MIFT Study 1 Analyses

Descriptive Statistics

  • stable:

    Table continues below
      vars n mean sd median trimmed
    whitemulti_trust_dif 1 172 0.07703 0.5416 0.05 0.06866
    White 2 172 3.16 0.5696 3.125 3.156
    Multiracial 3 172 3.083 0.5688 3.075 3.082
    Black 4 172 3.147 0.6911 3.213 3.182
    trust_survey 5 177 5.003 0.8495 5.1 5.017
    trust_therm 6 177 72.42 20.05 75 73.34
    atma 7 177 3.602 0.5185 3.565 3.582
    feel_therm 8 177 75.59 19.2 77 76.53
    authen 9 177 5.573 0.9506 5.75 5.617
    race_ess 10 177 3.819 1.245 4 3.891
    pol_or 11 177 3.073 1.742 3 2.93
    Table continues below
      mad min max range skew
    whitemulti_trust_dif 0.3336 -1.625 2.075 3.7 0.3512
    White 0.4448 1.1 5 3.9 0.002589
    Multiracial 0.5004 1.6 5 3.4 0.09065
    Black 0.5374 1.025 5 3.975 -0.4093
    trust_survey 0.8896 2.1 7 4.9 -0.3613
    trust_therm 23.72 0 100 100 -0.4798
    atma 0.5157 1.913 5 3.087 0.2713
    feel_therm 22.24 0 100 100 -0.5477
    authen 1.112 2.25 7 4.75 -0.4685
    race_ess 0.9266 1 6.75 5.75 -0.5099
    pol_or 1.483 1 7 6 0.4717
      kurtosis se
    whitemulti_trust_dif 2.504 0.0413
    White 1.205 0.04343
    Multiracial 0.6047 0.04337
    Black 0.8893 0.0527
    trust_survey 0.01425 0.06385
    trust_therm -0.308 1.507
    atma 0.2957 0.03897
    feel_therm -0.0686 1.443
    authen 0.004316 0.07145
    race_ess -0.1948 0.09361
    pol_or -0.7958 0.1309

  • fluid:

    Table continues below
      vars n mean sd median trimmed
    whitemulti_trust_dif 1 179 0.1111 0.4736 0.05 0.09078
    White 2 179 3.065 0.5146 3.05 3.077
    Multiracial 3 179 2.954 0.4961 3 2.978
    Black 4 179 3.051 0.5911 3.075 3.072
    trust_survey 5 181 4.99 0.8156 5 4.994
    trust_therm 6 181 71.4 18.12 75 71.08
    atma 7 181 3.5 0.4915 3.478 3.492
    feel_therm 8 181 74.27 18.52 76 74.47
    authen 9 181 5.421 0.9452 5.5 5.45
    race_ess 10 181 3.791 1.222 4 3.844
    pol_or 11 181 2.801 1.614 2 2.641
    Table continues below
      mad min max range skew
    whitemulti_trust_dif 0.3336 -1.85 2.05 3.9 0.4136
    White 0.4448 1.1 4.35 3.25 -0.5168
    Multiracial 0.4077 1 4 3 -0.6924
    Black 0.4818 1.025 4.7 3.675 -0.4769
    trust_survey 0.8896 1.5 7 5.5 -0.4737
    trust_therm 22.24 23 100 77 -0.09943
    atma 0.5157 2.174 4.783 2.609 0.08524
    feel_therm 22.24 26 100 74 -0.1666
    authen 1.112 2.5 7 4.5 -0.4124
    race_ess 1.112 1 6.875 5.875 -0.3603
    pol_or 1.483 1 7 6 0.5273
      kurtosis se
    whitemulti_trust_dif 3.058 0.0354
    White 1.625 0.03847
    Multiracial 1.561 0.03708
    Black 0.6772 0.04418
    trust_survey 0.8097 0.06063
    trust_therm -0.9839 1.347
    atma -0.1394 0.03654
    feel_therm -1.183 1.377
    authen -0.08457 0.07026
    race_ess -0.3309 0.09086
    pol_or -0.7234 0.12

GEE Model for Speeded Trust Task

From preregistration:

For the speeded trust task, we will fit a GEE regression model with trustworthiness ratings for each target face as the outcome and target race (within subjects—will be dummy coded with White as the comparison group), condition (between subjects), and their interactions as predictors.

Findings:

  • The comparison group is White faces in the stable condition. The intercept is 3.1 on a scale from 1 to 5 (3 midpoint) with a higher score indicating greater trust.
  • Main effects
    • There is a main effect of condition such that participants in the fluid condition report slightly less trust overall than participants in the stable condition.
    • There is also a main effect of the Multiracial dummy code such that participants in both conditions report slightly less trust for Multiracial faces relative to White faces.
    • There is also a main effect of the Black dummy code such that participants in both conditions report slightly less trust for Black faces relative to White faces.
  • All of these main effects are qualified by significant interactions (the interpretation of the interactions changed a bit).
    • There is an interaction between condition and the Multiracial dummy code such that, in the fluid condition, participants show significantly lower trust for Multiracial people than White people but there is no significant difference in the stable condition.
    • There is an interaction between condition and the Black dummy code such that people in the fluid condition show less trust for faces that are not Black than participants in the stable condition. There is no significant difference between conditions when faces are Black. This one doesn’t really make much sense to me.
##           (Intercept)            condition1            race_multi 
##           3.114003831          -0.047059387          -0.093357280 
##            race_black condition1:race_multi condition1:race_black 
##          -0.013201628          -0.016920498          -0.000270594
Estimates for the speeded trust model
term estimate std.error statistic p.value NA
(Intercept) 3.1140 0.0093 335.5444 0.0287 108.5113
condition1 -0.0471 0.0093 -5.0708 0.0287 -1.6398
race_multi -0.0934 0.0131 -7.1132 0.0269 -3.4712
race_black -0.0132 0.0131 -1.0059 0.0423 -0.3119
condition1:race_multi -0.0169 0.0131 -1.2892 0.0269 -0.6291
condition1:race_black -0.0003 0.0131 -0.0206 0.0423 -0.0064
Emmeans for condition * race_multi
condition race_multi emmean SE df asymp.LCL asymp.UCL
stable 0 3.1546 0.0342 Inf 3.0876 3.2216
fluid 0 3.0602 0.0316 Inf 2.9982 3.1222
stable 1 3.0782 0.0582 Inf 2.9641 3.1922
fluid 1 2.9499 0.0493 Inf 2.8534 3.0465
Post hoc tests for race_multi * condition
contrast condition estimate SE df z.ratio p.value
race_multi1 - race_multi0 stable -0.0764 0.0407 Inf -1.8760 0.0607
race_multi1 - race_multi0 fluid -0.1103 0.0351 Inf -3.1403 0.0017
Emmeans for condition * race_black
condition race_black emmean SE df asymp.LCL asymp.UCL
stable 0 3.1228 0.0376 Inf 3.0491 3.1966
fluid 0 3.0118 0.0332 Inf 2.9467 3.0769
stable 1 3.1099 0.0671 Inf 2.9784 3.2414
fluid 1 2.9983 0.0556 Inf 2.8893 3.1073
Post hoc tests for condition * condition
contrast race_black estimate SE df z.ratio p.value
fluid - stable 0 -0.1110 0.0502 Inf -2.2120 0.0270
fluid - stable 1 -0.1116 0.0872 Inf -1.2802 0.2005

T-tests for main dependent variables for differences between conditions

From preregistration:

For each of the five other dependent variables (listed below), we will run independent samples T-tests to determine if there is any difference between participants who read the fluidity article and participants who read the stability article.

Trust Survey

Findings:

There is not a significant difference in explicit trust on the trust survey between conditions.

Trust survey by condition (continued below)
Test statistic df P value Alternative hypothesis
0.1513 354.6 0.8798 two.sided
mean in group stable mean in group fluid
5.003 4.99

Trust Thermometer

Findings:

There is not a significant difference in explicit trust on the trust thermometer between conditions.

Trust thermometer by condition (continued below)
Test statistic df P value Alternative hypothesis
0.5076 350.7 0.6121 two.sided
mean in group stable mean in group fluid
72.42 71.4

Attitudes towards multiracial adults

Findings:

There is not a significant difference in attitudes towards multiracial adults between conditions. Though it is worth noting that there is a marginal difference in line with our hypothesis (more negative attitudes in fluid condition).

Attitudes towards multiracial adults by condition (continued below)
Test statistic df P value Alternative hypothesis mean in group stable
1.908 354 0.05716 two.sided 3.602
mean in group fluid
3.5

Feeling thermometer

Findings:

There is not a significant difference in feelings towards multiracial people on the feeling thermometer between conditions.

Feeling thermometer by condition (continued below)
Test statistic df P value Alternative hypothesis
0.6658 354.8 0.506 two.sided
mean in group stable mean in group fluid
75.59 74.27

Authenticity

Findings:

There is not a significant difference in perceived authenticity of multiracial people between conditions.

Authenticity by condition (continued below)
Test statistic df P value Alternative hypothesis
1.519 355.7 0.1297 two.sided
mean in group stable mean in group fluid
5.573 5.421

Exploratory Analyses.

From Preregistration:

“We will run an exploratory independent samples T-test to determine if there is any difference in race essentialism between participants who read the fluidity article and participants who read the stability article. If we find that there are no differences in race essentialism between conditions, we will examine whether race essentialism moderates the relationship between which article participants read and each of the six dependent variables.”

Race Essentialism

T-test

Findings:

There is not a significant difference in race essentialism between conditions Thus, we will move forward to examine whether race essentialism moderates the relationship between condition and the dependent variables.

Race essentialism by condition (continued below)
Test statistic df P value Alternative hypothesis
0.2182 355.4 0.8274 two.sided
mean in group stable mean in group fluid
3.819 3.791

Speeded trust task by condition + race essentialism as moderator

From preregistration:

“For the speeded trust task, we will first calculate a White/Multiracial trust difference score by subtracting participant’s average trust ratings for all multiracial targets from participant’s average trust ratings for all White targets, with a higher number indicating greater trust towards White targets relative to multiracial targets. We then will fit a linear regression model with the White/Multiracial trust difference score as the outcome and condition, race essentialism, and the interaction term between the two as predictors. If we find that the interactions is significant, we will conduct simple slopes analysis to probe the interaction.”

Findings:

  • Main Effects
    • There is no main effect of condition. Confused by this as there is when analyzed with the GEE.
    • There is a main effect of race essentialism such that participants who are more essentialist about race show greater trust towards White targets relative to Multiracial targets.
  • Interaction
    • There is no significant interaction. In other words, race essentialism does not moderate the relationship between condition and trust.
Estimates for the speeded trust model
term estimate std.error statistic p.value
(Intercept) -0.2416 0.0857 -2.8201 0.0051
condition1 0.0606 0.0857 0.7074 0.4798
race_ess 0.0885 0.0215 4.1201 0.0000
condition1:race_ess -0.0113 0.0215 -0.5258 0.5993
Summary for the speeded trust model
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.0484 0.0402 0.4973 5.8884 0.0006 3 -250.859 511.717 531.021 85.8243 347 351

Other main DVs by condition + race essentialism as moderator

From preregistration:

“For each of the five other dependent variables, we will fit an individual linear regression model including condition, race essentialism, and the interaction term between the two as predictors. If we find a significant interaction between condition and race essentialism for any of the models, we will conduct simple slopes analysis for that model to probe the interaction.”

Trust survey

Findings:

  • Main Effects
    • There is no main effect of condition.
    • There is a main effect of race essentialism such that participants who are more essentialist about race show less trust towards multiracial people.
  • Interaction
    • There is no significant interaction. In other words, race essentialism does not moderate the relationship between condition and trust.
Estimates for the trust survey model
term estimate std.error statistic p.value
(Intercept) 5.4749 0.1408 38.8855 0.0000
condition1 0.0921 0.1408 0.6541 0.5135
race_ess -0.1259 0.0352 -3.5762 0.0004
condition1:race_ess -0.0264 0.0352 -0.7505 0.4535
Summary for the trust survey model
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.0363 0.0281 0.8196 4.4455 0.0044 3 -434.762 879.524 898.927 237.815 354 358

Trust Thermometer

Findings:

  • Main Effects
    • There is no main effect of condition.
    • There is a main effect of race essentialism such that participants who are more essentialist about race show less trust towards multiracial people.
  • Interaction
    • There is no significant interaction. In other words, race essentialism does not moderate the relationship between condition and trust.
Estimates for the trust thermometer model
term estimate std.error statistic p.value
(Intercept) 80.5490 3.2544 24.7508 0.0000
condition1 -0.2977 3.2544 -0.0915 0.9272
race_ess -2.2705 0.8138 -2.7899 0.0056
condition1:race_ess -0.0651 0.8138 -0.0800 0.9363
Summary for the trust thermometer model
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.0222 0.0139 18.9453 2.6831 0.0466 3 -1559.05 3128.09 3147.5 127060 354 358

Attitudes towards multiracial adults

Findings:

  • Main Effects
    • There is no main effect of condition.
    • There is a main effect of race essentialism such that participants who are more essentialist about race show show more negative attitudes towards multiracial people.
  • Interaction
    • There is no significant interaction. In other words, race essentialism does not moderate the relationship between condition and attitudes towards multiracial adults.
Estimates for the attitudes towards multiracial adults model
term estimate std.error statistic p.value
(Intercept) 3.7421 0.0863 43.3493 0.0000
condition1 -0.0834 0.0863 -0.9661 0.3346
race_ess -0.0502 0.0216 -2.3243 0.0207
condition1:race_ess 0.0083 0.0216 0.3861 0.6996
Summary for the attitudes towards multiracial adults model
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.0255 0.0172 0.5025 3.0826 0.0275 3 -259.633 529.267 548.669 89.4001 354 358

Feeling thermometer

Findings:

  • Main Effects
    • There is no main effect of condition.
    • There is a main effect of race essentialism such that participants who are more essentialist about race show more negative attitudes towards multiracial people.
  • Interaction
    • There is no significant interaction. In other words, race essentialism does not moderate the relationship between condition and attitudes.
Estimates for the feeling thermometer model
term estimate std.error statistic p.value
(Intercept) 82.4914 3.2175 25.6380 0.0000
condition1 2.2008 3.2175 0.6840 0.4944
race_ess -1.9903 0.8046 -2.4736 0.0138
condition1:race_ess -0.7604 0.8046 -0.9450 0.3453
Summary for the feeling thermometer model
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.0206 0.0122 18.7308 2.4758 0.0612 3 -1554.97 3119.94 3139.34 124198 354 358

Authenticity

Findings:

  • Main Effects
    • There is no main effect of condition.
    • There is a main effect of race essentialism such that participants who are more essentialist about race perceive multiracial people as less authentic.
  • Interaction
    • There is no significant interaction. In other words, race essentialism does not moderate the relationship between condition and authenticity.
Estimates for the authenticity model
term estimate std.error statistic p.value
(Intercept) 6.1840 0.1587 38.9668 0.0000
condition1 -0.1199 0.1587 -0.7557 0.4503
race_ess -0.1804 0.0397 -4.5460 0.0000
condition1:race_ess 0.0108 0.0397 0.2733 0.7848
Summary for the authenticity model
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.0615 0.0535 0.9239 7.7295 0.0001 3 -477.615 965.23 984.633 302.142 354 358

Race & political orientation

From preregistration:

Additionally, we will conduct exploratory analyses to examine whether participants’ responses vary by participant race and political orientation..

Personally I would not read into the participant race effects too much due to the small Ns in some cells.

Participant Race
White Black Latine Asian Other
271 35 18 22 9

Speeded trust task by condition + participant race and political orientation as moderators

From preregistration:

For the speeded trust task, we will again use the White/Multiracial trust difference score as described above. We then will fit a linear regression model with the White/Multiracial trust difference score as the outcome and condition, participant race, political orientation, and the two two-way interaction terms between condition, and participant race/political orientation (i.e., condition X participant race, condition X political orientation) as predictors. If we find that any of the interactions are significant, we will conduct simple slopes analysis to probe the interaction.

Findings:

  • Main Effects
    • There is no main effect of condition
    • There are no main effects of participant race
    • There is a main effect of political orientation such that participants that are more conservative show greater trust for white faces relative to multiracial faces.
  • Interactions
    • There are no interactions between participant race and condition
    • There is no interaction between condition and political orientation.
White, Black, Latine, Asian and Other
Estimates for the speeded trust model
term estimate std.error statistic p.value
(Intercept) -0.1724 0.0681 -2.5301 0.0119
condition1 0.1030 0.0681 1.5109 0.1318
race_factor1 -0.0685 0.0440 -1.5558 0.1207
race_factor2 0.0162 0.0421 0.3837 0.7014
race_factor3 0.0440 0.0292 1.5044 0.1334
race_factor4 -0.0295 0.0360 -0.8193 0.4132
pol_or 0.0828 0.0159 5.2209 0.0000
condition1:race_factor1 0.0248 0.0440 0.5644 0.5729
condition1:race_factor2 0.0452 0.0421 1.0744 0.2834
condition1:race_factor3 -0.0428 0.0292 -1.4658 0.1436
condition1:race_factor4 -0.0160 0.0360 -0.4445 0.6570
condition1:pol_or -0.0235 0.0159 -1.4831 0.1390
Summary for the speeded trust model
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.1076 0.0784 0.4883 3.6841 0.0001 11 -238.248 502.495 552.574 80.1231 336 348

Other main DVs by condition + participant race and political orientation as moderators

From preregistration:

For each of the five other dependent variables, we will fit an individual linear regression model including condition, participant race, political orientation, and the two-way interaction terms between condition and each demographic variable (i.e., condition X participant race, condition X political orientation) as predictors. If we find a significant effect for either of the interaction terms for any of the models, we will conduct simple slopes analysis for that model to probe the interaction.

Trust survey

Findings:

  • Main Effects
    • There is no main effect of condition
    • There are no main effects of participant race aside from Black participants. Black participants show lower trust for multiracial people.
    • There is a main effect of political orientation such that participants that are more conservative show lower trust for multiracial people.
  • Interactions
    • There are no interactions between participant race and condition
    • There is no interaction between condition and political orientation
White, Black, Latine, Asian and Other
Estimates for the trust survey model
term estimate std.error statistic p.value
(Intercept) 5.1043 0.1129 45.2268 0.0000
condition1 -0.0379 0.1129 -0.3361 0.7370
race_factor1 -0.1914 0.0732 -2.6160 0.0093
race_factor2 0.0123 0.0701 0.1752 0.8610
race_factor3 -0.0673 0.0486 -1.3837 0.1673
race_factor4 -0.0066 0.0598 -0.1106 0.9120
pol_or -0.0950 0.0261 -3.6423 0.0003
condition1:race_factor1 -0.1017 0.0732 -1.3898 0.1655
condition1:race_factor2 0.0248 0.0701 0.3536 0.7238
condition1:race_factor3 0.0754 0.0486 1.5503 0.1220
condition1:race_factor4 0.0178 0.0598 0.2982 0.7657
condition1:pol_or 0.0145 0.0261 0.5573 0.5777
Summary for the trust survey model
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.0798 0.0503 0.8127 2.705 0.0023 11 -423.989 873.978 924.316 226.538 343 355

Trust Thermometer

Findings:

  • Main Effects
    • There is no main effect of condition
    • There are no main effects of participant race aside from Asian participants. Asian participants show lower trust for multiracial people.
    • There is a main effect of political orientation such that participants that are more conservative show less trust for multiracial people.
  • Interactions
    • There are no interactions between participant race and condition
    • There is no interaction between condition and political orientation
White, Black, Latine, Asian and Other
Estimates for the trust thermometer model
term estimate std.error statistic p.value
(Intercept) 74.0191 2.6147 28.3085 0.0000
condition1 -1.4210 2.6147 -0.5435 0.5872
race_factor1 -1.3787 1.6948 -0.8135 0.4165
race_factor2 0.0106 1.6232 0.0065 0.9948
race_factor3 -2.4172 1.1268 -2.1453 0.0326
race_factor4 -0.4964 1.3857 -0.3582 0.7204
pol_or -1.8237 0.6042 -3.0181 0.0027
condition1:race_factor1 -1.7756 1.6948 -1.0477 0.2955
condition1:race_factor2 -0.4378 1.6232 -0.2697 0.7876
condition1:race_factor3 1.6171 1.1268 1.4352 0.1521
condition1:race_factor4 1.1699 1.3857 0.8443 0.3991
condition1:pol_or 0.4852 0.6042 0.8030 0.4225
Summary for the trust thermometer model
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.0585 0.0283 18.8281 1.9362 0.0341 11 -1539.67 3105.34 3155.67 121592 343 355

Attitudes towards multiracial adults

Findings:

  • Main Effects
    • There is no main effect of condition
    • There are no main effects of participant race aside from Black participants. Black participants show more negative attitudes towards multiracial people than White participants.
    • There is no main effect of political orientation.
  • Interactions
    • There are no interactions between participant race and condition
    • There is no interaction between condition and political orientation
White, Black, Latine, Asian and Other
Estimates for the attitudes towards multiracial adults model
term estimate std.error statistic p.value
(Intercept) 3.4702 0.0697 49.8159 0.0000
condition1 -0.0400 0.0697 -0.5742 0.5662
race_factor1 -0.1200 0.0452 -2.6578 0.0082
race_factor2 -0.0312 0.0432 -0.7208 0.4715
race_factor3 0.0095 0.0300 0.3167 0.7517
race_factor4 -0.0341 0.0369 -0.9234 0.3565
pol_or -0.0166 0.0161 -1.0334 0.3021
condition1:race_factor1 -0.0049 0.0452 -0.1087 0.9135
condition1:race_factor2 -0.0495 0.0432 -1.1440 0.2534
condition1:race_factor3 0.0432 0.0300 1.4381 0.1513
condition1:race_factor4 0.0470 0.0369 1.2744 0.2034
condition1:pol_or 0.0076 0.0161 0.4734 0.6362
Summary for the attitudes towards multiracial adults model
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.0538 0.0235 0.5016 1.7741 0.0571 11 -252.696 531.391 581.729 86.3042 343 355

Feeling thermometer

Findings:

  • Main Effects
    • There is no main effect of condition
    • There are no main effects of participant race aside from Asian participants. Asian participants show more negative attitudes towards multiracial people than White participants.
    • There is a main effect of political orientation such that participants that are more conservative show more negative feelings towards multiracial people.
  • Interactions
    • There are no interactions between participant race and condition.
    • There is no interaction between condition and political orientation
White, Black, Latine, Asian and Other
Estimates for the feeling thermometer model
term estimate std.error statistic p.value
(Intercept) 77.7497 2.5667 30.2916 0.0000
condition1 -1.5771 2.5667 -0.6144 0.5393
race_factor1 -2.0650 1.6636 -1.2413 0.2154
race_factor2 0.5355 1.5934 0.3361 0.7370
race_factor3 -2.6335 1.1061 -2.3809 0.0178
race_factor4 -0.6549 1.3602 -0.4814 0.6305
pol_or -2.1199 0.5931 -3.5740 0.0004
condition1:race_factor1 -1.0773 1.6636 -0.6475 0.5177
condition1:race_factor2 -0.5358 1.5934 -0.3363 0.7369
condition1:race_factor3 1.1686 1.1061 1.0565 0.2915
condition1:race_factor4 0.9056 1.3602 0.6658 0.5060
condition1:pol_or 0.4697 0.5931 0.7920 0.4289
Summary for the feeling thermometer model
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.071 0.0412 18.4823 2.3829 0.0075 11 -1533.09 3092.18 3142.51 117167 343 355

Authenticity

Findings:

  • Main Effects
    • There is no main effect of condition
    • There are no main effects of participant race aside from Black participants. Black participants perceive multiracial people as less authentic than White participants.
    • There is a main effect of political orientation such that participants that are more conservative perceive multiracial people as less authentic.
  • Interactions
    • There are no interactions between participant race and condition.
    • There is no interaction between condition and political orientation
White, Black, Latine, Asian and Other
Estimates for the authenticity model
term estimate std.error statistic p.value
(Intercept) 5.6123 0.1261 44.5152 0.0000
condition1 -0.1258 0.1261 -0.9979 0.3191
race_factor1 -0.3267 0.0817 -3.9975 0.0001
race_factor2 0.0343 0.0783 0.4383 0.6614
race_factor3 -0.0173 0.0543 -0.3187 0.7501
race_factor4 -0.0201 0.0668 -0.3011 0.7635
pol_or -0.1176 0.0291 -4.0371 0.0001
condition1:race_factor1 -0.0456 0.0817 -0.5585 0.5769
condition1:race_factor2 -0.0669 0.0783 -0.8545 0.3934
condition1:race_factor3 0.0479 0.0543 0.8815 0.3786
condition1:race_factor4 0.1162 0.0668 1.7396 0.0828
condition1:pol_or 0.0308 0.0291 1.0577 0.2909
Summary for the authenticity model
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.1149 0.0866 0.9078 4.0495 0 11 -463.298 952.595 1002.93 282.695 343 355

Additional Exploratory Descriptives looking at main Dvs by participant race

  • White:

    Table continues below
      vars n mean sd median trimmed
    whitemulti_trust_dif 1 264 0.1132 0.519 0.075 0.09298
    White 2 264 3.168 0.5267 3.138 3.164
    Multiracial 3 264 3.055 0.5296 3.025 3.053
    Black 4 264 3.127 0.6449 3.175 3.156
    trust_survey 5 271 5.071 0.8156 5.2 5.08
    trust_therm 6 271 72.76 18.97 75 73.1
    atma 7 271 3.595 0.516 3.565 3.582
    feel_therm 8 271 76.15 18.64 80 76.98
    authen 9 271 5.585 0.9213 5.75 5.624
    race_ess 10 271 3.78 1.282 4 3.851
    pol_or 11 271 2.989 1.714 3 2.839
    Table continues below
      mad min max range skew
    whitemulti_trust_dif 0.3336 -1.85 2.075 3.925 0.4582
    White 0.4262 1.1 5 3.9 -0.01588
    Multiracial 0.4448 1.2 5 3.8 0.02322
    Black 0.556 1.025 5 3.975 -0.4322
    trust_survey 0.8896 2.1 7 4.9 -0.3306
    trust_therm 22.24 0 100 100 -0.3757
    atma 0.5157 1.913 5 3.087 0.1061
    feel_therm 22.24 0 100 100 -0.5009
    authen 0.7413 2.75 7 4.25 -0.4259
    race_ess 1.112 1 6.875 5.875 -0.4329
    pol_or 1.483 1 7 6 0.4614
      kurtosis se
    whitemulti_trust_dif 2.98 0.03194
    White 1.279 0.03241
    Multiracial 0.9763 0.03259
    Black 0.6574 0.03969
    trust_survey -0.01961 0.04955
    trust_therm -0.4341 1.152
    atma 0.2125 0.03134
    feel_therm -0.2696 1.132
    authen -0.224 0.05596
    race_ess -0.3173 0.07788
    pol_or -0.8882 0.1041

  • Black:

    Table continues below
      vars n mean sd median trimmed
    whitemulti_trust_dif 1 35 -0.03 0.4438 0 -0.005172
    White 2 35 2.888 0.6321 3 2.93
    Multiracial 3 35 2.918 0.5757 2.975 2.955
    Black 4 35 3.162 0.6532 3.1 3.157
    trust_survey 5 35 4.691 0.9147 5 4.759
    trust_therm 6 35 70.14 20.16 71 70.52
    atma 7 35 3.357 0.4269 3.348 3.325
    feel_therm 8 35 72.2 20.43 70 72.31
    authen 9 35 4.943 1.047 5 5.009
    race_ess 10 35 4.261 0.9579 4.375 4.302
    pol_or 11 35 2.943 1.697 3 2.759
    Table continues below
      mad min max range skew
    whitemulti_trust_dif 0.2965 -1.425 1.05 2.475 -0.6684
    White 0.4448 1.1 4.4 3.3 -0.8013
    Multiracial 0.3336 1 4.2 3.2 -1.01
    Black 0.4077 1.025 5 3.975 -0.1933
    trust_survey 0.8896 1.5 6.2 4.7 -1.182
    trust_therm 28.17 30 100 70 -0.1753
    atma 0.4512 2.565 4.304 1.739 0.6286
    feel_therm 29.65 35 100 65 0.02912
    authen 0.7413 2.25 6.75 4.5 -0.5817
    race_ess 0.556 2 6 4 -0.5097
    pol_or 1.483 1 7 6 0.648
      kurtosis se
    whitemulti_trust_dif 1.857 0.07502
    White 1.998 0.1068
    Multiracial 2.636 0.09731
    Black 2.778 0.1104
    trust_survey 2.311 0.1546
    trust_therm -1.124 3.407
    atma -0.244 0.07216
    feel_therm -1.505 3.453
    authen 0.3079 0.177
    race_ess 0.004889 0.1619
    pol_or -0.2651 0.2868

  • Latine:

    Table continues below
      vars n mean sd median trimmed
    whitemulti_trust_dif 1 18 -0.009722 0.5842 -0.0125 -0.02187
    White 2 18 2.861 0.5004 2.975 2.883
    Multiracial 3 18 2.871 0.4184 2.9 2.866
    Black 4 18 2.986 0.6347 3 2.998
    trust_survey 5 18 4.994 0.7787 5.1 5.019
    trust_therm 6 18 73.67 19.42 75.5 74.5
    atma 7 18 3.442 0.562 3.413 3.418
    feel_therm 8 18 78.06 17.84 80 78.44
    authen 9 18 5.528 0.9621 5.5 5.547
    race_ess 10 18 3.597 1.044 3.562 3.602
    pol_or 11 18 2.333 1.455 2 2.188
    Table continues below
      mad min max range skew
    whitemulti_trust_dif 0.3521 -1.275 1.45 2.725 0.1618
    White 0.4262 1.825 3.55 1.725 -0.6074
    Multiracial 0.4077 2.1 3.725 1.625 0.0933
    Black 0.5004 1.375 4.4 3.025 -0.2561
    trust_survey 0.9637 3.6 6 2.4 -0.3004
    trust_therm 23.72 34 100 66 -0.3494
    atma 0.4512 2.435 4.826 2.391 0.3272
    feel_therm 22.24 50 100 50 -0.2701
    authen 0.556 3.75 7 3.25 -0.1793
    race_ess 0.834 1.5 5.625 4.125 0.06359
    pol_or 1.483 1 6 5 0.8533
      kurtosis se
    whitemulti_trust_dif 0.8556 0.1377
    White -0.8959 0.1179
    Multiracial -0.8162 0.09863
    Black 0.912 0.1496
    trust_survey -1.362 0.1836
    trust_therm -1.048 4.578
    atma 0.1102 0.1325
    feel_therm -1.335 4.205
    authen -0.7046 0.2268
    race_ess -0.2793 0.246
    pol_or -0.1919 0.343

  • Asian:

    Table continues below
      vars n mean sd median trimmed
    whitemulti_trust_dif 1 22 0.1909 0.4674 0.15 0.1708
    White 2 22 3.09 0.5096 3.05 3.079
    Multiracial 3 22 2.899 0.5839 2.925 2.938
    Black 4 22 2.815 0.5883 2.912 2.878
    trust_survey 5 22 4.705 0.7371 4.65 4.733
    trust_therm 6 22 63.18 17.42 63 63.11
    atma 7 22 3.5 0.4474 3.478 3.495
    feel_therm 8 22 65.45 17.28 66 65.28
    authen 9 22 5.273 0.8342 5.125 5.292
    race_ess 10 22 3.483 1.165 3.688 3.493
    pol_or 11 22 2.5 1.371 2 2.389
    Table continues below
      mad min max range skew
    whitemulti_trust_dif 0.5189 -0.6 1.275 1.875 0.3943
    White 0.5189 2.025 4.15 2.125 0.1512
    Multiracial 0.7598 1.675 3.65 1.975 -0.4719
    Black 0.6116 1.45 3.675 2.225 -0.7905
    trust_survey 0.9637 3.1 5.6 2.5 -0.2579
    trust_therm 19.27 26 100 74 0.01099
    atma 0.3868 2.696 4.391 1.696 0.1335
    feel_therm 21.5 33 100 67 0.09723
    authen 0.9266 3.75 6.5 2.75 -0.06825
    race_ess 1.297 1.625 5.625 4 -0.08792
    pol_or 1.483 1 5 4 0.37
      kurtosis se
    whitemulti_trust_dif -0.5158 0.09966
    White -0.6262 0.1086
    Multiracial -0.9505 0.1245
    Black 0.009586 0.1254
    trust_survey -1.201 0.1571
    trust_therm -0.4796 3.715
    atma -0.7839 0.09538
    feel_therm -1.009 3.684
    authen -1.296 0.1779
    race_ess -1.346 0.2484
    pol_or -1.274 0.2924

  • Other:

    Table continues below
      vars n mean sd
    whitemulti_trust_dif 1 9 -0.00000000000000006759 0.3034
    White 2 9 2.819 0.5135
    Multiracial 3 9 2.819 0.6048
    Black 4 9 2.886 0.6531
    trust_survey 5 9 4.767 1.078
    trust_therm 6 9 67 19.92
    atma 7 9 3.329 0.354
    feel_therm 8 9 68.67 20.2
    authen 9 9 5.25 1.008
    race_ess 10 9 3.917 1.077
    pol_or 11 9 3.444 1.81
    Table continues below
      median trimmed mad min
    whitemulti_trust_dif 0.05 -0.00000000000000004934 0.1483 -0.625
    White 2.725 2.819 0.2965 1.95
    Multiracial 3 2.819 0.556 1.6
    Black 3.15 2.886 0.4448 1.85
    trust_survey 4.6 4.767 0.8896 3.3
    trust_therm 65 67 22.24 39
    atma 3.435 3.329 0.3868 2.739
    feel_therm 65 68.67 22.24 45
    authen 5.5 5.25 1.112 4
    race_ess 3.75 3.917 0.3706 1.75
    pol_or 4 3.444 1.483 1
      max range skew kurtosis se
    whitemulti_trust_dif 0.35 0.975 -0.7574 -0.542 0.1011
    White 3.6 1.65 0.1395 -1.019 0.1712
    Multiracial 3.5 1.9 -0.694 -0.7509 0.2016
    Black 3.675 1.825 -0.568 -1.339 0.2177
    trust_survey 6.4 3.1 0.3205 -1.488 0.3594
    trust_therm 100 61 0.2009 -1.473 6.64
    atma 3.826 1.087 -0.201 -1.439 0.118
    feel_therm 100 55 0.2088 -1.796 6.733
    authen 7 3 0.173 -1.375 0.3359
    race_ess 5.5 3.75 -0.3997 -0.4772 0.359
    pol_or 7 6 0.3042 -0.6039 0.6035

Non-preregistered Analyses

Examining the Attitudes Towards Multiracial Adults scale by subscale

It appears that it is the self-esteem subscale that is driving any marginal by-condition differens as this is the only significant t-test between conditions

Self-esteem subscale

Attitudes towards multiracial adults by condition SE (continued below)
Test statistic df P value Alternative hypothesis
2.345 350.5 0.01961 * two.sided
mean in group stable mean in group fluid
3.402 3.238

Multiracial Heritage subscale

Attitudes towards multiracial adults by condition MH (continued below)
Test statistic df P value Alternative hypothesis
0.7131 354.6 0.4763 two.sided
mean in group stable mean in group fluid
3.506 3.461

Psychological Adjustment subscale

Attitudes towards multiracial adults by condition PA (continued below)
Test statistic df P value Alternative hypothesis
1.499 355.9 0.1346 two.sided
mean in group stable mean in group fluid
3.702 3.613

Multiracial Identity subscale

Attitudes towards multiracial adults by condition MI (continued below)
Test statistic df P value Alternative hypothesis
1.365 353.2 0.1732 two.sided
mean in group stable mean in group fluid
3.798 3.702

Exploring moderation in the GEE model

Essentialism

Strangely, when essentialism is included in the GEE model, all of the effects disappear, including the effect of race essentialism.

##                    (Intercept)                     condition1 
##                      3.4502463                     -1.0550662 
##                     race_multi                       race_ess 
##                     -0.5925969                     -0.1059355 
##                     race_black          condition1:race_multi 
##                     -0.3081812                     -0.2701861 
##            condition1:race_ess            race_multi:race_ess 
##                      0.2142430                      0.1506636 
##          condition1:race_black            race_ess:race_black 
##                      0.4187384                      0.1231109 
## condition1:race_multi:race_ess condition1:race_ess:race_black 
##                      0.0598983                     -0.0918636
Estimates for the speeded trust model with essentialism
term estimate std.error statistic p.value NA
(Intercept) 3.4502 0.1795 19.2257 0.5704 6.0484
condition1 -1.0551 0.1795 -5.8791 0.5704 -1.8496
race_multi -0.5926 0.2538 -2.3349 0.4355 -1.3608
race_ess -0.1059 0.0395 -2.6836 0.1288 -0.8223
race_black -0.3082 0.2538 -1.2143 0.8425 -0.3658
condition1:race_multi -0.2702 0.2538 -1.0646 0.4355 -0.6204
condition1:race_ess 0.2142 0.0395 5.4273 0.1288 1.6631
race_multi:race_ess 0.1507 0.0558 2.6988 0.1015 1.4840
condition1:race_black 0.4187 0.2538 1.6499 0.8425 0.4970
race_ess:race_black 0.1231 0.0558 2.2052 0.1975 0.6234
condition1:race_multi:race_ess 0.0599 0.0558 1.0729 0.1015 0.5900
condition1:race_ess:race_black -0.0919 0.0558 -1.6455 0.1975 -0.4651

Political Orientation

Two-way interaction between multiracial dummy code and political orientation qualified by three way interaction between the multiracial dummy code, political orientation, and condition. It appears that in the fluid condition for non-white faces, the link between coversative political orientation and trust is strongest. There might be a better way to look at a three-way interaction like this.

##                  (Intercept)                   condition1 
##                   2.49375921                  -0.24014079 
##                   race_multi                       pol_or 
##                   0.00737895                   0.13578704 
##                   race_black        condition1:race_multi 
##                   0.50557632                  -0.15672105 
##            condition1:pol_or            race_multi:pol_or 
##                   0.04340704                   0.00490605 
##        condition1:race_black            pol_or:race_black 
##                  -0.12202368                  -0.08412882 
## condition1:race_multi:pol_or condition1:pol_or:race_black 
##                   0.02568605                   0.03495118
Estimates for the speeded trust model with essentialism
term estimate std.error statistic p.value NA
(Intercept) 2.4938 0.0621 40.1565 0.2408 10.3547
condition1 -0.2401 0.0621 -3.8669 0.2408 -0.9971
race_multi 0.0074 0.0878 0.0840 0.1900 0.0388
pol_or 0.1358 0.0182 7.4640 0.0574 2.3674
race_black 0.5056 0.0878 5.7567 0.2739 1.8458
condition1:race_multi -0.1567 0.0878 -1.7845 0.1900 -0.8249
condition1:pol_or 0.0434 0.0182 2.3860 0.0574 0.7568
race_multi:pol_or 0.0049 0.0257 0.1907 0.0478 0.1027
condition1:race_black -0.1220 0.0878 -1.3894 0.2739 -0.4455
pol_or:race_black -0.0841 0.0257 -3.2700 0.0606 -1.3879
condition1:race_multi:pol_or 0.0257 0.0257 0.9984 0.0478 0.5379
condition1:pol_or:race_black 0.0350 0.0257 1.3585 0.0606 0.5766 
##  condition race_multi pol_or.trend     SE  df asymp.LCL asymp.UCL
##  stable             0       0.0328 0.0610 Inf  -0.08678     0.152
##  fluid              0       0.1546 0.0779 Inf   0.00197     0.307
##  stable             1       0.0121 0.0692 Inf  -0.12363     0.148
##  fluid              1       0.1852 0.1002 Inf  -0.01115     0.382
## 
## Results are averaged over the levels of: race_black 
## Covariance estimate used: robust.variance 
## Confidence level used: 0.95

Examining implicit attitudes as a mediator for explicit attitudes

It appears that the white/multiracial trust difference score independently predicts both the trust thermometer and the feeling thermometer ### Trust Survey

## lavaan 0.6.15 ended normally after 21 iterations
## 
##   Estimator                                         ML
##   Optimization method                           NLMINB
##   Number of model parameters                         7
## 
##                                                   Used       Total
##   Number of observations                           351         358
## 
## Model Test User Model:
##                                                       
##   Test statistic                                 0.000
##   Degrees of freedom                                 0
## 
## Model Test Baseline Model:
## 
##   Test statistic                                 4.247
##   Degrees of freedom                                 3
##   P-value                                        0.236
## 
## User Model versus Baseline Model:
## 
##   Comparative Fit Index (CFI)                    1.000
##   Tucker-Lewis Index (TLI)                       1.000
## 
## Loglikelihood and Information Criteria:
## 
##   Loglikelihood user model (H0)               -688.370
##   Loglikelihood unrestricted model (H1)       -688.370
##                                                       
##   Akaike (AIC)                                1390.740
##   Bayesian (BIC)                              1417.766
##   Sample-size adjusted Bayesian (SABIC)       1395.559
## 
## Root Mean Square Error of Approximation:
## 
##   RMSEA                                          0.000
##   90 Percent confidence interval - lower         0.000
##   90 Percent confidence interval - upper         0.000
##   P-value H_0: RMSEA <= 0.050                       NA
##   P-value H_0: RMSEA >= 0.080                       NA
## 
## Standardized Root Mean Square Residual:
## 
##   SRMR                                           0.000
## 
## Parameter Estimates:
## 
##   Standard errors                            Bootstrap
##   Number of requested bootstrap draws              500
##   Number of successful bootstrap draws             500
## 
## Regressions:
##                          Estimate  Std.Err  z-value  P(>|z|) ci.lower ci.upper
##   trust_survey ~                                                              
##     condition  (c)         -0.016    0.087   -0.183    0.855   -0.199    0.146
##   whitemulti_trust_dif ~                                                      
##     condition  (a)          0.034    0.054    0.633    0.527   -0.073    0.145
##   trust_survey ~                                                              
##     whtmlt_tr_ (b)         -0.169    0.114   -1.476    0.140   -0.400    0.052
##    Std.lv  Std.all
##                   
##    -0.016   -0.010
##                   
##     0.034    0.034
##                   
##    -0.169   -0.104
## 
## Intercepts:
##                    Estimate  Std.Err  z-value  P(>|z|) ci.lower ci.upper
##    .trust_survey      5.042    0.140   36.016    0.000    4.758    5.345
##    .whtmlt_trst_df    0.043    0.089    0.483    0.629   -0.129    0.220
##    Std.lv  Std.all
##     5.042    6.105
##     0.043    0.085
## 
## Variances:
##                    Estimate  Std.Err  z-value  P(>|z|) ci.lower ci.upper
##    .trust_survey      0.675    0.058   11.604    0.000    0.563    0.791
##    .whtmlt_trst_df    0.257    0.029    8.867    0.000    0.200    0.316
##    Std.lv  Std.all
##     0.675    0.989
##     0.257    0.999
## 
## R-Square:
##                    Estimate
##     trust_survey      0.011
##     whtmlt_trst_df    0.001
## 
## Defined Parameters:
##                    Estimate  Std.Err  z-value  P(>|z|) ci.lower ci.upper
##     ab               -0.006    0.010   -0.568    0.570   -0.027    0.016
##     total            -0.022    0.087   -0.250    0.802   -0.208    0.136
##    Std.lv  Std.all
##    -0.006   -0.003
##    -0.022   -0.013
## [1] lhs      op       rhs      mi       epc      sepc.lv  sepc.all sepc.nox
## <0 rows> (or 0-length row.names)

Trust Thermometer

## lavaan 0.6.15 ended normally after 30 iterations
## 
##   Estimator                                         ML
##   Optimization method                           NLMINB
##   Number of model parameters                         7
## 
##                                                   Used       Total
##   Number of observations                           351         358
## 
## Model Test User Model:
##                                                       
##   Test statistic                                 0.000
##   Degrees of freedom                                 0
## 
## Model Test Baseline Model:
## 
##   Test statistic                                13.759
##   Degrees of freedom                                 3
##   P-value                                        0.003
## 
## User Model versus Baseline Model:
## 
##   Comparative Fit Index (CFI)                    1.000
##   Tucker-Lewis Index (TLI)                       1.000
## 
## Loglikelihood and Information Criteria:
## 
##   Loglikelihood user model (H0)              -1785.314
##   Loglikelihood unrestricted model (H1)      -1785.314
##                                                       
##   Akaike (AIC)                                3584.629
##   Bayesian (BIC)                              3611.654
##   Sample-size adjusted Bayesian (SABIC)       3589.448
## 
## Root Mean Square Error of Approximation:
## 
##   RMSEA                                          0.000
##   90 Percent confidence interval - lower         0.000
##   90 Percent confidence interval - upper         0.000
##   P-value H_0: RMSEA <= 0.050                       NA
##   P-value H_0: RMSEA >= 0.080                       NA
## 
## Standardized Root Mean Square Residual:
## 
##   SRMR                                           0.000
## 
## Parameter Estimates:
## 
##   Standard errors                            Bootstrap
##   Number of requested bootstrap draws              500
##   Number of successful bootstrap draws             500
## 
## Regressions:
##                          Estimate  Std.Err  z-value  P(>|z|) ci.lower ci.upper
##   trust_therm ~                                                               
##     condition  (c)         -0.954    1.969   -0.484    0.628   -4.853    2.975
##   whitemulti_trust_dif ~                                                      
##     condition  (a)          0.034    0.052    0.654    0.513   -0.076    0.138
##   trust_therm ~                                                               
##     whtmlt_tr_ (b)         -7.174    2.228   -3.220    0.001  -11.549   -2.617
##    Std.lv  Std.all
##                   
##    -0.954   -0.025
##                   
##     0.034    0.034
##                   
##    -7.174   -0.191
## 
## Intercepts:
##                    Estimate  Std.Err  z-value  P(>|z|) ci.lower ci.upper
##    .trust_therm      74.140    3.126   23.720    0.000   68.299   80.516
##    .whtmlt_trst_df    0.043    0.086    0.499    0.618   -0.140    0.224
##    Std.lv  Std.all
##    74.140    3.890
##     0.043    0.085
## 
## Variances:
##                    Estimate  Std.Err  z-value  P(>|z|) ci.lower ci.upper
##    .trust_therm     349.651   20.224   17.289    0.000  309.668  389.062
##    .whtmlt_trst_df    0.257    0.030    8.507    0.000    0.201    0.321
##    Std.lv  Std.all
##   349.651    0.963
##     0.257    0.999
## 
## R-Square:
##                    Estimate
##     trust_therm       0.037
##     whtmlt_trst_df    0.001
## 
## Defined Parameters:
##                    Estimate  Std.Err  z-value  P(>|z|) ci.lower ci.upper
##     ab               -0.244    0.368   -0.664    0.507   -0.938    0.556
##     total            -1.198    1.979   -0.605    0.545   -5.359    2.615
##    Std.lv  Std.all
##    -0.244   -0.006
##    -1.198   -0.031
## [1] lhs      op       rhs      mi       epc      sepc.lv  sepc.all sepc.nox
## <0 rows> (or 0-length row.names)

Attitudes towards multiracial adults

## lavaan 0.6.15 ended normally after 16 iterations
## 
##   Estimator                                         ML
##   Optimization method                           NLMINB
##   Number of model parameters                         7
## 
##                                                   Used       Total
##   Number of observations                           351         358
## 
## Model Test User Model:
##                                                       
##   Test statistic                                 0.000
##   Degrees of freedom                                 0
## 
## Model Test Baseline Model:
## 
##   Test statistic                                 4.338
##   Degrees of freedom                                 3
##   P-value                                        0.227
## 
## User Model versus Baseline Model:
## 
##   Comparative Fit Index (CFI)                    1.000
##   Tucker-Lewis Index (TLI)                       1.000
## 
## Loglikelihood and Information Criteria:
## 
##   Loglikelihood user model (H0)               -513.462
##   Loglikelihood unrestricted model (H1)       -513.462
##                                                       
##   Akaike (AIC)                                1040.923
##   Bayesian (BIC)                              1067.949
##   Sample-size adjusted Bayesian (SABIC)       1045.742
## 
## Root Mean Square Error of Approximation:
## 
##   RMSEA                                          0.000
##   90 Percent confidence interval - lower         0.000
##   90 Percent confidence interval - upper         0.000
##   P-value H_0: RMSEA <= 0.050                       NA
##   P-value H_0: RMSEA >= 0.080                       NA
## 
## Standardized Root Mean Square Residual:
## 
##   SRMR                                           0.000
## 
## Parameter Estimates:
## 
##   Standard errors                            Bootstrap
##   Number of requested bootstrap draws              500
##   Number of successful bootstrap draws             500
## 
## Regressions:
##                          Estimate  Std.Err  z-value  P(>|z|) ci.lower ci.upper
##   atma ~                                                                      
##     condition  (c)         -0.104    0.055   -1.906    0.057   -0.213   -0.001
##   whitemulti_trust_dif ~                                                      
##     condition  (a)          0.034    0.053    0.637    0.524   -0.070    0.142
##   atma ~                                                                      
##     whtmlt_tr_ (b)         -0.015    0.064   -0.237    0.812   -0.135    0.114
##    Std.lv  Std.all
##                   
##    -0.104   -0.104
##                   
##     0.034    0.034
##                   
##    -0.015   -0.015
## 
## Intercepts:
##                    Estimate  Std.Err  z-value  P(>|z|) ci.lower ci.upper
##    .atma              3.709    0.086   42.995    0.000    3.546    3.878
##    .whtmlt_trst_df    0.043    0.089    0.481    0.630   -0.130    0.217
##    Std.lv  Std.all
##     3.709    7.390
##     0.043    0.085
## 
## Variances:
##                    Estimate  Std.Err  z-value  P(>|z|) ci.lower ci.upper
##    .atma              0.249    0.019   13.130    0.000    0.209    0.282
##    .whtmlt_trst_df    0.257    0.029    8.754    0.000    0.206    0.314
##    Std.lv  Std.all
##     0.249    0.989
##     0.257    0.999
## 
## R-Square:
##                    Estimate
##     atma              0.011
##     whtmlt_trst_df    0.001
## 
## Defined Parameters:
##                    Estimate  Std.Err  z-value  P(>|z|) ci.lower ci.upper
##     ab               -0.001    0.004   -0.125    0.900   -0.010    0.009
##     total            -0.105    0.055   -1.916    0.055   -0.210    0.001
##    Std.lv  Std.all
##    -0.001   -0.001
##    -0.105   -0.105
## [1] lhs      op       rhs      mi       epc      sepc.lv  sepc.all sepc.nox
## <0 rows> (or 0-length row.names)

Feeling thermometer

## lavaan 0.6.15 ended normally after 30 iterations
## 
##   Estimator                                         ML
##   Optimization method                           NLMINB
##   Number of model parameters                         7
## 
##                                                   Used       Total
##   Number of observations                           351         358
## 
## Model Test User Model:
##                                                       
##   Test statistic                                 0.000
##   Degrees of freedom                                 0
## 
## Model Test Baseline Model:
## 
##   Test statistic                                15.177
##   Degrees of freedom                                 3
##   P-value                                        0.002
## 
## User Model versus Baseline Model:
## 
##   Comparative Fit Index (CFI)                    1.000
##   Tucker-Lewis Index (TLI)                       1.000
## 
## Loglikelihood and Information Criteria:
## 
##   Loglikelihood user model (H0)              -1779.466
##   Loglikelihood unrestricted model (H1)      -1779.466
##                                                       
##   Akaike (AIC)                                3572.932
##   Bayesian (BIC)                              3599.958
##   Sample-size adjusted Bayesian (SABIC)       3577.751
## 
## Root Mean Square Error of Approximation:
## 
##   RMSEA                                          0.000
##   90 Percent confidence interval - lower         0.000
##   90 Percent confidence interval - upper         0.000
##   P-value H_0: RMSEA <= 0.050                       NA
##   P-value H_0: RMSEA >= 0.080                       NA
## 
## Standardized Root Mean Square Residual:
## 
##   SRMR                                           0.000
## 
## Parameter Estimates:
## 
##   Standard errors                            Bootstrap
##   Number of requested bootstrap draws              500
##   Number of successful bootstrap draws             500
## 
## Regressions:
##                          Estimate  Std.Err  z-value  P(>|z|) ci.lower ci.upper
##   feel_therm ~                                                                
##     condition  (c)         -1.385    1.998   -0.693    0.488   -5.515    2.473
##   whitemulti_trust_dif ~                                                      
##     condition  (a)          0.034    0.051    0.666    0.505   -0.064    0.137
##   feel_therm ~                                                                
##     whtmlt_tr_ (b)         -7.353    2.212   -3.323    0.001  -11.417   -2.975
##    Std.lv  Std.all
##                   
##    -1.385   -0.037
##                   
##     0.034    0.034
##                   
##    -7.353   -0.198
## 
## Intercepts:
##                    Estimate  Std.Err  z-value  P(>|z|) ci.lower ci.upper
##    .feel_therm       77.911    3.152   24.717    0.000   71.332   84.028
##    .whtmlt_trst_df    0.043    0.087    0.494    0.621   -0.122    0.217
##    Std.lv  Std.all
##    77.911    4.148
##     0.043    0.085
## 
## Variances:
##                    Estimate  Std.Err  z-value  P(>|z|) ci.lower ci.upper
##    .feel_therm      338.191   20.398   16.580    0.000  296.177  375.194
##    .whtmlt_trst_df    0.257    0.029    8.838    0.000    0.201    0.311
##    Std.lv  Std.all
##   338.191    0.959
##     0.257    0.999
## 
## R-Square:
##                    Estimate
##     feel_therm        0.041
##     whtmlt_trst_df    0.001
## 
## Defined Parameters:
##                    Estimate  Std.Err  z-value  P(>|z|) ci.lower ci.upper
##     ab               -0.250    0.375   -0.668    0.504   -0.963    0.545
##     total            -1.635    2.015   -0.812    0.417   -5.721    2.361
##    Std.lv  Std.all
##    -0.250   -0.007
##    -1.635   -0.044
## [1] lhs      op       rhs      mi       epc      sepc.lv  sepc.all sepc.nox
## <0 rows> (or 0-length row.names)

Authenticity

## lavaan 0.6.15 ended normally after 19 iterations
## 
##   Estimator                                         ML
##   Optimization method                           NLMINB
##   Number of model parameters                         7
## 
##                                                   Used       Total
##   Number of observations                           351         358
## 
## Model Test User Model:
##                                                       
##   Test statistic                                 0.000
##   Degrees of freedom                                 0
## 
## Model Test Baseline Model:
## 
##   Test statistic                                 5.728
##   Degrees of freedom                                 3
##   P-value                                        0.126
## 
## User Model versus Baseline Model:
## 
##   Comparative Fit Index (CFI)                    1.000
##   Tucker-Lewis Index (TLI)                       1.000
## 
## Loglikelihood and Information Criteria:
## 
##   Loglikelihood user model (H0)               -732.563
##   Loglikelihood unrestricted model (H1)       -732.563
##                                                       
##   Akaike (AIC)                                1479.125
##   Bayesian (BIC)                              1506.151
##   Sample-size adjusted Bayesian (SABIC)       1483.944
## 
## Root Mean Square Error of Approximation:
## 
##   RMSEA                                          0.000
##   90 Percent confidence interval - lower         0.000
##   90 Percent confidence interval - upper         0.000
##   P-value H_0: RMSEA <= 0.050                       NA
##   P-value H_0: RMSEA >= 0.080                       NA
## 
## Standardized Root Mean Square Residual:
## 
##   SRMR                                           0.000
## 
## Parameter Estimates:
## 
##   Standard errors                            Bootstrap
##   Number of requested bootstrap draws              500
##   Number of successful bootstrap draws             500
## 
## Regressions:
##                          Estimate  Std.Err  z-value  P(>|z|) ci.lower ci.upper
##   authen ~                                                                    
##     condition  (c)         -0.176    0.096   -1.846    0.065   -0.361    0.011
##   whitemulti_trust_dif ~                                                      
##     condition  (a)          0.034    0.051    0.668    0.504   -0.063    0.141
##   authen ~                                                                    
##     whtmlt_tr_ (b)         -0.141    0.100   -1.402    0.161   -0.328    0.063
##    Std.lv  Std.all
##                   
##    -0.176   -0.094
##                   
##     0.034    0.034
##                   
##    -0.141   -0.076
## 
## Intercepts:
##                    Estimate  Std.Err  z-value  P(>|z|) ci.lower ci.upper
##    .authen            5.785    0.152   38.042    0.000    5.457    6.112
##    .whtmlt_trst_df    0.043    0.084    0.511    0.609   -0.129    0.208
##    Std.lv  Std.all
##     5.785    6.162
##     0.043    0.085
## 
## Variances:
##                    Estimate  Std.Err  z-value  P(>|z|) ci.lower ci.upper
##    .authen            0.868    0.068   12.671    0.000    0.725    1.013
##    .whtmlt_trst_df    0.257    0.030    8.649    0.000    0.198    0.314
##    Std.lv  Std.all
##     0.868    0.985
##     0.257    0.999
## 
## R-Square:
##                    Estimate
##     authen            0.015
##     whtmlt_trst_df    0.001
## 
## Defined Parameters:
##                    Estimate  Std.Err  z-value  P(>|z|) ci.lower ci.upper
##     ab               -0.005    0.009   -0.520    0.603   -0.027    0.013
##     total            -0.181    0.096   -1.896    0.058   -0.376    0.007
##    Std.lv  Std.all
##    -0.005   -0.003
##    -0.181   -0.096
## [1] lhs      op       rhs      mi       epc      sepc.lv  sepc.all sepc.nox
## <0 rows> (or 0-length row.names)

Analyses including only White participants

Note: All analyses NOT presented here did not have notable differences from the main analyses

Exploratory Analyses.

Race Essentialism as moderator

Attitudes towards multiracial adults

Findings:

The difference between this analysis and the main analysis is that there is no significant main effect of race essentialism in this analysis while there was in the analysis with the full sample.

Estimates for the attitudes towards multiracial adults model
term estimate std.error statistic p.value
(Intercept) 3.7703 0.0972 38.8056 0.0000
condition1 -0.0801 0.0972 -0.8240 0.4107
race_ess -0.0464 0.0243 -1.9045 0.0579
condition1:race_ess 0.0078 0.0243 0.3208 0.7486
Summary for the attitudes towards multiracial adults model
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.0232 0.0122 0.5128 2.1138 0.0988 3 -201.536 413.072 431.083 70.2171 267 271

Political orientation as a moderator

Speeded trust task by condition + political orientation as moderator

Findings:

Interesting, in contrast with the analyses with the main sample, here we see a main effect of condition in our predicted direction. Further, there is an interaction between condition and political orientation such that the relationship between conservative political orientation and the bias against multiracial faces is stronger in the stable condition.

Estimates for the speeded trust model
term estimate std.error statistic p.value
(Intercept) -0.1904 0.0605 -3.1483 0.0018
condition1 0.1290 0.0605 2.1340 0.0338
pol_or 0.1011 0.0177 5.7025 0.0000
condition1:pol_or -0.0353 0.0177 -1.9906 0.0476
Summary for the speeded trust model
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.1224 0.1123 0.489 12.0916 0 3 -183.724 377.448 395.327 62.1729 260 264 
##  condition pol_or.trend     SE  df lower.CL upper.CL
##  stable          0.1364 0.0253 260   0.0866    0.186
##  fluid           0.0658 0.0249 260   0.0168    0.115
## 
## Confidence level used: 0.95
Attitudes towards multiracial adults

Findings:

Here, of note there is a marginal effect of condition in the expected direction (there was not in analyses with main sample).

White, Black, Latine, Asian and Other
Estimates for the attitudes towards multiracial adults model
term estimate std.error statistic p.value
(Intercept) 3.6555 0.0632 57.8778 0.0000
condition1 -0.1209 0.0632 -1.9140 0.0567
pol_or -0.0189 0.0183 -1.0328 0.3026
condition1:pol_or 0.0226 0.0183 1.2295 0.2200
Summary for the attitudes towards multiracial adults model
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.0191 0.008 0.5139 1.729 0.1614 3 -202.109 414.219 432.229 70.5149 267 271
Authenticity

Findings:

Here, of note there is also a marginal effect of condition in the expected direction (there was not in analyses with main sample).

White, Black, Latine, Asian and Other
Estimates for the authenticity model
term estimate std.error statistic p.value
(Intercept) 5.9257 0.1108 53.4700 0.0000
condition1 -0.1934 0.1108 -1.7456 0.0820
pol_or -0.1118 0.0322 -3.4735 0.0006
condition1:pol_or 0.0365 0.0322 1.1344 0.2576
Summary for the authenticity model
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.0526 0.042 0.9017 4.9458 0.0023 3 -354.489 718.978 736.988 217.108 267 271

Analyses including only Black participants

GEE Model for Speeded Trust Task

Findings:

Interestingly, the condition main effect is not significant here, though it was in the main analyses. There are also no main effects of the dummy coded race variables nor interactions, though there were with the main sample.

##           (Intercept)            condition1            race_multi 
##             2.9062135            -0.0951023             0.0248904 
##            race_black condition1:race_multi condition1:race_black 
##             0.2567251            -0.0790570            -0.0261696
Estimates for the speeded trust model
term estimate std.error statistic p.value NA
(Intercept) 2.9062 0.0287 101.3055 0.0992 29.2892
condition1 -0.0951 0.0287 -3.3151 0.0992 -0.9585
race_multi 0.0249 0.0406 0.6135 0.0694 0.3588
race_black 0.2567 0.0406 6.3279 0.1161 2.2121
condition1:race_multi -0.0791 0.0406 -1.9486 0.0694 -1.1397
condition1:race_black -0.0262 0.0406 -0.6450 0.1161 -0.2255
Emmeans for condition * race_multi
condition race_multi emmean SE df asymp.LCL asymp.UCL
stable 0 3.1428 0.1036 Inf 2.9396 3.3459
fluid 0 2.9264 0.1255 Inf 2.6804 3.1724
stable 1 3.2467 0.1310 Inf 2.9899 3.5035
fluid 1 2.8722 0.1532 Inf 2.5719 3.1725
Post hoc tests for race_multi * condition
contrast condition estimate SE df z.ratio p.value
race_multi1 - race_multi0 stable 0.1039 0.0951 Inf 1.0927 0.2745
race_multi1 - race_multi0 fluid -0.0542 0.1010 Inf -0.5364 0.5917
Emmeans for condition * race_black
condition race_black emmean SE df asymp.LCL asymp.UCL
stable 0 3.0533 0.1171 Inf 2.8238 3.2828
fluid 0 2.7840 0.1288 Inf 2.5315 3.0365
stable 1 3.3362 0.1811 Inf 2.9813 3.6910
fluid 1 3.0146 0.1462 Inf 2.7281 3.3011
Post hoc tests for condition * condition
contrast race_black estimate SE df z.ratio p.value
fluid - stable 0 -0.2693 0.1741 Inf -1.5467 0.1219
fluid - stable 1 -0.3216 0.2327 Inf -1.3820 0.1670

Exploratory Analyses.

Race Essentialism

T-test

Findings:

Interesting, there is actually quite a large difference in race essentialism here. Black participants in the fluid condition report less race essentialism that Black participants in the stable condition.

Race essentialism by condition (continued below)
Test statistic df P value Alternative hypothesis
3.001 28.43 0.005547 * * two.sided
mean in group stable mean in group fluid
4.706 3.84
Race essentialism as moderator

Of note, there is no main effect of race essentialism for Black participants for any of the following DVs, though there were for the full sample.

Speeded trust task
Estimates for the speeded trust model
term estimate std.error statistic p.value
(Intercept) 0.5782 0.4431 1.3048 0.2016
condition1 0.2846 0.4431 0.6423 0.5254
race_ess -0.1491 0.0973 -1.5333 0.1353
condition1:race_ess -0.0614 0.0973 -0.6317 0.5322
Summary for the speeded trust model
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.1672 0.0866 0.4242 2.074 0.1239 3 -17.523 45.046 52.8228 5.5778 31 35
Trust survey
Estimates for the trust survey model
term estimate std.error statistic p.value
(Intercept) 5.1741 0.9179 5.6371 0.0000
condition1 0.7361 0.9179 0.8020 0.4287
race_ess -0.1357 0.2015 -0.6737 0.5055
condition1:race_ess -0.2402 0.2015 -1.1921 0.2423
Summary for the trust survey model
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.1587 0.0773 0.8786 1.9497 0.1421 3 -43.0105 96.0209 103.798 23.9319 31 35
Trust Thermometer
Estimates for the trust thermometer model
term estimate std.error statistic p.value
(Intercept) 46.3571 20.4028 2.2721 0.0302
condition1 32.7295 20.4028 1.6042 0.1188
race_ess 4.7640 4.4782 1.0638 0.2956
condition1:race_ess -8.2151 4.4782 -1.8345 0.0762
Summary for the trust thermometer model
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.144 0.0612 19.5308 1.7384 0.1796 3 -151.559 313.117 320.894 11825 31 35
Attitudes towards multiracial adults
Estimates for the attitudes towards multiracial adults model
term estimate std.error statistic p.value
(Intercept) 3.8289 0.4452 8.6011 0.0000
condition1 0.0093 0.4452 0.0209 0.9835
race_ess -0.1131 0.0977 -1.1571 0.2561
condition1:race_ess -0.0291 0.0977 -0.2979 0.7678
Summary for the attitudes towards multiracial adults model
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.0914 0.0035 0.4261 1.0397 0.3887 3 -17.6844 45.3689 53.1456 5.6294 31 35
Feeling thermometer
Estimates for the feeling thermometer model
term estimate std.error statistic p.value
(Intercept) 63.8742 21.4541 2.9773 0.0056
condition1 23.3033 21.4541 1.0862 0.2858
race_ess 1.3559 4.7089 0.2879 0.7753
condition1:race_ess -6.0603 4.7089 -1.2870 0.2076
Summary for the feeling thermometer model
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.0783 -0.0109 20.5371 0.8777 0.4633 3 -153.317 316.635 324.411 13075 31 35
Authenticity
Estimates for the authenticity model
term estimate std.error statistic p.value
(Intercept) 5.6394 1.0844 5.2005 0.0000
condition1 0.5273 1.0844 0.4863 0.6302
race_ess -0.1807 0.2380 -0.7594 0.4534
condition1:race_ess -0.1882 0.2380 -0.7906 0.4352
Summary for the authenticity model
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.1035 0.0168 1.038 1.1933 0.3285 3 -48.8458 107.692 115.468 33.4034 31 35

Political orientation as moderator

Of note, there is no main effect of political orientation for Black participants for any of the following DVs, though there were for the full sample.

Speeded trust task
Estimates for the speeded trust model
term estimate std.error statistic p.value
(Intercept) -0.0079 0.1640 -0.0480 0.9620
condition1 0.1572 0.1640 0.9588 0.3451
pol_or -0.0070 0.0484 -0.1443 0.8862
condition1:pol_or -0.0236 0.0484 -0.4879 0.6290
Summary for the speeded trust model
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.0465 -0.0457 0.4539 0.5042 0.6822 3 -19.8903 49.7805 57.5573 6.3857 31 35
Trust Thermometer
White, Black, Latine, Asian and Other
Estimates for the trust thermometer model
term estimate std.error statistic p.value
(Intercept) 73.9851 7.3570 10.0564 0.0000
condition1 -6.3404 7.3570 -0.8618 0.3954
pol_or -1.3074 2.1709 -0.6023 0.5514
condition1:pol_or 0.7252 2.1709 0.3341 0.7406
Summary for the trust thermometer model
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.0695 -0.0205 20.3629 0.7719 0.5185 3 -153.019 316.038 323.815 12854 31 35
Feeling thermometer
White, Black, Latine, Asian and Other
Estimates for the feeling thermometer model
term estimate std.error statistic p.value
(Intercept) 81.1048 7.3926 10.9710 0.0000
condition1 -2.3943 7.3926 -0.3239 0.7482
pol_or -2.9947 2.1814 -1.3728 0.1797
condition1:pol_or -0.0908 2.1814 -0.0416 0.9671
Summary for the feeling thermometer model
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.0851 -0.0035 20.4614 0.9609 0.4235 3 -153.188 316.376 324.152 12978.7 31 35
Authenticity
White, Black, Latine, Asian and Other
Estimates for the authenticity model
term estimate std.error statistic p.value
(Intercept) 5.3939 0.3780 14.2706 0.0000
condition1 -0.0452 0.3780 -0.1196 0.9055
pol_or -0.1490 0.1115 -1.3363 0.1912
condition1:pol_or -0.0434 0.1115 -0.3891 0.6999
Summary for the authenticity model
r.squared adj.r.squared sigma statistic p.value df logLik AIC BIC deviance df.residual nobs
0.0894 0.0013 1.0462 1.0151 0.3993 3 -49.1183 108.237 116.013 33.9278 31 35