You are on page 1of 9

Journal of Experimental Social Psychology 82 (2019) 26–34

Contents lists available at ScienceDirect

Journal of Experimental Social Psychology


journal homepage: www.elsevier.com/locate/jesp

Wealthy Whites and poor Blacks: Implicit associations between racial groups T
and wealth predict explicit opposition toward helping the poor
Jazmin L. Brown-Iannuzzia, , Erin Cooleyb, Stephanie E. McKeea, Charly Hydena

a
University of Kentucky, United States of America
b
Colgate University, United States of America

ARTICLE INFO ABSTRACT

Handling editor: Sarah J Gervais The current research investigates whether automatic associations between race and social class predict delib-
Keywords: erative attitudes toward wealth redistribution policies. In three studies, we found that participants were sig-
Implicit attitudes nificantly more likely to associate African Americans (vs. White Americans) with the concept of poor in an
Prejudice implicit task. In addition, we found that this implicit association between poor and African Americans predicted
Political psychology explicit attitudes toward wealth redistribution (Study 1)—an effect mediated by the belief that redistributive
Social cognition policies benefit African Americans over White Americans (Study 2). Further, these associations held when
Income inequality controlling for implicit affective associations and explicit prejudice. Manipulating people's metacognitive beliefs
about the validity of their implicit associations shifted whether implicit Black-poor associations predicted sup-
port for wealth redistribution (Study 3). Overall, these findings suggest race/class associations are well-learned,
spontaneously activated, and may predict explicit policy attitudes toward wealth redistribution above and be-
yond implicit or explicit affective prejudice.

1. Introduction strongest predictor of opposition toward welfare policies was explicit


negative attitudes toward Black people (see also Gilens, 1996; Iyengar,
Economic inequality is rising to historic highs (Saez & Zucman, 1990). However, research has not investigated whether implicit asso-
2014; Wolff, 2002). Although most Americans would prefer a more ciations between Black people and poverty predict attitudes toward
equitable distribution of wealth (Norton & Ariely, 2011), redistributive redistribution.
policies remain controversial (e.g., Ashok, Kuziemko, & Washington,
2015; McCall & Kenworthy, 2009). This controversy may be, in part, 1.2. Implicit semantic associations between race and class
based on racial bias. That is, the extent to which people automatically
associate race and class may predict deliberative attitudes toward In contrast to explicit attitudes, which are deliberative and con-
wealth redistributive policies. trolled, implicit attitudes are spontaneous reactions that are difficult to
control (Gawronski & Bodenhausen, 2006). When the attitude object is
1.1. Explicit associations between race and class controversial and socially sensitive, implicit attitudes are more likely to
diverge due to social desirability concerns (e.g., Dovidio, Kawakami, &
Historical factors, including slavery and institutionalized racism, Gaertner, 2002; Fazio & Olson, 2003; Nosek, 2005). For example, ex-
created a wealth gap between Black and White Americans that persists plicit attitudes toward Black people have become more positive over
today (e.g., Pew, 2016; Quillian, Pager, Hexel, & Midtboen, 2017). The time (e.g., Bergsieker, Leslie, Constantine, & Fiske, 2012), yet implicit
macro-level racial wealth gap may inform explicit stereotypical asso- attitudes remain relatively more negative, suggesting that people may
ciations between being poor and Black (Brown-Iannuzzi, Dotsch, be controlling their explicit attitudes due to social desirability concerns
Cooley, & Payne, 2017; Cox & Devine, 2015; Freeman, Penner, (e.g., Greenwald, Poehlman, Uhlmann, & Banaji, 2009; Payne, Vuletich,
Saperstein, Scheutz, & Ambady, 2011; Lei & Bodenhausen, 2017). & Lundberg, 2017). In addition, research suggests that implicit attitudes
Further, opposition toward redistributive policies may be informed by can independently predict explicit attitudes and behavior (e.g.,
negative attitudes toward Blacks because people assume poor people are Cameron, Brown-Iannuzzi, & Payne, 2012; Greenwald et al., 2009).
Black. Consistent with this explanation, Gilens (1995) found that the Thus, implicit attitudes may be particularly important when


Corresponding author at: 171 Funkhouser Dr., University of Kentucky, Lexington, KY 40506, United States of America.
E-mail address: jazmin.bi@uky.edu (J.L. Brown-Iannuzzi).

https://doi.org/10.1016/j.jesp.2018.11.006
Received 9 February 2018; Received in revised form 28 November 2018; Accepted 28 November 2018
0022-1031/ © 2018 Elsevier Inc. All rights reserved.
J.L. Brown-Iannuzzi et al. Journal of Experimental Social Psychology 82 (2019) 26–34

investigating attitudes toward Black Americans. with open practices, all studies were pre-registered,1 and materials and
Traditionally, implicit attitudes were assessed on one dimension: data are available at an online repository.2 Studies were run using In-
valence (Cameron et al., 2012; Greenwald et al., 2009). However, im- quisit 5 (Millisecond, 2016).
plicit negativity toward Black people may reflect prejudice or the belief
that Black Americans are mistreated in the U.S. (Andreychik & Gill, 2. Study 1 method
2012; also Lee, Lindquist, & Payne, 2018; Uhlmann, Brescoll, & Paluck,
2006). Because implicit semantic associations are more specific, they 2.1. Power and participants
may, at times, predict explicit attitudes better than implicit valence
associations (Gawronski & Ye, 2014; Imhoff, Schmidt, Bernhardt, We determined we needed a minimum sample of 199 to have ade-
Dierksmeier, & Banse, 2011; Sava et al., 2012; Ye & Gawronski, 2018). quate power (0.80) to detect a small effect (d = 0.20) in a repeated
Based on this research, we reasoned that the specificity of implicit se- measure design (G ∗ Power 3; Faul, Erdfelder, Lang, & Buchner, 2007).
mantic associations (i.e., Black/poor) may be important when pre- We needed a minimum sample of 117 to have adequate power (0.80) to
dicting explicit redistribution policy attitudes. detect a small-to-medium effect (f2 = 0.085) in a multiple regression
model. To account for attrition, we aimed to collect N = 220 and to end
1.3. The belief that redistribution benefits black people data collection when we anticipated having N = 300.3
Participants (N = 284) were recruited from the psychology parti-
Implicit race/class associations may only be relevant to policy cipant pool. Following standard procedures (Payne & Lundberg, 2014;
attitudes to the extent people assume redistributive policies pre- Payne, Cheng, Govorun, & Stewart, 2005), five participants were ex-
ferentially benefit Black Americans. Research suggests that when cluded for only pressing one button during the Semantic Misattribution
people think about redistributive policies (particularly welfare), they Procedure (SMP; Imhoff et al., 2011; Payne et al., 2005). In addition,
tend to imagine these policies benefit Black people (e.g., Brown- because the SMP utilizes Chinese symbols, we investigated whether
Iannuzzi et al., 2017; Gilens, 1995, 1996; Sears & Cirin, 1985; Wetts & participants should be excluded based on being able to read or under-
Willer, 2018). However, to our knowledge research has not directly stand Chinese (Payne & Lundberg, 2014). No participants reported
measured the belief that redistributive policies benefit Black Amer- being able to read Chinese, so no participants were excluded on this
icans. Thus, measuring this belief may provide insight into the psy- basis. Previous research has found that Blacks tend to have different
chological mechanism linking implicit racial/class associations to explanations for being poor (Kluegel & Smith, 1986) and different at-
explicit policy attitudes. titudes toward redistributive policies than do other racial/ethnic groups
In addition, validating or invaliding implicit race/class associa- (Dawson, 1984).4 Thus, because our hypotheses were central to non-
tions may add insight into the relation between implicit associations Black participants, we excluded Black participants. Our final sample
and explicit policy attitudes. In particular, people's thoughts about was N = 243 (46 men, 197 women). Participants self-reported being
their implicit associations (i.e., their metacognitions) may affect the White (219), Hispanic (15), Asian (6), Native American (2), and one
implicit-explicit attitude relation because these metacognitions may participant did not report race/ethnicity. Participants identified being
reinforce or diminish the initial association (e.g., Gawronski & Republican (113), Democrat (72), independent (47), and “other” (10),
Bodenhausen, 2006; Petty, Briñol, & DeMaree, 2007). For example, and one participant did not report political affiliation.
previous research has shown that people may invalidate an auto-
matic attitude because it is based on a recently discredited view 2.2. Procedure
(e.g., Gregg, Seibt, & Banaji, 2006; Petty, Tormala, Briñol, & Jarvis,
2006) or because they believed their implicit affective reaction was Participants completed 80 trials of the SMP (Imhoff et al., 2011;
unintentional (Cooley, Payne, Loersch, & Lei, 2015; Cooley, Payne, Payne et al., 2005). On each trial, participants were presented with a
& Phillips, 2014). There are many ways that metacognitions can prime photo (175 ms), a blank screen (125 ms), a target photo (100 ms),
influence the implicit-explicit relation such as by temporarily re- and then a visual noise mask remained on the screen until a response
ducing the activation of the implicit association (e.g., Mitchell, De was entered.5 The target photos were Chinese ideographs (Payne et al.,
Houwer, & Lovibond, 2009) or temporarily reducing the influence of 2005). Participants were asked to determine whether the Chinese
the implicit attitude on subsequent judgments (e.g., Gawronski &
Bodenhausen, 2006). Regardless of the process by which metacog-
1
nitions influence the implicit-explicit association, the result would Study 1 preregistration: http://aspredicted.org/blind.php?x=g5uq9a.
be the same - implicit semantic race/class associations should no Study 2 preregistration: https://osf.io/938ad/register/564d31db8c5e4a7
longer correlate with explicit policy attitudes when people are led to c9694b2be.Study 3 preregistration: http://aspredicted.org/blind.php?x=
pe62wt.
think of those associations as invalid. 2
To find study materials and data, please visit: https://osf.io/xrk6j/.
3
Deviation from the desired sample size reflects issues related to collecting
1.4. The current research participants from the university participant pool. Specifically, the university
advises researchers to run all posted experimental sessions to avoid student
We extend upon previous research in several ways. First, in Study 1, complaints. We aimed to over-sample because we did not know exactly how
we investigate whether participants have implicit semantic associations many participants we would be able to keep in our sample once data collection
between race (Black and White Americans) and social class (Poor and ended.
Rich). We also investigate whether these implicit semantic associations 4
For exploratory purposes, we analyzed the data using all participants and
predict explicit attitudes toward redistribution, controlling for explicit different combinations of participant exclusions across all studies. By in large,
prejudice. In Study 2, we investigate whether the belief that Black (vs. the results remain consistent with what is reported in the current manuscript.
White) Americans benefit from redistributive policies mediates the re- The only difference is in Study 3 where the results seem to hinge on excluding
African American participants. Based on the research and theories reported in
lation between implicit race/class associations and explicit attitudes
the manuscript, this finding is consistent with our expectation. All of these
toward redistribution. Further, we investigate whether implicit se-
analyses in the Supplemental Materials.
mantic associations predict explicit attitudes, controlling for implicit 5
We sought to present the prime supraliminally because previous research
valence associations. Finally, in Study 3, we test whether metacognitive suggests that a quick, yet clear presentation of the prime may enhance the
thoughts about the validity of one's implicit race/class associations alter misattribution of emotions/attitudes onto the ambiguous target item (Cameron
the implicit-explicit attitude correspondence. et al., 2012). Thus, we decided to use the presentation length described in the
We report all measures, manipulations, and exclusions. Consistent manuscript.

27
J.L. Brown-Iannuzzi et al. Journal of Experimental Social Psychology 82 (2019) 26–34

ideographs meant ‘rich’ (coded 0) or ‘poor’ (coded 1). There were 40 control variables were added, Black/poor automatic associations still
critical trials where the prime photo was either of a Black or White male predicted less support for redistribution, although significance dropped
(α = 0.88),6 and 40 non-critical trials where the prime photo was a to marginal, b = −0.09, SE = 0.05, p = .088.
positive or negative image from the International Affective Picture
System (Lang, Bradley, & Cuthbert, 2008). We created one index that 4. Discussion
reflected the average response preceded by Black primes and one index
that reflected the average response preceded by White primes (higher Overall, these results are consistent with our hypotheses - partici-
numbers indicated a greater proportion of ‘poor’ responses). pants were more likely to implicitly associate poor with Black (vs. White)
Next, participants were asked about their support for redistributive Americans, and the extent to which people implicitly associated race
policies (Brown-Iannuzzi, Lundberg, Kay, & Payne, 2015). A sample with class predicted opposition to redistributive policies. Next, we sought
item is: “In general, the wealthy should be taxed to provide benefits to to extend our findings by investigating whether the belief that Blacks are
the poor,” (1 = strongly disagree; 6 = strongly agree; all items are re- more likely than Whites to benefit from redistribution policies mediated
ported in the Supplementary materials). We averaged these items to the implicit-explicit relation found in Study 1. Further, we investigated
create one index which assessed support for redistribution (higher whether implicit semantic associations uniquely predicted explicit policy
numbers indicated greater support; α = 0.89, M = 3.64, SD = 1.03). attitudes, controlling for implicit valence associations using the Affect
We measured explicit attitudes toward Blacks using the Symbolic Misattribution Procedure (AMP; Payne et al., 2005).
Racism Scale (higher numbers indicated more symbolic racism; Henry &
Sears, 2002; α = 0.65, M = 0.04, SD = 0.52), party affiliation, and poli-
5. Study 2 method
tical ideology (1 = very liberal; 6 = very conservative; M = 3.05, SD = 1.02)
as previous research found these measures predict attitudes toward redis-
5.1. Power and participants
tribution (e.g., Gilens, 1995). We assessed demographic information in-
cluding whether participants could understand Chinese. For exploratory
We determined that we needed a minimum sample of 199 to have
purposes, we included feeling thermometers assessing how warm or cold
adequate power (0.80) to detect a small effect (d = 0.20) in a repeated
participants felt toward Black, White, Asian, and Hispanic people (0 = very
measure design (Faul et al., 2007). In addition, we would need at least
cold/unfavorable; 100 = very warm/favorable), and internal and external
185 to have adequate power (0.80) to detect a small-to-medium effect
motivations to respond without prejudice (Plant & Devine, 1998).
(f2 = 0.085) in a multiple regression model. To account for attrition, we
aimed to collect N = 400.8
Participants (N = 431) were recruited from MTurk. Again, we ex-
3. Results
cluded Black individuals. We also excluded 53 participants who could
read Chinese, 27 who pressed one button throughout the AMP, and 17
We hypothesized that participants would be more likely to associate
participants who pressed one button throughout the SMP. Our final
Black (vs. White) primes with the concept of poor and conducted a
sample was N = 306 (144 men, 155 women, 1 ‘other’, 6 participants
repeated measure ANOVA with prime type (Black vs. White) as the
did not report gender). Participants were White (247), Asian (22),
within-subject factor. Consistent with our hypothesis, participants were
Hispanic (20), multiracial (6), ‘other’ (3), Native American (2), and 6
significantly more likely to think the Chinese ideographs meant ‘poor’
participants did not report race/ethnicity. Participants identified being
when preceded by a Black prime (M = 0.63, SD = 0.27, 95% CI: [0.60,
Democrat (128), independent (84), Republican (76), or “other” (12),
0.67]) than when preceded by a White prime (M = 0.49, SD = 0.24,
and 6 participants did not report political affiliation.
95% CI: [0.46, 0.52]), F(1, 242) = 59.45, p < .001, ηp2 = 0.20.
Next, we inspected the bivariate correlations (see Table 1). Results
revealed that greater Black/poor automatic associations were asso- 5.2. Procedure
ciated with less support for redistribution, r(243) = −0.18, p = .006,
95% CI: [−0.30, −0.04]. There was no relation between White/poor First, participants completed the implicit measures. To avoid order
automatic associations and support for redistribution, r(243) = −0.06, effects, participants were randomly assigned to complete the AMP or
p = .39, 95% CI: [−0.18, 0.06]. SMP first. The SMP procedure and scoring were identical to the design in
We regressed support for redistribution onto the proportion of ‘poor’ Study 1 (α = 0.77). The AMP was similar to the SMP, except responses
responses toward each prime type to investigate whether prime type were ‘pleasant’ (coded 0) and ‘unpleasant’ (coded 1; α = 0.82). All of the
uniquely predicted support for redistribution (see Table 2). All variables prime photos, target photos, and timing parameters were the same in the
were standardized before analysis. Consistent with the bivariate cor- AMP as in the SMP. Similar to the SMP, we created one index that re-
relation results, Black/poor automatic associations uniquely predicted flected the average response preceded by Black primes and one index
less support for redistribution (see Fig. 1), b = −0.19, SE = 0.07, that reflected the average response preceded by Black primes (higher
p = .009. There was no relation between White/poor automatic asso- numbers indicated a greater proportion of ‘unpleasant’ responses).
ciations and support for redistribution, b = 0.002, SE = 0.07, p = .98. Next, participants completed a three-item measure that assessed
For exploratory purposes, we controlled for party affiliation, poli- their belief that Blacks are more likely to receive the benefits of wealth
tical ideology and symbolic racism, variables that are known to predict redistribution than are Whites. These items were: (1) “African
support for redistribution (e.g., Gilens, 1995). Political ideology and Americans are more likely to receive welfare assistance than are
symbolic racism were standardized, and party affiliation was dummy Whites,” (2) “African Americans are more likely to benefit from wealth
coded (1 = Republican) prior to entering them into model.7 When these redistribution than are Whites,” and (3) “Welfare programs are de-
signed to benefit African Americans” (1 = strongly disagree; 6 = strongly
agree). We averaged these items to create one index (higher numbers
6
Primes were taken from B.K. Payne's image database and were previously indicated a greater belief that Blacks benefited from redistribution;
pretested to be matched on perceived attractiveness, age, likeability, and pro- α = 0.82, M = 3.44, SD = 1.24).
totypicality of the individual's respective race/ethnicity (Payne, Brown-
Iannuzzi, & Panter, in preparation).
7 8
Across all studies, we investigated multicollinearity by considering the size We anticipated a higher attrition rate on MTurk than in our university
of the variance inflation factor (VIF) for each predictor. All VIFs remained participant pool, so we over-recruited by a sizeable number. Deviations from
below traditional cutoffs (Kutner, Nachtshem, & Neter, 2004; Sheather, 2009), the desired sample size goal were entirely due to Mechanical Turk Software and
suggesting the magnitude of multicollinearity was not high across all studies. were outside of our control.

28
J.L. Brown-Iannuzzi et al. Journal of Experimental Social Psychology 82 (2019) 26–34

Table 1
Study 1 correlations and 95% Confidence Intervals between measures of interest.
1 2 3 4 5

1 SMP Black primes –

2 SMP White primes 0.33⁎⁎ –


[0.20, 0.45]
3 Support for redistribution −0.18⁎⁎ −0.06 –
[−0.30, −0.04] [−0.18, 0.07]
4 Symbolic racism 0.14⁎ −0.10 −0.57⁎⁎ –
[0.01, 0.27] [−0.21, 0.01] [−0.66, −0.47]
5 Party affiliation (1 = Republican) 0.11 0.13⁎ −0.54⁎⁎ 0.49⁎⁎ –
[−0.02, 0.24] [0.004, 0.26] [−0.62, −0.45] [0.40, 0.57]
6 Political ideology 0.08 0.13⁎ −0.55⁎⁎ 0.53⁎⁎ 0.60⁎⁎
[−0.06, 0.21] [0.01, 0.25] [−0.64, −0.45] [0.42, 0.62] [0.53, 0.68]

Note: Support for redistribution is coded such that higher numbers represent more support. Symbolic racism is coded such that higher numbers represent more
symbolic racism. For simplicity and parsimony, we dummy coded Party Affiliation such that 1 = Republican and 0 = democrats, independents, and people identified
as another political affiliation or did not report. Political ideology is coded such that higher numbers represent more conservative (as opposed to liberal) ideology.
95% confidence intervals for the correlation coefficients appear below the coefficient in brackets. ⁎⁎ Denotes p < .01. ⁎ Denotes p < .05.

Table 2
Study 1 regression coefficients and associated standard errors, 95% confidence intervals, and significance levels when predicting support for redistribution.
b SE 95% CI p-Value b SE 95% CI p-value

SMP White primes 0.002 0.07 [−0.14, 0.14] .98 0.003 0.06 [−0.11, 0.12] .97
SMP African American primes −0.19 0.07 [−0.32, −0.05] .009 −0.09 0.05 [−0.20, 0.01] .088
Political ideology −0.23 0.07 [−0.36, −0.10] .001
Party affiliation (1 = Republican) −0.48 0.13 [−0.73, −0.23] < .001
Symbolic racism −0.63 0.12 [−0.86, −0.39] < .001

Note: Support for redistribution is coded such that higher numbers represent more support. Symbolic racism is coded such that higher numbers represent more
symbolic racism. Political ideology is coded such that higher numbers represent more conservative (as opposed to liberal) ideology. Party affiliation was dummy
coded such that 1 = Republican. All continuous variables were standardized prior to entering the model.

more opposition; α = 0.93, M = 3.08, SD = 1.32). Then, participants


completed the same nine items assessing support for redistributive
polices from Study 1 (α = 0.92, M = 4.28, SD = 1.20).
Finally, participants completed the Symbolic Racism Scale (Henry &
Sears, 2002; α = 0.88, M = 0.00, SD = 0.75), party affiliation and po-
litical ideology (M = 2.62, SD = 1.21). We also measured demographic
information including whether participants could understand Chinese.

6. Results

Consistent with previous research, participants were significantly


more likely to think the Chinese ideographs were unpleasant when
preceded by Black primes (M = 0.46, SD = 0.21, 95% CI: [0.44, 0.49])
than White primes (M = 0.42, SD = 0.20, 95% CI: [0.40, 0.45]), F(1,
305) = 8.49, p = .004, ηp2 = 0.03. And, replicating Study 1, partici-
pants were significantly more likely to think the Chinese ideographs
meant ‘poor’ when preceded by Black primes (M = 0.53, SD = 0.21,
95% CI: [0.50, 0.55]) than White primes (M = 0.48, SD = 0.19, 95%
CI: [0.46, 0.50]), F(1, 305) = 10.38, p = .001, ηp2 = 0.03.
Fig. 1. The partial plot predicting support for redistribution from the propor-
Table 3 presents bivariate correlations between the implicit and
tion of ‘poor’ responses when preceded by an African American prime, con-
trolling for the proportion of ‘poor’ responses when preceded by a White prime. explicit measures of interest. We found greater Black/poor automatic
All variables were standardized prior to entering the model. Line represents the associations were associated with a greater belief that redistribution
best fit line. policies benefit Blacks over Whites, r(302) = 0.22, p < .001, 95% CI:
[0.11, 0.33]. There was no relation between White/poor automatic
associations and the belief that redistribution policies benefit Blacks
Participants also completed items assessing their redistribution over Whites, r(302) = 0.11, p = .06, 95% CI: [−0.01, 0.22].9
policy preferences. First, participants completed six items assessing
their attitudes toward welfare policies and the poor. A sample item was: 9
Unexpectedly, the results revealed that more White/poor associations were
“Most people on welfare could get by without it if they really tried,” associated more negative attitudes toward welfare, r(304) = 0.16, p = .004,
(1 = strongly disagree; 6 = strongly agree; Lei & Bodenhausen, 2017; see 95% CI: [0.05, 0.27], and less support for redistribution, r(303) = −0.15,
Supplemental materials). We averaged these items to create one index p = .01, 95% CI: [−0.26, −0.04]. Given these findings were unexpected and
which assessed opposition toward welfare (higher numbers indicated did not replicate across studies, we hesitate to interpret them.

29
J.L. Brown-Iannuzzi et al. Journal of Experimental Social Psychology 82 (2019) 26–34

Table 3
Study 2 correlation between measures of interest with confidence intervals.
1 2 3 4 5 6 7 8 9

1 AMP White primes

2 AMP Black primes 0.42⁎⁎


[0.32, 0.50]
3 SMP White primes 0.27⁎⁎ 0.15⁎⁎
[0.16, 0.37] [0.04, 0.26]
4 SMP Black primes 0.04 0.35⁎⁎ 0.28⁎⁎
[−0.08, 0.15] [0.25, 0.45] [0.17, 0.38]
5 Black benefit 0.10 0.24⁎⁎ 0.11 0.22⁎⁎
[−0.01, 0.21] [0.13, 0.34] [−0.01, 0.22] [0.11, 0.33]
6 Negative welfare attitudes 0.09 0.14⁎ 0.16⁎⁎ 0.10 0.50⁎⁎
[−0.02, 0.20] [0.03, 0.35] [0.05, 0.27] [−0.01, [0.41, 0.58]
0.21]
7 Support for redistribution −0.13⁎ −0.15⁎⁎ −0.15⁎⁎ −0.08 −0.41⁎⁎ −0.64⁎⁎
[−0.24, [−0.26, [−0.26, [−0.19, [−0.50, [−0.70,
−0.02] −0.04] −0.04] 0.03] −0.31] −0.57]
8 Symbolic racism 0.01 0.13⁎ 0.10 0.12⁎ 0.48⁎⁎ 0.66⁎⁎ −0.63⁎⁎
[−0.10, 0.12] [0.02, 0.24] [−0.01, 0.21] [0.002, 0.23] [0.39, 0.56] [0.59, 0.72] [−0.69,
−0.55]
9 Political ideology 0.00 0.07 0.11 0.08 0.40⁎⁎ 0.56⁎⁎ −0.56⁎⁎ 0.63⁎⁎
[−0.11, 0.11] [−0.04, 0.18] [−0.01, 0.22] [−0.03, [0.30, 0.49] [0.47, 0.63] [−0.63, [0.56,
0.19] −0.48] 0.69]
10 Party affiliation 0.05 0.19⁎⁎ 0.09 0.09 0.27⁎⁎ 0.40⁎⁎ −0.39⁎⁎ 0.40⁎⁎ 0.63⁎⁎
(1 = Republican) [−0.07, 0.16] [0.08, 0.29] [−0.03, 0.20] [−0.02, [0.17, 0.38] [0.30, 0.49] [−0.48, [0.30, [0.55,
0.20] −0.29] 0.49] 0.69]

Note: Black Benefit indicates the belief that Blacks (vs. Whites) benefit from redistributive policies (higher numbers represent a greater belief). Support for redis-
tribution is coded such that higher numbers represent more support. Opposition toward welfare is coded such that higher numbers represent more opposition.
Symbolic racism is coded such that higher numbers represent more symbolic racism. Political ideology is coded such that higher numbers represent more con-
servative (as opposed to liberal) ideology. ⁎⁎ Denotes p < .01. ⁎ Denotes p < .05.

Table 4
Study 2 regression coefficients and associated standard errors, 95% confidence intervals, and significance levels when dependent variables of interest.
b SE 95% CI p-Value b SE 95% CI p-Value

Predicting the belief redistribution benefits African Americans (over White Americans)
AMP White primes 0.02 0.07 [−0.12, 0.16] 0.78 0.05 0.06 [−0.07, 0.17] 0.42
AMP Black primes 0.19 0.07 [0.05, 0.33] 0.008 0.13 0.06 [0, 0.25] 0.05
SMP White primes 0.04 0.07 [−0.09, 0.17] 0.53 −0.01 0.06 [−0.13, 0.11] 0.87
SMP Black primes 0.16 0.06 [0.03, 0.28] 0.016 0.13 0.06 [0.02, 0.25] 0.019
Symbolic racism 0.43 0.08 [0.28, 0.59] < 0.001
Political ideology 0.15 0.07 [0.01, 0.29] 0.03
Party aff. (1 = Rep.) −0.01 0.14 [−0.28, 0.26] 0.94

Predicting opposition toward welfare


AMP White primes 0.01 0.08 [−0.14, 0.17] 0.87 0.07 0.06 [−0.04, 0.19] 0.20
AMP Black primes 0.12 0.08 [−0.03, 0.27] 0.12 0.02 0.06 [−0.10, 0.13] 0.75
SMP White primes 0.16 0.07 [0.02, 0.31] 0.02 0.09 0.05 [−0.02, 0.19] 0.12
SMP Black primes 0.03 0.07 [−0.11, 0.17] 0.68 0.002 0.05 [−0.10, 0.10] 0.97
Symbolic racism 0.68 0.07 [0.54, 0.83] < 0.001
Political ideology 0.18 0.07 [0.05, 0.31] 0.006
Party aff. (1 = Rep.) 0.20 0.13 [−0.05, 0.44] 0.12

Predicting support for redistribution


AMP White primes −0.08 0.08 [−0.24, 0.08] 0.35 −0.15 0.07 [−0.29, −0.01] 0.03
AMP Black primes −0.13 0.08 [−0.29, 0.04] 0.13 −0.03 0.07 [−0.17, 0.11] 0.68
SMP White primes −0.14 0.08 [−0.30, 0.01] 0.06 −0.07 0.07 [−0.20, 0.06] 0.29
SMP Black primes −0.01 0.08 [−0.16, 0.14] 0.88 0.03 0.06 [−0.10, 0.15] 0.69
Symbolic racism −0.72 0.09 [−0.89, −0.54] < 0.001
Political ideology −0.30 0.08 [−0.46, −0.14] < 0.001
Party aff. (1 = Rep.) −0.14 0.16 [−0.44, 0.17] 0.38

Note: Opposition toward welfare is coded such that higher numbers represent more opposition. Symbolic racism is coded such that higher numbers represent more
symbolic racism. Political ideology is coded such that higher numbers represent more conservative (as opposed to liberal) ideology. Party affiliation was dummy
coded such that 1 = Republican. Party Affiliation is abbreviated Party aff., and Republican is abbreviated Rep. All continuous variables were standardized prior to
entering the model.

In addition, we investigated whether the implicit measures pre- each implicit measure, we entered the AMP and SMP responses toward
dicted: (1) the belief that redistribution policies benefit Blacks over each prime type into the regression model simultaneously. All variables
Whites, (2) attitudes toward welfare policies and the poor, and (3) were standardized prior to entering the model. The proportion of
support for redistribution generally. To measure the unique effect of Black/poor responses uniquely predicted the assumption that

30
J.L. Brown-Iannuzzi et al. Journal of Experimental Social Psychology 82 (2019) 26–34

redistribution policies benefit Blacks over Whites, b = 0.16, SE = 0.06, between implicit associations and explicit policy relations by in-
p = .016. Inconsistent with Study 1, however, Black/poor associations vestigating invalidity metacognitions. If people reject the idea that race
did not uniquely predict attitudes toward welfare policies, b = 0.03, connotes class, then they may no longer believe that redistribution
SE = 0.07, p = .68, or support for redistribution, b = −0.01, primarily benefits a specific race. As a result, we would expect the
SE = 0.08, p = .88. Finally, to investigate the robustness of these re- implicit-explicit correspondence to be mitigated among participants
lations, we controlled for political ideology, party affiliation, and who think of their implicit associations as invalid.
symbolic racism, variables that are known to predict support for re-
distribution (e.g., Gilens, 1995). Controlling for these variables did not 8. Study 3 method
substantively change the previously presented relations (see Table 4).
Finally, we investigated whether the belief that redistribution po- 8.1. Power and participants
licies benefit Blacks over Whites would mediate the association be-
tween implicit poor/Black associations and explicit policy attitudes. Using G⁎Power (Faul et al., 2007), we determined we needed a
Although there was no direct effect of semantic associations on explicit minimum sample of 199 to have adequate power (0.80) to detect a
policy attitudes, mediation can occur in the absence of a direct effect for small effect (d = 0.20) in a repeated measure design. In addition, we
a variety of reasons (e.g., Hayes, 2009; Shrout & Bolger, 2002). For needed as sample of at least 185 to have adequate power (0.80) to
example, mediation can occur in the absence of a direct effect when the detect a small-to-medium effect (f2 = 0.085) in a multiple regression
indirect effect (ab pathway) has a different sign from the direct effect model. To account for attrition, we aimed to collect at least N = 300.
(c′). To test this predicted pattern, we used the PROCESS macro with Participants (N = 332) were undergraduates recruited from the
10,000 bootstraps (model 4; Hayes, 2017). We controlled for the pro- psychology participant pool. Eight participants pressed one button
portion of ‘poor’ responses when preceded by White primes, and the during the SMP and five participants reported they could read Chinese
proportion of ‘unpleasant’ responses when preceded by Black and White and were excluded. Again, we excluded Black participants. Our final
primes. In a separate set of analyses, we again controlled for political sample was N = 272 (77 men, 194 women, 1 did not report gender).
ideology, party affiliation, and symbolic racism (see Supplemental Participants self-reported being White (249), Hispanic (12), Asian (10),
materials). The pattern of the mediation results was consistent with our and one participant did not report race/ethnicity. Participants identi-
hypothesis - the belief that redistribution policies benefit Blacks over fied being Republican (128), Democrat (78), independent (43), or
Whites mediated the association between implicit poor/Black associa- “other” (22), and one participant did not report political affiliation.
tions and explicit policy attitudes, indirect effect bWelfare attitudes = 0.08,
95% CI: [0.01, 0.16]; indirect effect bredistribution attitudes = −0.07, 95% CI:
[−0.14, −0.01] (see Fig. 2). 8.2. Procedure

First, participants completed the SMP (α = 0.89). Then, participants


7. Discussion were randomly assigned to the validation (coded 1) or invalidation
condition (coded 0). We created this manipulation based on previous
Overall, these results are consistent with our hypotheses. research (Gawronski & Bodenhausen, 2006), but we made the manip-
Participants were more likely to implicitly associate poor with Blacks ulation specific to the current metacognitive inferences. Participants
(vs. White). And, the extent to which people implicitly associated poor were told: “while completing the picture task, you may have had a ‘gut
with Blacks predicted the belief that redistributive policies benefit association’ between the pictures of Black and White individuals and
Blacks which in turn predicted negative attitudes toward these policies. the concepts of rich and poor. Research has found that this gut asso-
Further, these findings controlled for implicit valence associations, ciation usually does [NOT] accurately reflect people's genuine attitudes.
suggesting that implicit race/class semantic associations may be parti- For the purpose of this study, we'd like you to write 2–3 reasons why
cularly important when investigating these policy attitudes. the associations you made between Black and White individuals and the
In our final study, we sought to further understand the relation concepts of rich and poor during the picture task are [NOT] accurate

Fig. 2. Mediations depicting the association between negative attitudes toward welfare policies (top panel) and support for redistribution (bottom panel) and implicit
semantic associations between poor and African American primes, as mediated by the belief that African Americans (vs. White Americans) benefit from redistribution
policies. These results control for the proportion of ‘poor’ responses when preceded by a White American prime, and the proportion of ‘unpleasant’ responses when
preceded by an African American or White American prime. All continuous variables were standardized prior to entering the model.

31
J.L. Brown-Iannuzzi et al. Journal of Experimental Social Psychology 82 (2019) 26–34

reflections of your attitudes.” Preacher, Curran, & Bauer, 2006). The simple slope relating the pro-
After the metacognition task, participants were asked about their portion of ‘poor’ responses toward Black primes and opposition toward
support for welfare policies (α = 0.89, M = 3.28, SD = 1.09) and re- welfare policies was significant in the valid condition, b = −0.21,
distributive policies (α = 0.89, M = 3.77, SD = 0.99) as in Study 2. We SE = 0.08, p = .006. This means that the extent to which participants
also measured Symbolic Racism (α = 0.63, M = 0.06, SD = 0.52; implicitly associated Black/poor predicted less support for redistribu-
Henry & Sears, 2002), party affiliation, and political ideology tion. Critically, this association dropped to non-significance in the in-
(M = 3.10, SD = 1.04). We measured demographic information in- valid condition, b = −0.06, SE = 0.06, p = .37. And the same basic
cluding whether participants could read Chinese. For exploratory pur- pattern emerged when decomposing the interaction predicting support
poses, we also included feeling thermometers toward Black, White, for redistribution policies. The extent to which participants implicitly
Asian and Hispanic people (0 = very cold/unfavorable; 100 = very associated Black/poor predicted less support for redistributive policies
warm/favorable), and internal and external motivations to respond in the valid condition, b = −0.26, SE = 0.08, p = .001. However, this
without prejudice (Plant & Devine, 1998). association dropped to non-significance in the invalid condition,
b = −0.02, SE = 0.06, p = .73. Finally, we examined whether in-
validity metacognitions dampened implicit-explicit correspondence by
9. Results
leading people to question whether race connotes social class. To test
this, a research assistant, blind to the hypothesis, coded whether or not
First, we investigated whether participants would be more likely to
participants' open-ended responses explicitly rejected the idea that race
associate poor with Black (vs. White) primes. Replicating the previous
connotes class (1 = rejected). As intended, participants in the invalid
studies, participants were significantly more likely to think the Chinese
condition were significantly more likely to explicitly reject the race/
ideographs meant ‘poor’ when preceded by Black primes (M = 0.61,
class relation than participants in the valid condition, χ2(1) = 18.79,
SD = 0.26, 95% CI: [0.57, 0.64]) than when preceded by White primes
p < .001. It is important to note that the manipulation did not lead to a
(M = 0.50, SD = 0.24, 95% CI: [0.47, 0.53]), F(1, 271) = 36.78,
main effect policy attitudes, tWelfare(270) = −0.31, p = .76;
p < .001, ηp2 = 0.12.
tRedistribution(270) = −0.73, p = .46. However, we did not expect this
Next, we investigated whether the metacognition condition mod-
main effect as the manipulation sought to moderate the implicit-explicit
erated the implicit-explicit association. We hypothesized that partici-
association.
pants in the invalidate (vs. validate) condition would have a reduced
association between implicit semantic associations and explicit policy
attitudes. We investigated these hypotheses by running two regressions
10. Discussion
where the dependent variable was regressed onto the proportion of
‘poor’ responses toward each prime type (Black and White), condition,
In Study 3, participants in the invalidity condition were more likely
and the interaction between condition and each prime type. All con-
to reject the idea that race connotes social class than participants in the
tinuous variables were standardized prior to entering the model.
validity condition. In addition, after controlling for explicit prejudice
Condition was dummy coded (valid condition = 1). Counter to our
and political ideology, variables known to be associated with redis-
hypothesis, there was not a significant interaction between condition
tributive policy preferences, there was no relation between implicit
and responses preceded by Black primes (see Table 5). However, when
race/class associations and explicit policy preferences when partici-
political ideology, party affiliation, and symbolic racism, variables that
pants were in the invalidity (vs. validity) condition. Together, this
are known to predict redistributive policy attitudes (e.g., Carmines &
suggests the metacognitions moderate the implicit-explicit correspon-
Stimson, 1989; Gilens, 1995), were added to the model the Condition
dence.
by Black primes interaction emerged (see Table 5).
To further probe these interactions, separate regression lines were
estimated for participants in the valid and invalid condition (see Fig. 3;

Table 5
Study 3 regression coefficients and associated standard errors, 95% confidence intervals, and significance levels when dependent variables of interest.
b SE 95% CI p-Value b SE 95% CI p-Value

Predicting opposition toward welfare


Condition (1 = valid) 0.03 0.12 [−0.20, 0.27] 0.79 0.06 0.09 [−0.13, 0.24] 0.54
SMP White primes −0.01 0.09 [−0.18, 0.17] 0.96 0.04 0.07 [−0.09, 0.18] 0.52
SMP Black primes 0.13 0.09 [−0.04, 0.31] 0.14 −0.06 0.07 [−0.19, 0.08] 0.40
Condition × White primes −0.19 0.13 [−0.45, 0.07] 0.16 −0.15 0.10 [−0.35, 0.05] 0.14
Condition × Black primes 0.14 0.13 [−0.12, 0.40] 0.29 0.28 0.10 [0.08, 0.48] 0.007
Symbolic racism 0.95 0.11 [0.74, 1.16] < 0.001
Political ideology 0.21 0.07 [0.08, 0.34] 0.002
Party aff. (1 = Rep.) 0.10 0.13 [−0.16, 0.36] 0.45

Predicting support for redistribution


Condition (1 = Valid) 0.09 0.12 [−0.14, 0.32] 0.45 0.003 0.09 [−0.17, 0.18] 0.98
SMP White primes −0.002 0.09 [−0.18, 0.17] 0.98 −0.06 0.07 [−0.19, 0.07] 0.35
SMP Black primes −0.21 0.09 [−0.38, −0.04] 0.016 −0.02 0.07 [−0.15, 0.11] 0.74
Condition × White primes 0.12 0.13 [−0.14, 0.37] 0.36 0.11 0.10 [−0.08, 0.30] 0.27
Condition × Black primes −0.12 0.13 [−0.37, 0.14] 0.37 −0.23 0.10 [−0.43, −0.04] 0.018
Symbolic racism −0.58 0.10 [−0.79, −0.38] < 0.001
Political ideology −0.29 0.07 [−0.41, −0.16] < 0.001
Party aff. (1 = Rep.) −0.41 0.13 [−0.66, −0.16] 0.002

Note: Condition was dummy coded such that invalid condition was coded 0 and valid condition was coded 1. Support for redistribution is coded such that higher
numbers represent more support. Opposition toward welfare is coded such that higher numbers represent more opposition. Symbolic racism is coded such that higher
numbers represent more symbolic racism. Political ideology is coded such that higher numbers represent more conservative (as opposed to liberal) ideology. Party
affiliation was dummy coded such that 1 = Republican. Party Affiliation is abbreviated Party aff., and Republican is abbreviated Rep. All continuous variables were
standardized prior to entering the model.

32
J.L. Brown-Iannuzzi et al. Journal of Experimental Social Psychology 82 (2019) 26–34

Fig. 3. The graph on the left displays the interaction between implicit semantic associations when preceded by African American primes and condition when
predicting opposition toward welfare policies. Higher numbers on the y-axis indicates more opposition toward welfare policies. The graph on the right displays the
interaction between implicit semantic associations when preceded by African American primes and condition when predicting support for redistributive policies.
Higher numbers on the y-axis indicates more support for redistributive policies. Condition was dummy coded such that 0 = invalid and 1 = valid. All continuous
variables were standardized prior to entering the model.

11. General discussion 11.1. Limitations and future directions

Across three studies we found that, on average, participants were The current research is not without limitations. For example, there
more likely to associate poor with Blacks (vs. Whites). Additionally, we may be situations where implicit race/class attitudes predict positivity
found that the extent to which participants associated Black with poor toward wealth redistribution policies. In particular, Black participants,
predicted opposition toward welfare and redistributive policies, an ef- or those with positive attitudes toward Blacks, may interpret implicit
fect that was mediated by the belief that redistributive policies bene- poor/Black associations as reflective of structural discrimination. In this
fited Blacks over Whites. Finally, when participants were randomly case, Black/poor associations might result in positive (rather than ne-
assigned to invalidate their implicit race/class associations, there was gative) attitudes toward redistribution. Future research should in-
no longer an association between implicit race/class associations and vestigate this possibility.
explicit policy preferences.
Across the three studies, the results revealed an inconsistent asso- 12. Conclusion
ciation between implicit Black-poor associations and explicit policy
support. To investigate the average correlation across studies, we con- The current research represents an important first step toward un-
ducted mini-meta analyses using a few different approaches (see derstanding implicit race/class associations. We found that people tend
Supplemental materials; Goh, Hall, & Rosenthal, 2016). Overall, we to implicitly associate Blacks with the concept of being poor, as opposed
found a small yet significant correlation suggesting that more Black- to rich. In addition, these implicit semantic associations predicted op-
poor implicit associations were associated with more negative attitudes position toward redistributive policies. Overall, these findings suggest
toward welfare and redistribution, average r's ≥ 0.15, p's ≤ 0.05. These that automatic beliefs about wealthy Whites and poor Blacks may affect
findings suggest that there is an overall correlation between the Black- wealth redistribution policy attitudes.
poor association and negative attitudes toward redistribution or welfare
policies. Appendix A. Supplementary data
These findings can be understood in the context of implicit pre-
judice and implicit stereotypes. Previous research suggests that al- Supplementary data to this article can be found online at https://
though implicit stereotypes (e.g., implicit semantic associations) may doi.org/10.1016/j.jesp.2018.11.006.
be related to implicit prejudice (e.g., implicit affective associations),
they may also be independent of one another (e.g., Amodio & Devine, References
2006; Amodio & Frith, 2006; Cacioppo, Gardner, & Berntson, 1999).
Further, implicit stereotypes and prejudice may have unique effects on Amodio, D. M., & Devine, P. G. (2006). Stereotyping and evaluation in implicit race bias:
explicit attitudes and behaviors (Amodio & Devine, 2006). The current evidence for independent constructs and unique effects on behavior. Journal of
Personality and Social Psychology, 91(4), 652.
research suggests that both implicit stereotypes and implicit prejudice Amodio, D. M., & Frith, C. D. (2006). Meeting of minds: The medial frontal cortex and
are associated with policy attitudes toward redistribution. In addition, a social cognition. Nature Reviews Neuroscience, 7(4), 268–277.
similar pattern of effects emerges when we investigate the relation Andreychik, M., & Gill, M. J. (2012). Do negative implicit associations indicate negative
attitudes? Social explanations moderate whether ostensible “negative” associations
between implicit prejudice and policy attitudes controlling for implicit are prejudice-based or empathy-based. Journal of Experimental Social Psychology,
stereotypes (see Supplemental analyses). Together, this suggests that 48(5), 1082–1093. https://doi.org/10.1016/j.jesp.2012.05.006.
both implicit prejudice and implicit stereotypes uniquely predict ex- Ashok, V., Kuziemko, I., & Washington, E. (2015). Support for redistribution in an age of
rising inequality: New stylized facts and some tentative explanations. Brookings
plicit policy attitudes toward redistribution, and thus it may be fruitful Papers on Economic Activity, 367–405.
to consider both when predicting explicit attitudes and behaviors (e.g., Bergsieker, H. B., Leslie, L. M., Constantine, V. S., & Fiske, S. T. (2012). Stereotyping by
Amodio & Devine, 2006). omission: Eliminate the negative, accentuate the positive. Journal of Personality and
Social Psychology, 102(6), 1214–1238. https://doi.org/10.1037/a0027717.
Brown-Iannuzzi, J. L., Dotsch, R., Cooley, E., & Payne, B. K. (2017). The relationship
between mental representations of welfare recipients and attitudes toward welfare

33
J.L. Brown-Iannuzzi et al. Journal of Experimental Social Psychology 82 (2019) 26–34

policies. Psychological Science, 28(1), 92–103. https://doi.org/10.1177/ A-8. University of Florida. Gainesville: FL.
0956797616674999. Lee, K., Lindquist, K. A., & Payne, B. K. (2018). Constructing bias: Conceptualization
Brown-Iannuzzi, J. L., Lundberg, K. B., Kay, A. C., & Payne, B. K. (2015). Subjective status breaks the link between implicit bias and fear of Black Americans. Emotion, 18(6),
shapes political preferences. Psychological Science, 26(1), 15–26. https://doi.org/10. 855–871. https://doi.org/10.1037/emo0000347.
1177/0956797614553947. Lei, R. F., & Bodenhausen, G. V. (2017). Racial assumptions color the mental re-
Cacioppo, J. T., Gardner, W. L., & Berntson, G. G. (1999). The affect system has parallel presentation of social class. Frontiers in Psychology, 8. https://doi.org/10.3389/fpsyg.
and integrative processing components: Form follows function. Journal of personality 2017.00519.
and Social Psychology, 76(5), 839–855. McCall, L., & Kenworthy, L. (2009). Americans' social policy preferences in the era of
Cameron, C., Brown-Iannuzzi, J. L., & Payne, B. (2012). Sequential priming measures of rising inequality. Perspectives on Politics, 7(3), 459–484. https://doi.org/10.1017/
implicit social cognition: A meta-analysis of associations with behavior and explicit S1537592709990818.
attitudes. Personality and Social Psychology Review, 16(4), 330–350. https://doi.org/ Mitchell, C. J., De Houwer, J., & Lovibond, P. F. (2009). The propositional nature of
10.1177/1088868312440047. human associative learning. Behavioral and Brain Sciences, 32(2), 183–190. https://
Carmines, E. G., & Stimson, J. A. (1989). Issue Evolution. Princeton, NJ: Princeton doi.org/10.1017/S0140525X09000855.
University Press. Norton, M. I., & Ariely, D. (2011). Building a better America—One wealth quintile at a
Cooley, E., Payne, B. K., Loersch, C., & Lei, R. (2015). Who owns implicit attitudes? time. Perspectives on Psychological Science, 6(1), 9–12.
Testing a metacognitive perspective. Personality and Social Psychology Bulletin, 41(1), Nosek, B. A. (2005). Moderators of the relationship between implicit and explicit eva-
103–115. https://doi.org/10.1177/0146167214559712. luation. Journal of Experimental Psychology. General, 134(4), 565–584. https://doi.
Cooley, E., Payne, B. K., & Phillips, J. K. (2014). Implicit bias and the illusion of conscious org/10.1037/0096-3445.134.4.565.
ill will. Social Psychological and Personality Science, 5(4), 500–507. https://doi.org/10. Payne, K., & Lundberg, K. (2014). The affect misattribution procedure: Ten years of
1177/1948550613506123. evidence on reliability, validity, and mechanisms. Social and Personality Psychology
Cox, W. T. L., & Devine, P. G. (2015). Stereotypes possess heterogeneous directionality: A Compass, 8(12), 672–686.
theoretical and empirical exploration of stereotype structure and content. PLoS ONE, Payne, B. K., Cheng, C. M., Govorun, O., & Stewart, B. D. (2005). An inkblot for attitudes:
10(3), e0122292. https://doi.org/10.1371/journal.pone.0122292. Affect misattribution as implicit measurement. Journal of Personality and Social
Dovidio, J. F., Kawakami, K., & Gaertner, S. L. (2002). Implicit and explicit prejudice and Psychology, 89(3), 277–293. https://doi.org/10.1037/0022-3514.89.3.277.
interracial interaction. Journal of Personality and Social Psychology, 82(1), 62–68. Payne, B. K., Vuletich, H. A., & Lundberg, K. B. (2017). The bias of crowds: How implicit
https://doi.org/10.1037//0022-3514.82.1.62. bias bridges personal and systemic prejudice. Psychological Inquiry, 28(4), 233–248.
Faul, F., Erdfelder, E., Lang, A. G., & Buchner, A. (2007). G⁎Power 3: A flexible statistical https://doi.org/10.1080/1047840X.2017.1335568.
power analysis program for the social, behavioral, and biomedical sciences. Behavior Petty, R. E., Briñol, P., & DeMaree, K. G. (2007). The meta-cognitive model (MCM) of
Research Methods, 39(2), 175–191. https://doi.org/10.3758/BF03193146. attitudes: Implications for attitude measurement, change, and strength. Social
Fazio, R. H., & Olson, M. A. (2003). Implicit measures in social cognition research: Their Cognition, 25(5), 1–15. https://doi.org/10.1521/soco.2007.25.5.657.
meaning and use. Annual Review of Psychology, 54(1), 297–327. https://doi.org/10. Petty, R. E., Tormala, Z. L., Briñol, P., & Jarvis, W. B. G. (2006). Implicit ambivalence
1146/annurev.psych.54.101601.145225. from attitude change: An exploration of the PAST model. Journal of Personality and
Freeman, J. B., Penner, A. M., Saperstein, A., Scheutz, M., & Ambady, N. (2011). Looking Social Psychology, 90(1), 21–41. https://doi.org/10.1037/0022-3514.90.1.21.
the part: Social status cues shape race perception. PLoS ONE, 6(9), e25107. https:// Pew Research Center Social & Demographic Trends (2016). On views of race and in-
doi.org/10.1371/journal.pone.0025107. equality, Blacks and Whites are worlds apart. Retrieved from http://www.
Gawronski, B., & Bodenhausen, G. V. (2006). Associative and propositional processes in pewsocialtrends.org/2016/06/27/on-views-of-race-and-inequality-blacks-and-
evaluation: An integrative review of implicit and explicit attitude change. whites-are-worlds-apart/.
Psychological Bulletin, 132(5), 692–731. https://doi.org/10.1037/0033-2909.132.5. Plant, E. A., & Devine, P. G. (1998). Internal and external motivation to respond without
692. prejudice. Journal of Personality and Social Psychology, 75(3), 811–832. https://doi.
Gawronski, B., & Ye, Y. (2014). What drives priming effects in the affect misattribution org/10.1037/0022-3514.75.3.811.
procedure? Personality and Social Psychology Bulletin, 40(1), 3–15. https://doi.org/10. Preacher, K. J., Curran, P. J., & Bauer, D. J. (2006). Computational tools for probing
1177/0146167213502548. interaction effects in multiple linear regression, multilevel modeling, and latent curve
Gilens, M. (1995). Racial attitudes and opposition to welfare. Journal of Politics, 57(4), analysis. Journal of Educational and Behavioral Statistics, 31(4), 437–448. https://doi.
994–1014. https://doi.org/10.2307/2960399. org/10.3102/10769986031004437.
Gilens, M. (1996). “Race coding” and White opposition to welfare. American Political Quillian, L., Pager, D., Hexel, O., & Midtboen, A. H. (2017). Meta-analysis of field ex-
Science Review, 90(3), 593–604. https://doi.org/10.2307/2082611. periments shows no change in racial discrimination in hiring over time. Proceedings of
Goh, J. X., Hall, J. A., & Rosenthal, R. (2016). Mini meta-analysis of your own studies: the National Academy of Sciences, 114(41), 10870–10875. https://doi.org/10.1073/
Some arguments on why and a primer on how. Social and Personality Psychology pnas.1706255114.
Compass, 10(10), 535–549. Saez, E., & Zucman, E. (2014). Wealth inequality in the United States since 1913:
Greenwald, A. G., Poehlman, T. A., Uhlmann, E. L., & Banaji, M. R. (2009). Understanding Evidence from capitalized income tax data. Working Paper No. 20625. Retrieved
and using the Implicit Association Test: III. Meta-analysis of predictive validity. from http://www.nber.org/papers/w20625.
Journal of Personality and Social Psychology, 97(1), 17–41. https://doi.org/10.1037/ Sava, F. A., Maricuτoiu, L. P., Rusu, S., Macsinga, I., Vîrgă, D., Cheng, C. M., & Payne, B.
a0015575. K. (2012). An inkblot for the implicit assessment of personality: The semantic mis-
Gregg, A. P., Seibt, B., & Banaji, M. R. (2006). Easier done than undone: Asymmetry in the attribution procedure. European Journal of Personality, 26(6), 613–628. https://doi.
malleability of implicit preferences. Journal of Personality and Social Psychology, org/10.1002/per.1861.
90(1), 1–20. https://doi.org/10.1037/0022-3514.90.1.1. Sears, D. O., & Cirin, J. (1985). Tax revolt: Something for nothing in California. Cambridge,
Hayes, A. F. (2009). Beyond Baron and Kenny: Statistical mediation analysis in the new MA: Harvard University Press.
millennium. Communication Monographs, 76(4), 408–420. https://doi.org/10.1080/ Shrout, P. E., & Bolger, N. (2002). Mediation in experimental and nonexperimental stu-
03637750903310360. dies: New procedures and recommendations. Psychological Methods, 7(4), 422–445.
Hayes, A. F. (2017). Introduction to mediation, moderations, and conditional process analysis: https://doi.org/10.1037/1082-989X.7.4.422.
A regression-based approach. New York: The Guilford Press. Uhlmann, E. L., Brescoll, V. L., & Paluck, E. L. (2006). Are members of low status groups
Henry, P. J., & Sears, D. O. (2002). The symbolic racism 2000 scale. Political Psychology, perceived as bad, or badly off? Egalitarian negative associations and automatic
23(2), 253–283. https://doi.org/10.1111/0162-895X.00281. prejudice. Journal of Experimental Social Psychology, 42(4), 491–499. https://doi.org/
Imhoff, R., Schmidt, A. F., Bernhardt, J., Dierksmeier, A., & Banse, R. (2011). An inkblot 10.1016/j.jesp.2004.10.003.
for sexual preference: A semantic variant of the Affect Misattribution Procedure. Wetts, R., & Willer, R. (2018). Privilege on the precipice: Perceived racial status threats
Cognition and Emotion, 25(4), 676–690. https://doi.org/10.1080/02699931.2010. lead White Americans to oppose welfare programs. Social Forces. https://doi.org/10.
508260. 1093/sf/soy046.
Iyengar, S. (1990). Framing responsibility for political issues: The case of poverty. Political Wolff, E. N. (2002). Top heavy: The increasing inequality of wealth in America and what can
Behavior, 12(1), 19–40. https://doi.org/10.1007/BF00992330. be done about it. New York, NY: New Press.
Kluegel, J. R., & Smith, E. R. (1986). Beliefs about inequality: Americans' views of what is and Ye, Y., & Gawronski, B. (2018). Validating the semantic misattribution procedure as an
what ought to be. New York, NY: Aline De Grader. implicit measure of gender stereotyping. European Journal of Social Psychology, 48(3),
Lang, P. J., Bradley, M. M., & Cuthbert, B. N. (2008). International affective picture 348–364. https://doi.org/10.1002/ejsp.2337.
system (IAPS): Affective ratings of pictures and instruction manual. Technical report

34