2. Models
A. partyLikely ~ intro_cc + policy + DvR + BvE + DRvBE + polXDR + polXBE + polXDRvBE + (1 | country)
Likelihood of attending party regressed on contrast coded introspection condition accounting for random intercepts for country (the model would not converge with random slopes)
intro_cc_.5: introspection = +.5; control = -.5
Across countries, on average as you move from control to introspection condition, there is a .26 decrease in the likelihood of attending the party, \(\beta_1\) = -.26, t(6282) = -5.71, p < .001.
summary(m1 <- lmer(partyLikely ~ intro_cc_.5 + policy + DvR + BvE + DRvBE + polXDR + polXBE + polXDRvBE + ( 1 | country_factor), data = SG))
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: partyLikely ~ intro_cc_.5 + policy + DvR + BvE + DRvBE + polXDR +
## polXBE + polXDRvBE + (1 | country_factor)
## Data: SG
##
## REML criterion at convergence: 25616.2
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.4594 -0.7860 -0.3535 0.7061 2.7621
##
## Random effects:
## Groups Name Variance Std.Dev.
## country_factor (Intercept) 0.1965 0.4433
## Residual 3.3883 1.8407
## Number of obs: 6297, groups: country_factor, 7
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 2.705e+00 1.693e-01 6.004e+00 15.978 3.79e-06 ***
## intro_cc_.5 -2.649e-01 4.643e-02 6.282e+03 -5.705 1.22e-08 ***
## policy 1.157e-02 2.324e-02 6.283e+03 0.498 0.6187
## DvR 2.625e-02 3.297e-02 6.282e+03 0.796 0.4260
## BvE 3.409e-02 3.270e-02 6.282e+03 1.043 0.2972
## DRvBE 5.622e-02 2.322e-02 6.282e+03 2.421 0.0155 *
## polXDR -2.055e-03 3.297e-02 6.282e+03 -0.062 0.9503
## polXBE 1.984e-02 3.270e-02 6.282e+03 0.607 0.5440
## polXDRvBE -1.940e-02 2.323e-02 6.282e+03 -0.835 0.4036
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) in__.5 policy DvR BvE DRvBE polXDR polXBE
## intro_cc_.5 0.002
## policy 0.001 -0.008
## DvR 0.001 -0.004 0.002
## BvE -0.004 0.007 0.006 0.000
## DRvBE -0.001 0.002 -0.018 -0.006 -0.022
## polXDR 0.000 -0.003 0.006 0.024 0.000 -0.003
## polXBE 0.001 0.015 -0.021 0.000 -0.013 0.007 0.000
## polXDRvBE -0.003 -0.006 -0.010 -0.003 0.007 0.006 -0.005 -0.022
## Warning: `select_vars()` is deprecated as of dplyr 0.8.4.
## Please use `tidyselect::vars_select()` instead.
## This warning is displayed once every 8 hours.
## Call `lifecycle::last_warnings()` to see where this warning was generated.
B. (2 DF test) partyLikely ~ intro_cc + introSpread_c + introExperienceSympt_c + policy + DvR + BvE + DRvBE + polXDR + polXBE + polXDRvBE + (1 | country)
2 DF test: Likelihood of attending party regressed on contrast coded introspection condition, introspection intervention rating for likelihood of spreading Covid-19 and experiencing symptoms of Covid-19, while controlling for contrast coded party over policy condition. Again, this also accounts for random intercepts for country (the model would not converge with random slopes)
Predictors excluded:
introSpread (mean centered): How much should your likelihood of spreading Covid-19 influence your decision to attend the gathering? (1 = not at all; 7 = extremely)
introExpSymp (mean centered): How much should your likelihood of experiencing severe symptoms from Covid-19 influence your decision to attend the gathering? (1 = not at all; 7 = extremely)
Results:
As expected, for our 2 degree of freedom test, there is a significant effect of the introspection question items. As a person rates higher that the spread and experiencing symptoms of Covid-19 should influence their decision to attend the party, they are less likely to attend the party, PRE = .18, F(2,6247) = 693.76, p < .001.
modelCompare(mc, ma)
## SSE (Compact) = 22163.04
## SSE (Augmented) = 18135.07
## Delta R-Squared = 0.1806141
## Partial Eta-Squared (PRE) = 0.1817428
## F(2,6247) = 693.7596, p = 8.153119e-273
C. partyLikely ~ intro_cc + avgIntro_c + policy + DvR + BvE + DRvBE + polXDR + polXBE + polXDRvBE + (1 | country)
Likelihood of attending party regressed on contrast coded introspection condition, and the mean centered average for introspection intervention ratings, while controlling for contrast coded party over policy condition. Again, this also accounts for random intercepts for country (the model would not converge with random slopes)
Predictors excluded
avg_screen (mean centered): the average of introspection items for how much you should consider the likelihood of spreading Covid-19 to others and experiencing symptoms when deciding whether to attend the gathering
Results
Controlling for party over policy condition, and when including random intercepts for country, there is a significant effect for introspection condition even when controlling for mean centered average rating for the introspection items. As you move from control to introspection condition, there is a .21 decrease in the likelihood of attending the party, \(\beta_1\) = -.21, t(6241.15) = -4.90, p < .001.
SG$avg_screen <- (SG$introExpSymp + SG$introSpread)/2
SG$avg_screen_c <- SG$avg_screen - mean(SG$avg_screen, na.rm = T)
summary(m<-lmer(partyLikely ~ intro_cc_.5*avg_screen_c + policy + DvR + BvE + DRvBE + polXDR + polXBE + polXDRvBE + (1 | country), data = SG))
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: partyLikely ~ intro_cc_.5 * avg_screen_c + policy + DvR + BvE +
## DRvBE + polXDR + polXBE + polXDRvBE + (1 | country)
## Data: SG
##
## REML criterion at convergence: 24165.2
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -2.8429 -0.5929 -0.2782 0.5833 3.2420
##
## Random effects:
## Groups Name Variance Std.Dev.
## country (Intercept) 0.1916 0.4377
## Residual 2.7501 1.6583
## Number of obs: 6258, groups: country, 7
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 2.68781 0.16685 6.00769 16.109 3.60e-06 ***
## intro_cc_.5 -0.20556 0.04197 6241.15032 -4.898 9.94e-07 ***
## avg_screen_c -0.48725 0.01286 6246.94048 -37.887 < 2e-16 ***
## policy 0.02333 0.02101 6241.65698 1.110 0.2669
## DvR 0.03841 0.02976 6241.20355 1.291 0.1969
## BvE 0.02831 0.02959 6241.12371 0.957 0.3388
## DRvBE 0.05028 0.02098 6241.19538 2.396 0.0166 *
## polXDR -0.01507 0.02976 6241.18038 -0.506 0.6127
## polXBE 0.01557 0.02960 6241.19586 0.526 0.5988
## polXDRvBE -0.03011 0.02099 6241.29471 -1.434 0.1515
## intro_cc_.5:avg_screen_c -0.04119 0.02517 6241.47504 -1.637 0.1017
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) in__.5 avg_s_ policy DvR BvE DRvBE polXDR polXBE
## intro_cc_.5 0.001
## avg_scren_c 0.002 -0.032
## policy 0.001 -0.007 -0.013
## DvR 0.001 -0.002 -0.009 0.001
## BvE -0.003 0.007 0.002 0.006 0.000
## DRvBE 0.000 -0.001 0.005 -0.020 -0.004 -0.022
## polXDR 0.000 -0.003 0.015 0.004 0.025 0.000 -0.003
## polXBE 0.001 0.015 0.004 -0.021 0.000 -0.015 0.006 0.000
## polXDRvBE -0.003 -0.005 0.018 -0.007 -0.002 0.007 0.005 -0.003 -0.023
## intr__.5:__ -0.003 -0.001 0.030 0.002 -0.012 0.007 -0.004 -0.006 -0.006
## pXDRBE
## intro_cc_.5
## avg_scren_c
## policy
## DvR
## BvE
## DRvBE
## polXDR
## polXBE
## polXDRvBE
## intr__.5:__ -0.009
3. test for moderators
A. education (alone)
Note: I exclude the 85 participants from Brazil that chose #18 for their education.
Education is a marginally significant moderator, \(\beta\) = -.03, t(6057) = -1.73, p = .084.
SG$edu_c <- SG$education - mean(SG$education, na.rm = T)
SG$ideology_c <- SG$ideology - mean(SG$ideology, na.rm = T)
#exclude the weird 18 data points
SG2 <- SG[SG$education != 18,]
summary(m<-lmer(partyLikely ~ intro_cc_.5*edu_c + policy + DvR + BvE + DRvBE + polXDR + polXBE + polXDRvBE + (1 | country), data = SG2))
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: partyLikely ~ intro_cc_.5 * edu_c + policy + DvR + BvE + DRvBE +
## polXDR + polXBE + polXDRvBE + (1 | country)
## Data: SG2
##
## REML criterion at convergence: 24695.6
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.4627 -0.7798 -0.3515 0.6994 2.7666
##
## Random effects:
## Groups Name Variance Std.Dev.
## country (Intercept) 0.1891 0.4348
## Residual 3.3734 1.8367
## Number of obs: 6074, groups: country, 7
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 2.696e+00 1.662e-01 6.012e+00 16.226 3.42e-06 ***
## intro_cc_.5 -2.509e-01 4.718e-02 6.057e+03 -5.319 1.08e-07 ***
## edu_c -1.848e-02 8.810e-03 6.062e+03 -2.098 0.0360 *
## policy 1.748e-02 2.363e-02 6.058e+03 0.740 0.4595
## DvR 2.009e-02 3.350e-02 6.057e+03 0.600 0.5487
## BvE 3.835e-02 3.322e-02 6.057e+03 1.154 0.2484
## DRvBE 5.836e-02 2.360e-02 6.057e+03 2.473 0.0134 *
## polXDR -7.927e-03 3.352e-02 6.057e+03 -0.237 0.8130
## polXBE 2.893e-02 3.323e-02 6.057e+03 0.871 0.3840
## polXDRvBE -2.625e-02 2.360e-02 6.057e+03 -1.112 0.2661
## intro_cc_.5:edu_c -3.015e-02 1.745e-02 6.057e+03 -1.727 0.0841 .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) in__.5 edu_c policy DvR BvE DRvBE polXDR polXBE
## intro_cc_.5 0.001
## edu_c 0.002 -0.017
## policy 0.001 -0.008 -0.017
## DvR 0.001 -0.005 -0.010 0.002
## BvE -0.004 0.007 0.001 0.008 0.000
## DRvBE -0.001 0.000 0.004 -0.020 -0.008 -0.025
## polXDR -0.001 -0.002 0.026 0.008 0.029 0.001 -0.003
## polXBE 0.001 0.012 -0.006 -0.024 0.000 -0.012 0.008 0.000
## polXDRvBE -0.004 -0.006 -0.001 -0.010 -0.002 0.009 0.009 -0.008 -0.025
## intr_c_.5:_ -0.002 0.018 0.057 -0.013 -0.005 0.000 -0.017 -0.005 -0.004
## pXDRBE
## intro_cc_.5
## edu_c
## policy
## DvR
## BvE
## DRvBE
## polXDR
## polXBE
## polXDRvBE
## intr_c_.5:_ -0.015
B. Eduction and average of symbolic ideology
Results
the interaction between education and introspection remains marginally significant, \(\beta\) = -.03, t(6039) = -1.81, p = .070.
There is no significant 2-way interaction between introspection and ideology, p = .677, or 3-way interaction between introspection, ideology, and education, p = .162.
summary(m<-lmer(partyLikely ~ intro_cc_.5*edu_c*ideology_c + policy + DvR + BvE + DRvBE + polXDR + polXBE + polXDRvBE + (1 | country), data = SG2))
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: partyLikely ~ intro_cc_.5 * edu_c * ideology_c + policy + DvR +
## BvE + DRvBE + polXDR + polXBE + polXDRvBE + (1 | country)
## Data: SG2
##
## REML criterion at convergence: 24569.7
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.6709 -0.7785 -0.3605 0.6933 2.8909
##
## Random effects:
## Groups Name Variance Std.Dev.
## country (Intercept) 0.1502 0.3876
## Residual 3.3254 1.8236
## Number of obs: 6059, groups: country, 7
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 2.700e+00 1.485e-01 6.011e+00 18.182 1.75e-06
## intro_cc_.5 -2.569e-01 4.695e-02 6.038e+03 -5.472 4.64e-08
## edu_c -1.473e-02 8.781e-03 6.043e+03 -1.678 0.0934
## ideology_c 1.594e-01 1.646e-02 6.043e+03 9.684 < 2e-16
## policy 1.988e-02 2.349e-02 6.039e+03 0.846 0.3975
## DvR 2.055e-02 3.331e-02 6.038e+03 0.617 0.5374
## BvE 4.073e-02 3.302e-02 6.038e+03 1.233 0.2175
## DRvBE 5.255e-02 2.347e-02 6.038e+03 2.239 0.0252
## polXDR -2.593e-03 3.333e-02 6.038e+03 -0.078 0.9380
## polXBE 3.046e-02 3.303e-02 6.038e+03 0.922 0.3565
## polXDRvBE -3.035e-02 2.347e-02 6.038e+03 -1.293 0.1959
## intro_cc_.5:edu_c -3.156e-02 1.741e-02 6.039e+03 -1.812 0.0700
## intro_cc_.5:ideology_c 1.150e-02 3.206e-02 6.038e+03 0.359 0.7199
## edu_c:ideology_c -2.444e-03 5.869e-03 6.039e+03 -0.416 0.6771
## intro_cc_.5:edu_c:ideology_c -1.642e-02 1.173e-02 6.038e+03 -1.400 0.1615
##
## (Intercept) ***
## intro_cc_.5 ***
## edu_c .
## ideology_c ***
## policy
## DvR
## BvE
## DRvBE *
## polXDR
## polXBE
## polXDRvBE
## intro_cc_.5:edu_c .
## intro_cc_.5:ideology_c
## edu_c:ideology_c
## intro_cc_.5:edu_c:ideology_c
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation matrix not shown by default, as p = 15 > 12.
## Use print(x, correlation=TRUE) or
## vcov(x) if you need it
C. Eduction and party identity
Results
the interaction between education and introspection remains marginally significant, \(\beta\) = -.034, t(5950) = -1.88, p = .060.
There is no significant 2-way interaction between introspection and party identity for participant or 3-way interaction between introspection, participant party, and education.
summary(m<-lmer(partyLikely ~ intro_cc_.5*edu_c*ptIvRD*ptRvD + policy + DvR + BvE + DRvBE + polXDR + polXBE + polXDRvBE + (1 | country), data = SG2))
## fixed-effect model matrix is rank deficient so dropping 4 columns / coefficients
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: partyLikely ~ intro_cc_.5 * edu_c * ptIvRD * ptRvD + policy +
## DvR + BvE + DRvBE + polXDR + polXBE + polXDRvBE + (1 | country)
## Data: SG2
##
## REML criterion at convergence: 24282.8
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.5622 -0.7931 -0.3550 0.6930 2.8341
##
## Random effects:
## Groups Name Variance Std.Dev.
## country (Intercept) 0.1712 0.4138
## Residual 3.3510 1.8306
## Number of obs: 5975, groups: country, 7
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 2.698e+00 1.585e-01 6.030e+00 17.024 2.51e-06 ***
## intro_cc_.5 -2.593e-01 4.866e-02 5.950e+03 -5.328 1.03e-07 ***
## edu_c -2.010e-02 9.108e-03 5.954e+03 -2.207 0.0274 *
## ptIvRD -1.235e-02 3.237e-02 5.956e+03 -0.381 0.7029
## ptRvD -3.670e-01 6.514e-02 5.956e+03 -5.634 1.84e-08 ***
## policy 1.968e-02 2.376e-02 5.951e+03 0.828 0.4075
## DvR 2.232e-02 3.375e-02 5.950e+03 0.661 0.5084
## BvE 4.573e-02 3.333e-02 5.950e+03 1.372 0.1702
## DRvBE 5.275e-02 2.373e-02 5.950e+03 2.222 0.0263 *
## polXDR -1.146e-02 3.377e-02 5.950e+03 -0.339 0.7344
## polXBE 2.314e-02 3.334e-02 5.950e+03 0.694 0.4877
## polXDRvBE -1.568e-02 2.373e-02 5.950e+03 -0.661 0.5088
## intro_cc_.5:edu_c -3.399e-02 1.809e-02 5.950e+03 -1.880 0.0602 .
## intro_cc_.5:ptIvRD -1.271e-02 6.373e-02 5.950e+03 -0.199 0.8420
## edu_c:ptIvRD 9.786e-03 1.190e-02 5.950e+03 0.823 0.4108
## intro_cc_.5:ptRvD -9.867e-02 1.275e-01 5.950e+03 -0.774 0.4389
## edu_c:ptRvD 2.553e-02 2.363e-02 5.951e+03 1.081 0.2800
## intro_cc_.5:edu_c:ptIvRD 2.718e-02 2.378e-02 5.950e+03 1.143 0.2530
## intro_cc_.5:edu_c:ptRvD -8.835e-03 4.718e-02 5.950e+03 -0.187 0.8515
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation matrix not shown by default, as p = 19 > 12.
## Use print(x, correlation=TRUE) or
## vcov(x) if you need it
## fit warnings:
## fixed-effect model matrix is rank deficient so dropping 4 columns / coefficients
## Error: Confidence intervals could not be computed.
## * Reason: "non-conformable arguments"
## * Source: mm %*% vcm
## Scale for 'colour' is already present. Adding another scale for 'colour',
## which will replace the existing scale.
D. ses ladder
SES ladder is not a significant moderator, p = .622.
summary(m<-lmer(partyLikely ~ intro_cc_.5*SESladder + policy + DvR + BvE + DRvBE + polXDR + polXBE + polXDRvBE + (1 | country), data = SG))
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: partyLikely ~ intro_cc_.5 * SESladder + policy + DvR + BvE +
## DRvBE + polXDR + polXBE + polXDRvBE + (1 | country)
## Data: SG
##
## REML criterion at convergence: 25132.1
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.4731 -0.7844 -0.3584 0.7036 2.7876
##
## Random effects:
## Groups Name Variance Std.Dev.
## country (Intercept) 0.1946 0.4412
## Residual 3.3822 1.8391
## Number of obs: 6178, groups: country, 7
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 2.861e+00 1.832e-01 8.380e+00 15.619 1.73e-07 ***
## intro_cc_.5 -3.200e-01 1.473e-01 6.161e+03 -2.172 0.0299 *
## SESladder -2.942e-02 1.304e-02 6.167e+03 -2.257 0.0240 *
## policy 1.444e-02 2.346e-02 6.162e+03 0.616 0.5381
## DvR 2.526e-02 3.329e-02 6.161e+03 0.759 0.4479
## BvE 3.692e-02 3.297e-02 6.161e+03 1.120 0.2629
## DRvBE 5.938e-02 2.343e-02 6.161e+03 2.534 0.0113 *
## polXDR -6.651e-03 3.329e-02 6.161e+03 -0.200 0.8417
## polXBE 2.467e-02 3.298e-02 6.161e+03 0.748 0.4544
## polXDRvBE -2.479e-02 2.343e-02 6.161e+03 -1.058 0.2901
## intro_cc_.5:SESladder 1.253e-02 2.539e-02 6.161e+03 0.493 0.6217
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) in__.5 SESldd policy DvR BvE DRvBE polXDR polXBE
## intro_cc_.5 0.002
## SESladder -0.392 -0.003
## policy -0.004 0.006 0.013
## DvR 0.001 -0.013 0.001 0.004
## BvE -0.004 0.022 0.000 0.009 0.000
## DRvBE 0.007 0.009 -0.020 -0.020 -0.009 -0.025
## polXDR 0.006 0.010 -0.014 0.009 0.024 0.001 -0.005
## polXBE 0.008 -0.011 -0.016 -0.024 0.000 -0.017 0.010 0.000
## polXDRvBE -0.004 -0.010 0.002 -0.011 -0.005 0.010 0.003 -0.009 -0.025
## int__.5:SES -0.001 -0.948 0.004 -0.009 0.013 -0.021 -0.009 -0.012 0.016
## pXDRBE
## intro_cc_.5
## SESladder
## policy
## DvR
## BvE
## DRvBE
## polXDR
## polXBE
## polXDRvBE
## int__.5:SES 0.008
E. collectivism/individualism scale
Collectivism/Individualism is not a significant moderator, p = .600.
SG$CIScale_c <- SG$CIScale - mean(SG$CIScale, na.rm = T)
summary(m<-lmer(partyLikely ~ intro_cc_.5*CIScale_c + policy + DvR + BvE + DRvBE + polXDR + polXBE + polXDRvBE + (1 | country), data = SG))
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: partyLikely ~ intro_cc_.5 * CIScale_c + policy + DvR + BvE +
## DRvBE + polXDR + polXBE + polXDRvBE + (1 | country)
## Data: SG
##
## REML criterion at convergence: 24771.7
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.4696 -0.7809 -0.3540 0.7178 2.7515
##
## Random effects:
## Groups Name Variance Std.Dev.
## country (Intercept) 0.1983 0.4454
## Residual 3.3704 1.8359
## Number of obs: 6095, groups: country, 7
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 2.698e+00 1.701e-01 5.999e+00 15.866 3.98e-06 ***
## intro_cc_.5 -2.522e-01 4.706e-02 6.078e+03 -5.359 8.68e-08 ***
## CIScale_c 3.764e-02 2.470e-02 6.012e+03 1.524 0.1276
## policy 1.798e-02 2.358e-02 6.079e+03 0.763 0.4457
## DvR 2.038e-02 3.347e-02 6.078e+03 0.609 0.5427
## BvE 2.706e-02 3.311e-02 6.078e+03 0.817 0.4138
## DRvBE 5.828e-02 2.354e-02 6.078e+03 2.476 0.0133 *
## polXDR 3.463e-03 3.348e-02 6.078e+03 0.103 0.9176
## polXBE 1.651e-02 3.312e-02 6.078e+03 0.498 0.6182
## polXDRvBE -3.007e-02 2.355e-02 6.078e+03 -1.277 0.2016
## intro_cc_.5:CIScale_c 2.359e-02 4.502e-02 6.078e+03 0.524 0.6003
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) in__.5 CIScl_ policy DvR BvE DRvBE polXDR polXBE
## intro_cc_.5 0.001
## CIScale_c 0.002 0.004
## policy 0.001 -0.010 0.024
## DvR 0.001 -0.002 -0.011 -0.001
## BvE -0.004 0.006 -0.017 0.006 0.001
## DRvBE -0.001 0.003 -0.001 -0.018 -0.006 -0.024
## polXDR -0.001 -0.005 0.021 0.007 0.023 0.000 0.000
## polXBE 0.001 0.012 0.012 -0.023 0.000 -0.012 0.007 0.001
## polXDRvBE -0.003 -0.006 0.000 -0.012 0.000 0.007 0.006 -0.006 -0.024
## in__.5:CIS_ 0.000 0.000 -0.003 -0.008 -0.003 -0.018 -0.003 -0.008 -0.020
## pXDRBE
## intro_cc_.5
## CIScale_c
## policy
## DvR
## BvE
## DRvBE
## polXDR
## polXBE
## polXDRvBE
## in__.5:CIS_ -0.005
F. party ID
Party ID of participant is not a significant moderator by itself.
summary(m<-lmer(partyLikely ~ intro_cc_.5*ptIvRD*ptRvD + policy + DvR + BvE + DRvBE + polXDR + polXBE + polXDRvBE + (1 | country), data = SG))
## fixed-effect model matrix is rank deficient so dropping 2 columns / coefficients
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: partyLikely ~ intro_cc_.5 * ptIvRD * ptRvD + policy + DvR + BvE +
## DRvBE + polXDR + polXBE + polXDRvBE + (1 | country)
## Data: SG
##
## REML criterion at convergence: 25154.3
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.5490 -0.7900 -0.3545 0.7022 2.7955
##
## Random effects:
## Groups Name Variance Std.Dev.
## country (Intercept) 0.1776 0.4214
## Residual 3.3656 1.8346
## Number of obs: 6191, groups: country, 7
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 2.710e+00 1.612e-01 6.019e+00 16.809 2.75e-06 ***
## intro_cc_.5 -2.742e-01 4.786e-02 6.172e+03 -5.730 1.05e-08 ***
## ptIvRD -2.489e-02 3.182e-02 6.178e+03 -0.782 0.4342
## ptRvD -4.049e-01 6.414e-02 6.178e+03 -6.313 2.92e-10 ***
## policy 1.402e-02 2.337e-02 6.173e+03 0.600 0.5486
## DvR 2.773e-02 3.321e-02 6.172e+03 0.835 0.4038
## BvE 4.004e-02 3.280e-02 6.172e+03 1.221 0.2222
## DRvBE 5.083e-02 2.335e-02 6.172e+03 2.177 0.0295 *
## polXDR -3.932e-03 3.323e-02 6.172e+03 -0.118 0.9058
## polXBE 1.646e-02 3.281e-02 6.172e+03 0.502 0.6158
## polXDRvBE -9.487e-03 2.336e-02 6.172e+03 -0.406 0.6846
## intro_cc_.5:ptIvRD -1.095e-02 6.263e-02 6.172e+03 -0.175 0.8612
## intro_cc_.5:ptRvD -5.517e-02 1.254e-01 6.172e+03 -0.440 0.6601
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation matrix not shown by default, as p = 13 > 12.
## Use print(x, correlation=TRUE) or
## vcov(x) if you need it
## fit warnings:
## fixed-effect model matrix is rank deficient so dropping 2 columns / coefficients
G. political ideology (symbolic beliefs)
No general symbolic beliefs are significant moderators, \(\beta\) = .01, t(6264) = .41, p = .683.
SG$symBelGen_c <- SG$symBelGen - mean(SG$symBelGen, na.rm = T)
summary(m<-lmer(partyLikely ~ intro_cc_.5*ideology_c + policy + DvR + BvE + DRvBE + polXDR + polXBE + polXDRvBE + (1 | country), data = SG))
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: partyLikely ~ intro_cc_.5 * ideology_c + policy + DvR + BvE +
## DRvBE + polXDR + polXBE + polXDRvBE + (1 | country)
## Data: SG
##
## REML criterion at convergence: 25458.2
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.7017 -0.7813 -0.3545 0.7064 2.8883
##
## Random effects:
## Groups Name Variance Std.Dev.
## country (Intercept) 0.1549 0.3936
## Residual 3.3341 1.8260
## Number of obs: 6281, groups: country, 7
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 2.710e+00 1.507e-01 6.000e+00 17.988 1.90e-06 ***
## intro_cc_.5 -2.710e-01 4.612e-02 6.264e+03 -5.876 4.41e-09 ***
## ideology_c 1.683e-01 1.618e-02 6.269e+03 10.402 < 2e-16 ***
## policy 1.369e-02 2.309e-02 6.265e+03 0.593 0.5533
## DvR 2.727e-02 3.275e-02 6.264e+03 0.833 0.4051
## BvE 3.606e-02 3.248e-02 6.264e+03 1.110 0.2669
## DRvBE 5.095e-02 2.307e-02 6.264e+03 2.209 0.0272 *
## polXDR 2.489e-03 3.275e-02 6.264e+03 0.076 0.9394
## polXBE 2.273e-02 3.248e-02 6.264e+03 0.700 0.4841
## polXDRvBE -2.498e-02 2.307e-02 6.265e+03 -1.083 0.2790
## intro_cc_.5:ideology_c 1.287e-02 3.153e-02 6.264e+03 0.408 0.6831
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) in__.5 idlgy_ policy DvR BvE DRvBE polXDR polXBE
## intro_cc_.5 0.002
## ideology_c 0.005 -0.011
## policy 0.001 -0.008 0.007
## DvR 0.001 -0.005 -0.006 0.002
## BvE -0.004 0.008 0.007 0.006 0.000
## DRvBE -0.001 0.002 -0.022 -0.018 -0.006 -0.022
## polXDR 0.000 -0.003 0.015 0.006 0.023 0.001 -0.004
## polXBE 0.001 0.015 0.006 -0.021 0.000 -0.013 0.007 0.000
## polXDRvBE -0.004 -0.006 -0.016 -0.010 -0.002 0.007 0.005 -0.005 -0.022
## intr_c_.5:_ -0.001 0.001 0.002 -0.017 0.010 -0.002 -0.006 0.001 0.006
## pXDRBE
## intro_cc_.5
## ideology_c
## policy
## DvR
## BvE
## DRvBE
## polXDR
## polXBE
## polXDRvBE
## intr_c_.5:_ 0.012
H. personal family risk perception
- riskContractCovid (mean centered): What do you believe is the likelihood you or close friends or family will personally contract Covid-19?
Results
Not a signficant moderator.
SG$riskContractCovid_c <- SG$riskContractCovid - mean(SG$riskContractCovid, na.rm = T)
summary(m<-lmer(partyLikely ~ intro_cc_.5* riskContractCovid_c + policy + DvR + BvE + DRvBE + polXDR + polXBE + polXDRvBE + (1 | country), data = SG))
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: partyLikely ~ intro_cc_.5 * riskContractCovid_c + policy + DvR +
## BvE + DRvBE + polXDR + polXBE + polXDRvBE + (1 | country)
## Data: SG
##
## REML criterion at convergence: 17078.5
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.8269 -0.7882 -0.3696 0.6891 2.8528
##
## Random effects:
## Groups Name Variance Std.Dev.
## country (Intercept) 0.2914 0.5398
## Residual 3.2503 1.8028
## Number of obs: 4235, groups: country, 7
##
## Fixed effects:
## Estimate Std. Error df t value
## (Intercept) 2.71528 0.20621 5.88625 13.168
## intro_cc_.5 -0.26749 0.05547 4218.18004 -4.822
## riskContractCovid_c -0.15043 0.01744 3652.10510 -8.625
## policy 0.02634 0.02779 4218.90506 0.948
## DvR -0.01987 0.03936 4218.09151 -0.505
## BvE 0.03764 0.03911 4218.20242 0.963
## DRvBE 0.03053 0.02774 4218.09881 1.100
## polXDR 0.04419 0.03934 4218.04972 1.123
## polXBE 0.02338 0.03911 4217.96552 0.598
## polXDRvBE -0.02573 0.02776 4218.42710 -0.927
## intro_cc_.5:riskContractCovid_c 0.02922 0.02624 4218.26436 1.114
## Pr(>|t|)
## (Intercept) 1.37e-05 ***
## intro_cc_.5 1.47e-06 ***
## riskContractCovid_c < 2e-16 ***
## policy 0.343
## DvR 0.614
## BvE 0.336
## DRvBE 0.271
## polXDR 0.261
## polXBE 0.550
## polXDRvBE 0.354
## intro_cc_.5:riskContractCovid_c 0.265
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) in__.5 rskCC_ policy DvR BvE DRvBE polXDR polXBE
## intro_cc_.5 0.002
## rskCntrctC_ -0.014 0.016
## policy 0.001 -0.016 0.020
## DvR 0.002 -0.005 0.028 -0.001
## BvE -0.002 0.017 0.007 0.007 0.000
## DRvBE -0.001 -0.002 -0.014 -0.013 -0.013 -0.019
## polXDR -0.001 0.000 0.018 0.012 0.030 0.000 0.000
## polXBE 0.001 0.009 -0.007 -0.020 0.000 0.001 0.008 0.000
## polXDRvBE -0.001 -0.007 0.022 -0.008 0.002 0.007 0.014 -0.011 -0.020
## int__.5:CC_ 0.000 -0.001 0.020 0.011 -0.003 0.006 0.011 -0.016 -0.031
## pXDRBE
## intro_cc_.5
## rskCntrctC_
## policy
## DvR
## BvE
## DRvBE
## polXDR
## polXBE
## polXDRvBE
## int__.5:CC_ 0.008
- riskExpSymp (mean centered): If you or close friends or family were to contract Covid-19, what do you believe is the likelihood that either you or someone close to you would develop serious health consequences such as great difficulty breathing, pneumonia, and dangerously high fever?
Results
Not a significant moderator.
SG$riskExpSymp_c <- SG$riskExpSymp - mean(SG$riskExpSymp, na.rm = T)
summary(m<-lmer(partyLikely ~ intro_cc_.5* riskExpSymp_c + policy + DvR + BvE + DRvBE + polXDR + polXBE + polXDRvBE + (1 | country), data = SG))
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: partyLikely ~ intro_cc_.5 * riskExpSymp_c + policy + DvR + BvE +
## DRvBE + polXDR + polXBE + polXDRvBE + (1 | country)
## Data: SG
##
## REML criterion at convergence: 16947.8
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.9312 -0.7408 -0.3590 0.6713 3.0686
##
## Random effects:
## Groups Name Variance Std.Dev.
## country (Intercept) 0.3295 0.574
## Residual 3.1564 1.777
## Number of obs: 4233, groups: country, 7
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 2.713e+00 2.189e-01 5.922e+00 12.392 1.86e-05 ***
## intro_cc_.5 -2.719e-01 5.467e-02 4.216e+03 -4.973 6.87e-07 ***
## riskExpSymp_c -2.216e-01 1.584e-02 4.012e+03 -13.990 < 2e-16 ***
## policy 3.044e-02 2.738e-02 4.217e+03 1.112 0.266
## DvR -4.266e-03 3.878e-02 4.216e+03 -0.110 0.912
## BvE 4.180e-02 3.855e-02 4.216e+03 1.084 0.278
## DRvBE 2.093e-02 2.735e-02 4.216e+03 0.765 0.444
## polXDR 4.472e-02 3.877e-02 4.216e+03 1.153 0.249
## polXBE 2.208e-02 3.853e-02 4.216e+03 0.573 0.567
## polXDRvBE -2.762e-02 2.736e-02 4.216e+03 -1.009 0.313
## intro_cc_.5:riskExpSymp_c 2.574e-02 2.556e-02 4.216e+03 1.007 0.314
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) in__.5 rskES_ policy DvR BvE DRvBE polXDR polXBE
## intro_cc_.5 0.002
## rskExpSymp_ -0.008 0.014
## policy 0.001 -0.017 0.001
## DvR 0.002 -0.006 -0.011 -0.002
## BvE -0.002 0.017 0.000 0.007 0.000
## DRvBE -0.001 -0.001 0.017 -0.014 -0.012 -0.019
## polXDR -0.001 -0.001 0.011 0.012 0.030 0.000 0.001
## polXBE 0.001 0.010 -0.005 -0.019 0.000 -0.001 0.008 0.000
## polXDRvBE -0.001 -0.007 0.024 -0.009 0.001 0.008 0.014 -0.011 -0.019
## int__.5:ES_ 0.000 -0.001 0.024 -0.013 -0.012 0.004 0.021 -0.010 -0.008
## pXDRBE
## intro_cc_.5
## rskExpSymp_
## policy
## DvR
## BvE
## DRvBE
## polXDR
## polXBE
## polXDRvBE
## int__.5:ES_ -0.001
- riskHealthCons (mean centered): How severe have the personal health consequences of Covid-19 been to you or your close friend or family member who contracted the virus?
Results
Not a signficant moderator.
SG$riskHealthCons_c <- SG$riskHealthCons - mean(SG$riskHealthCons, na.rm = T)
summary(m<-lmer(partyLikely ~ intro_cc_.5*riskHealthCons_c + policy + DvR + BvE + DRvBE + polXDR + polXBE + polXDRvBE + (1 | country), data = SG))
## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: partyLikely ~ intro_cc_.5 * riskHealthCons_c + policy + DvR +
## BvE + DRvBE + polXDR + polXBE + polXDRvBE + (1 | country)
## Data: SG
##
## REML criterion at convergence: 7224.1
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -1.7013 -0.8040 -0.3026 0.7834 2.6208
##
## Random effects:
## Groups Name Variance Std.Dev.
## country (Intercept) 0.3145 0.5608
## Residual 3.5083 1.8730
## Number of obs: 1753, groups: country, 6
##
## Fixed effects:
## Estimate Std. Error df t value Pr(>|t|)
## (Intercept) 2.88336 0.23613 4.94505 12.211 7.00e-05
## intro_cc_.5 -0.25998 0.08965 1737.00382 -2.900 0.00378
## riskHealthCons_c -0.10416 0.02489 1741.95188 -4.185 2.99e-05
## policy -0.04038 0.04490 1737.50662 -0.899 0.36858
## DvR 0.08381 0.06327 1737.28321 1.325 0.18547
## BvE 0.02821 0.06354 1737.22981 0.444 0.65714
## DRvBE 0.12900 0.04488 1737.60998 2.874 0.00410
## polXDR -0.11678 0.06349 1738.49217 -1.839 0.06603
## polXBE -0.00686 0.06362 1737.29314 -0.108 0.91414
## polXDRvBE -0.03491 0.04494 1737.66184 -0.777 0.43732
## intro_cc_.5:riskHealthCons_c 0.00515 0.04726 1737.42551 0.109 0.91323
##
## (Intercept) ***
## intro_cc_.5 **
## riskHealthCons_c ***
## policy
## DvR
## BvE
## DRvBE **
## polXDR .
## polXBE
## polXDRvBE
## intro_cc_.5:riskHealthCons_c
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Correlation of Fixed Effects:
## (Intr) in__.5 rskHC_ policy DvR BvE DRvBE polXDR polXBE
## intro_cc_.5 0.007
## rskHlthCns_ 0.016 0.010
## policy -0.010 0.018 0.035
## DvR -0.006 0.003 -0.004 0.011
## BvE -0.005 0.004 -0.007 -0.015 0.000
## DRvBE 0.001 0.022 -0.008 -0.025 0.006 -0.011
## polXDR -0.004 -0.018 -0.059 -0.007 -0.007 0.002 -0.011
## polXBE -0.005 0.033 0.001 -0.008 0.001 -0.053 -0.016 -0.001
## polXDRvBE -0.010 0.011 0.032 0.007 -0.008 -0.016 -0.030 0.008 -0.009
## int__.5:HC_ 0.006 0.002 0.017 0.015 -0.041 0.011 0.008 0.021 -0.019
## pXDRBE
## intro_cc_.5
## rskHlthCns_
## policy
## DvR
## BvE
## DRvBE
## polXDR
## polXBE
## polXDRvBE
## int__.5:HC_ -0.015
Social Gains Scenario
1. descriptives
violin plot
country means
across country means
2. Models
A. partyLikely ~ intro_cc + policy + DvR + BvE + DRvBE + polXDR + polXBE + polXDRvBE + (1 | country)
Likelihood of attending party regressed on contrast coded introspection condition accounting for random intercepts for country (the model would not converge with random slopes)
intro_cc_.5: introspection = +.5; control = -.5
Across countries, on average as you move from control to introspection condition, there is a .26 decrease in the likelihood of attending the party, \(\beta_1\) = -.26, t(6282) = -5.71, p < .001.
B. (2 DF test) partyLikely ~ intro_cc + introSpread_c + introExperienceSympt_c + policy + DvR + BvE + DRvBE + polXDR + polXBE + polXDRvBE + (1 | country)
2 DF test: Likelihood of attending party regressed on contrast coded introspection condition, introspection intervention rating for likelihood of spreading Covid-19 and experiencing symptoms of Covid-19, while controlling for contrast coded party over policy condition. Again, this also accounts for random intercepts for country (the model would not converge with random slopes)
Predictors excluded:
introSpread (mean centered): How much should your likelihood of spreading Covid-19 influence your decision to attend the gathering? (1 = not at all; 7 = extremely)
introExpSymp (mean centered): How much should your likelihood of experiencing severe symptoms from Covid-19 influence your decision to attend the gathering? (1 = not at all; 7 = extremely)
Results:
As expected, for our 2 degree of freedom test, there is a significant effect of the introspection question items. As a person rates higher that the spread and experiencing symptoms of Covid-19 should influence their decision to attend the party, they are less likely to attend the party, PRE = .18, F(2,6247) = 693.76, p < .001.
C. partyLikely ~ intro_cc + avgIntro_c + policy + DvR + BvE + DRvBE + polXDR + polXBE + polXDRvBE + (1 | country)
Likelihood of attending party regressed on contrast coded introspection condition, and the mean centered average for introspection intervention ratings, while controlling for contrast coded party over policy condition. Again, this also accounts for random intercepts for country (the model would not converge with random slopes)
Predictors excluded
avg_screen (mean centered): the average of introspection items for how much you should consider the likelihood of spreading Covid-19 to others and experiencing symptoms when deciding whether to attend the gathering
Results
Controlling for party over policy condition, and when including random intercepts for country, there is a significant effect for introspection condition even when controlling for mean centered average rating for the introspection items. As you move from control to introspection condition, there is a .21 decrease in the likelihood of attending the party, \(\beta_1\) = -.21, t(6241.15) = -4.90, p < .001.
3. test for moderators
A. education (alone)
Note: I exclude the 85 participants from Brazil that chose #18 for their education.
Education is a marginally significant moderator, \(\beta\) = -.03, t(6057) = -1.73, p = .084.
B. Eduction and average of symbolic ideology
Results
the interaction between education and introspection remains marginally significant, \(\beta\) = -.03, t(6039) = -1.81, p = .070.
There is no significant 2-way interaction between introspection and ideology, p = .677, or 3-way interaction between introspection, ideology, and education, p = .162.
C. Eduction and party identity
Results
the interaction between education and introspection remains marginally significant, \(\beta\) = -.034, t(5950) = -1.88, p = .060.
There is no significant 2-way interaction between introspection and party identity for participant or 3-way interaction between introspection, participant party, and education.
D. ses ladder
SES ladder is not a significant moderator, p = .622.
E. collectivism/individualism scale
Collectivism/Individualism is not a significant moderator, p = .600.
F. party ID
Party ID of participant is not a significant moderator by itself.
G. political ideology (symbolic beliefs)
No general symbolic beliefs are significant moderators, \(\beta\) = .01, t(6264) = .41, p = .683.
H. personal family risk perception
Results
Not a signficant moderator.
Results
Not a significant moderator.
Results
Not a signficant moderator.