: WPS [r(46) = 0.07, p = 0.670]; single-item measure of Black costs [r(49) = 0.17, p = 0.274]. This pattern of results may help to explain the negative, though nonsignificant, correlation between race-change requests and reparations support in the birth/no-costs condition (Table 2, row 1), in that participants in the birth/no-costs condition appeared not to be focused on racial disparities when rating their support for reparations. We further examine possible reasons for these patterns of within-condition correlations below, where participants' explanations of race-change requests are analyzed.

The patterns of correlations between race-change requests and other variables were relatively similar, despite the differences noted between the status/no-costs scenario and the birth/no-costs scenario. One of these was that, in the birth/no-costs scenario, race-change requests were marginally correlated with private CSE—a variable not related to perceptions of ongoing Black suffering. This finding suggests that Black suffering may have been discounted under the birth/no-costs scenario, and thus helps to explain the aforementioned negative association between compensation requests and reparations support in the birth/no-costs condition. The White CSE results are consistent with an interpretation that most birth/no-costs participants were not relying on perception of Black costs.

Study 5 helped to clarify our original status-change measure of perceptions of racial disparity. Overcoming several of the possible critiques of the status-change scenario (as accomplished by the birth/no-costs scenario) did not lead to substantial increases in race-change requests. Although we interpret the correlation of status-change requests with support for reparations as due to the shared link with perceptions of racial disparity, race-change requests were not correlated with simpler measures of Black cost and White privilege. Fortunately, the majority of participants in Studies 1 through 5 provided brief explanations for their race-change requests. Their explanations helped to shed light on the correlations of race-change requests with other measures of disparity.

PARTICIPANTS' EXPLANATIONS FOR THEIR RACE-CHANGE REQUESTS

Participants' race-change requests may have reflected a variety of motivations beyond perceptions of racial disparity. It is possible that low race-change requests were indicative of perceptions that the scenario allows the bulk of Black costs to be avoided (the change is made after the childhood experience; skin color does not change). Low requests may have also indicated opposition to being paid for one's racial status (due to pride-based or political concerns). Given that the race-change measure entailed a willingness to be paid, it is possible that race-change requests (particularly high requests) were indicative of greed. By coding participants' explanations for their race-change requests, we were able to address these and other alternative interpretations of the median race-change requests (Table 1) reported in Studies 1 through 5 and endnotes 9 and 10.

In total, 632 participants across the previously reported studies provided brief written explanations for their race-change (status-change only) requests immediately after making them.¹¹

In the valuation-irrelevant category, the following subcategories were exhaustive: (1) cited Black costs that would be avoided by the scenario; (2) cited issues involving pride in one's racial status; (3) cited perceptions of past injustices that require compensation; and (4) cited opposition to receiving government funding related to racial status. All responses that could not be categorized into one of the two basic categories were coded as nondiagnostic. Within this nondiagnostic category, the following explanations were coded either as indicative of difficulties with the scenario (misunderstandings, perceived ambiguities) or as entirely nondiagnostic (left blank, completely unclear or illegible).

Responses were coded by the first author and a research assistant who was blind to expectations about category frequencies. These two coders agreed on the placement into one of the three main categories 81.8% of the time (Kappa = 0.67, p < 0.001). When considering the eleven subcategories separately, agreement was similarly high (Kappa = 0.68, p < 0.001). These initial levels of agreement were deemed acceptable [see Landis and Koch (1977) for a discussion of the Kappa statistic], and disagreements were settled via discussion.

Results and Discussion

Generally, the coding data supported our contentions that the race-change measure is sensitive to perceptions of racial disparity (though sometimes indirectly), and that perceptions of Black costs are responsible for much of the variation in participant requests. Indeed, when considered apart from the actual race-change requests, the explanations themselves provide evidence consistent with our general hypothesis that Whites are largely unaware of racial disparities.

Table 4 contains descriptive information relating to the three main coding categories, as well as the eleven subcategories for all data collections (n = 632) that included the status-change scenario and also elicited participants' explanations for their race-change compensation requests (including the status/no-costs condition in Study 5). Overall, 361 (57.1%) of the responses were coded as valuation relevant, 178 (28.2%) as valuation irrelevant, and 93 (14.7%) as nondiagnostic of the given race-change request. Hence, when focusing only on the first two categories where meaningful responses were given, 67.0% of the responses were valuation relevant.

Frequency of various race-change explanations, and corresponding race-change requests

The relative percentages of the five subcategories within the 361 valuation-relevant responses were as follows: 79 (21.9%) indicated Black costs, 12 (3.3%) indicated White costs, 140 (38.8%) indicated racial parity, 53 (14.7%) explained their requests based on trivial technicalities, and 77 (21.3%) appeared to be motivated only by greed. Hence, approximately 78.1% of the valuation-relevant explanations seemed to indicate perceptions of either racial parity, or at least only trivial racial disparity. A small percentage of participants (n = 12; 1.9%), in fact appeared to perceive net benefits of being Black.

For each of the three main categories and the eleven subcategories, median race-change requests were calculated as in the previous studies (Table 4, columns 3 and 4). The median request for the valuation-relevant group was $1000, a number strikingly similar to the overall values reported in the previous studies. The corresponding median requests in the valuation-irrelevant and nondiagnostic category were $0 and $1000, respectively.

The median requests across the five valuation-relevant subcategories were as follows: perceived Black costs = $300,000; perceived White costs = $75; perceived racial parity = $37.50; scenario-specific Black costs = $1000; greed = $50,000. For the group whose explanations cited Black costs, race-change requests were correspondingly high. In contrast, for those seemingly unaware of Black costs (the four remaining groups), the median request was only $500.

In summary, participants, by and large, were not sensitive to the massive racial disparities that still exist today. For the rather small group of participants who were race sensitive (n = 79 out of 632), race-change requests were relatively high. For the remainder of the participants (excepting the “greed” group), race-change requests tended to be minimal.

The analysis of the race-change explanations may also help to shed light on the lack of correspondence between race-change requests and other measures of racial disparity in Studies 3, 4, and 5. When confronted with a simple question attempting to gauge racial disparity, participants most likely responded in a relatively reflexive fashion, indicating mild discrepancies. The race-change scenario, however, required participants to imagine a counterfactual scene, and gave them great latitude in choosing a set of factors with which to make their response. While we had hoped that more participants' explanations would involve direct assessments of racial disparities, a full one-third of participants gave nondiagnostic responses (notably, issues involving pride).

The race-change scenario was intended to mirror the issue of reparations (participants imagined being paid based on minority status). Although the issue of reparations involves past injustice, the race-change requests indicate that considerations of the past may play a relatively minor role in determining support for reparations. There are many different factors (direct perceptions of racial disparity being only one) that would seem to relate to the complicated issue of reparations. Our contingent valuation procedure appeared to elicit rumination regarding this myriad of factors, and this may simultaneously explain the positive correlation with reparations and the null correlations with simpler measures of racial disparity.

Finally, Table 5 examines the race-change explanations within the four conditions of Study 5. Although the within-cell sample sizes are quite small (ranging from forty-four to forty-nine), given eleven coding subcategories, several interesting findings emerged. Most strikingly, the three new conditions (birth/no costs, birth/costs, and Atria/costs) all dramatically increased the percentage of valuation-relevant explanations, and decreased the percentage of valuation-irrelevant responses. Also instructive is the fact that the percentage of participants acknowledging Black costs increases from column 1 to column 4. This finding may appear inconsistent with the null to negative correlations between race-change requests and reparations support reported in columns 4 through 6 in Table 2, but it is consistent with the main-effect findings of Study 4, in that cajoling participants into recognizing and acknowledging privileges and costs did not have a corresponding effect on support for reparations. There appears to be a difference between people who acknowledge racial disparities on their own (people who can be detected using the original status-change scenario), and those who are informed of racial disparities—a distinction we discuss below under Cognitive Frames and Racial-Policy Attitudes.

Frequencies (relative percentages) of various race-change explanations for the four conditions in Study 5

STUDY 6

Up to this point, we have focused on the relationship between White perceptions of Black costs and support for reparations. However, another pervasive rationale for opposing reparations is that the crime of slavery occurred too long ago. This sentiment has been echoed by antireparations advocates such as conservative pundit David Horowitz (2001) and Congressman Henry Hyde (Dawson and Popoff, 2004, p. 49). Although a full empirical investigation of this form of opposition is outside the purview of this paper, we wanted to provide an example of how the counterfactual scenario methodology that we employed in the preceding studies could be adapted to address this issue as well.

Those who oppose reparations on the grounds that slavery (the crime) occurred in the distant past reason that the original victims (the slaves themselves) cannot receive compensation, and the perpetrators (slave owners) died long ago. Hence, they feel that reparations would unfairly sanction present-day Whites for a crime that they did not commit, and unfairly reward present-day Blacks for hardships that they themselves did not suffer. The logic of this line of reasoning appears, on the surface, relatively straightforward, and so we wondered if people would oppose the payment of reparations for any long-ago sufferings, or whether their opposition might be more specific to slavery-related reparations. To examine this issue, we devised a simple scenario wherein participants were asked to imagine that a distant relative had been kidnapped. The culprits were said to have demanded a large ransom, which bankrupted the family and was used to start a business that still prospers to this day. After reading this scenario, participants were asked if they would agree to be included in a class-action lawsuit aimed at recovering wealth from the descendants of the kidnappers. If people oppose restitution for long-ago incidents, regardless of the particular antagonists and protagonists, then opposition to being included in the lawsuit should be extreme [similar to 90% opposition to reparations reported in the 2001 CNN/Gallup poll or the 96% reported in Dawson and Popoff (2004)].

COGNITIVE FRAMES AND RACIAL-POLICY ATTITUDES

Cognitive frames are mental structures that assimilate complex or bewildering stimuli into a simpler and more familiar understanding (Goffman 1974; Kinder 1998a). Extant work on racial-policy attitudes has highlighted the importance of such cognitive frames (Bonilla-Silva 2003; Henry and Sears, 2002; McConahay 1986).

As blatant discrimination and prejudice continue to decrease among Whites in general (Schuman et al., 1997), it is likely that cognitive frames related to the understanding and interpretation of economic (and other) disparities will become increasing powerful predictors of racial-policy attitudes. As our data demonstrate, most participants seemed to be relatively unaware of the extent of ongoing racial disparity. Those who were, however, as shown in Studies 3 and 4, supported reparations more strongly. More generally, to the extent that racial disparities are perceived as absent, minimal, or diminishing (as opposed to present, extreme, or persisting/widening), support for proactive racial policies such as reparations or affirmative action is likely to be slight.

Of course, acknowledgment of extreme and persisting racial disparities does not guarantee support for proactive racial policy. Indeed, our attempts to educate participants on racial disparity (Studies 4 and 5) fell on seemingly deaf ears. This failure of a simple injection of disparity information is unsurprising in light of the posited denial of racism and racial disparity. As noted above, explanations for racial disparities will also play a pivotal role in support for racial policy. For example, racial disparities may be easily explained as a by-product of a deficient Black culture—consistent with the work on color-blind racism (Bonilla-Silva 2003; Neville et al., 2000), symbolic racism (Henry and Sears, 2002), and modern racism (McConahay 1986), as well as other work relating to Whites' perceptions of Black culture (Bobo 1998, 2000; Bobo et al., 1997). To the extent that these elements of Black culture are biologized or essentialized, there may be little hope for any positive change (leading to reductions in support for social programs aimed at reducing racial disparities).

Regarding explanations for racial disparities and other racial phenomena, there is also the possibility that some will feel that no special explanations are required. For example, those with a social-dominance orientation feel that racial disparities are normal and acceptable phenomena (Sidanius et al., 1992; Sidanius et al., 1996; Sidanius and Pratto, 1999). In a similar vein, disparity-generating segregation may be explained in terms of social preferences (e.g., Blacks prefer to live near other Blacks), and hence does not require solutions.

Even if racial disparities are seen as a troubling result of past discrimination and present-day structural disadvantages, opinions regarding solutions to such problems may still prevent support for proactive programs such as reparations and affirmative action. Policies may be opposed based on perceived inconsistencies with cherished American ideals such as meritocracy (Son Hing et al., 2002) and the Protestant work ethic (Kinder 1998b; Kinder and Sanders, 1996). Generally speaking, proactive programs such as reparations necessarily entail targeting a given group for some advantage (a color-conscious approach to racial policy), and it is this targeting that is seen as especially inconsistent with “the American way.” As a result, a color-blind approach to racial policy has become increasingly popular among Whites (Bonilla-Silva 2003; Neville et al., 2000). The color-blind approach to racial policy, however, should actually exacerbate racial disparity, given persistent and systemic White privileges (Rothenberg 2004).

In summary, cognitive frames relating to perceptions of, explanations for, and solutions to racial disparities are likely to play an important role in support for proactive racial policy programs. As should be clear from the preceding discussion, awareness of the presence, extent, and persisting nature of disparities is a necessary but not sufficient precursor to support for proactive racial policies. In the present work, we established that most of our White participants did not appear to perceive persisting racial disparities—to the extent that they did, support for reparations generally increased. But correlations between race-change requests and reparations support were slight to moderate (between 0.13 and 0.45 across Studies 3, 4, and 5), perhaps due to the influence of cognitive frames related to explanations for and solutions to racial disparities. Similarly, our attempts to educate White participants about Black cost/White privilege may have failed to affect reparations support because we did not attempt to counter these same cognitive frames. Finally, the difference in requests between the Atria/costs and birth/costs scenario in Study 5 may indicate that when typical race labels are used (as opposed to a fictional country with nondescript majority and minority groups), an alternative set of cognitive frames is activated, leading to generally, though not significantly, lower compensation requests. These speculations deserve further empirical attention, and the newly developed race-change methodology should prove useful in this regard.

MULTICULTURALIST APPROACHES

Our work is also relevant to the debate between the color-blind and multiculturalist approaches to race. Park and Judd (2005) recently reviewed the work on the color-blind approach to race relations, which entails making race less salient by either encouraging a focus on individuals (Gaertner et al., 2000), or fostering recategorization of different groups into a larger and more inclusive group such as humanity (Gaertner et al., 1993). They suggested that the color-blind approach was fraught with peril due in large part to the pervasiveness of social categorization. And although the “color-blind” approach as studied in the person-perception literature is not entirely synonymous with the “color-blind” approach to racial policy, the societal implications are similar. Namely, both approaches divert attention from racial categories and, necessarily, from racial disparities and related race-based structural advantages and disadvantages. Work on multiculturalism (Park and Judd, 2005) suggests that a focus on race and racial categories can foster an appreciation of the unique gifts that various groups have to offer society. Similarly, a multicultural approach may allow an understanding of the unique challenges faced by certain groups in society—challenges that may warrant a more progressive racial policy.

FUTURE DIRECTIONS

Alternative methods of measuring perceptions of Black cost can be developed. We have employed contingent valuation measures (willingness-to-be-paid) because of their enhanced potential to evoke thoughtful cost/benefit reasoning. However, additional methods for measuring Black costs (e.g., cost-listing protocols and direct assessments of costs, economic and otherwise) may also prove useful in future studies.

We are also mindful that reparations means different things to different people. Polling has shown that certain types of reparations programs are more desirable than others to Whites [e.g., scholarship funds set up by corporations are seen as more desirable than are cash payments administered by the government (Viles 2002; Kravitz 1995)]. Future studies might include variations on the type of reparations plan (e.g., government sponsored vs. corporate sponsored; cash payments vs. social programs).

Studies 4 and 5 attempted to educate White participants either about Black costs or about White benefits, but these one-shot informational interventions had no effect on support for reparations. Furthermore, in the conditions in Study 5 where focus on racial disparities was the greatest (birth/no costs, birth/costs, and Atria/no costs), correlations between race-change requests and reparations support ranged from null to negative. As discussed above, these correlations may be due to a failure by participants to make situational attributions for racial disparities. In fact, prior research has established the tendency to prefer dispositional accounts (e.g., laziness) for outcomes rather than situational accounts (e.g., ongoing discrimination), due to the inability to adopt others' perspectives (Ross 1977). Techniques that would put people “into the shoes” of others may be promising. Such techniques include powerful narratives that may transport (Green and Brock, 2000) White readers into a vivid world of Black costs (à la Uncle Tom's Cabin), or even virtual reality technology (Blascovich et al., 2002), which can portray typical Black costs from a first-person perspective.

PUBLIC REACTIONS TO REPARATIONS

IMPLICATIONS FOR THE REPARATIONS DEBATE

The present approach does not emphasize past slavery-related damages per se, but rather perceptions of the ongoing costs of being identified, correctly, as Black. Study 6, however, provided evidence that participants did not disapprove of payments related to a long-ago damage; indeed, most participants were willing to be listed as claimants for compensation attributable to damage experienced by a distant ancestor. More importantly, to focus upon the ongoing consequences of past damages is in fact to address an injustice that has been perpetuated by the intractable intergenerational Black/White wealth disparity (Allen 1998; Oliver and Shapiro, 1995; Robinson 2001; Wolff 2001). One implication of our studies is that, if the perceived cost of being Black is minimal, and reparations are commensurate, then a reparations program becomes more financially feasible. However, if the perceived need for reparations is associated with the perceived cost of being Black, and if that cost is perceived to be slight, perhaps the need for reparations can be slighted as well.