Participants: 160 MTurk
Design: First, participants either completed a moral sentence prime or a control prime. All participants were told that they were outperformed by their friend on the midterm exam. They were then told that they discovered some information on a test that they and their friend hadn't covered together, and there there was a 10, 20, 30, 40, 50, 60, 70, 80, 90 or 100% chance that the information was going to show up on the exam. They were asked if they would call or not.
2 (moral prime sentences vs. control) x 10 (10, 20, 30, 40, 50, 60, 70, 80, 90, 100% chance of material on test; treated as a continuous variable).
Predictions: Those in the control condition would show a positive correlation between chance and calling, whereas those in the moral condition should all be equally likely to call (and more likely than the control condition) across chance.
Main effects
res1 = glm(call ~ chance, family = binomial, data = su)
summary(res1) #no main effect of chance on yes/no call
##
## Call:
## glm(formula = call ~ chance, family = binomial, data = su)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.958 0.564 0.593 0.623 0.654
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 1.3969 0.4574 3.05 0.0023 **
## chance 0.0361 0.0754 0.48 0.6322
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 145.25 on 159 degrees of freedom
## Residual deviance: 145.02 on 158 degrees of freedom
## AIC: 149
##
## Number of Fisher Scoring iterations: 4
res1.1 = glm(call ~ poly(chance, 2), family = binomial, data = su)
summary(res1.1) #no quadratic effect
##
## Call:
## glm(formula = call ~ poly(chance, 2), family = binomial, data = su)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.938 0.576 0.588 0.617 0.666
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 1.598 0.212 7.55 4.4e-14 ***
## poly(chance, 2)1 1.260 2.656 0.47 0.64
## poly(chance, 2)2 -0.316 2.666 -0.12 0.91
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 145.25 on 159 degrees of freedom
## Residual deviance: 145.00 on 157 degrees of freedom
## AIC: 151
##
## Number of Fisher Scoring iterations: 4
res2 = lm(continuous ~ chance, data = su)
summary(res2) #hm not replicating the main effects here
##
## Call:
## lm(formula = continuous ~ chance, data = su)
##
## Residuals:
## Min 1Q Median 3Q Max
## -3.177 -0.232 -0.012 0.851 0.988
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 3.9845 0.1919 20.76 <2e-16 ***
## chance 0.0275 0.0308 0.89 0.37
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 1.09 on 158 degrees of freedom
## Multiple R-squared: 0.00501, Adjusted R-squared: -0.00128
## F-statistic: 0.796 on 1 and 158 DF, p-value: 0.374
res3 = lm(continuous ~ poly(chance, 2), data = su)
summary(res3) #no quadratic effect
##
## Call:
## lm(formula = continuous ~ poly(chance, 2), data = su)
##
## Residuals:
## Min 1Q Median 3Q Max
## -3.222 -0.212 0.088 0.816 1.088
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 4.1375 0.0864 47.89 <2e-16 ***
## poly(chance, 2)1 0.9734 1.0928 0.89 0.37
## poly(chance, 2)2 -0.7430 1.0928 -0.68 0.50
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 1.09 on 157 degrees of freedom
## Multiple R-squared: 0.00794, Adjusted R-squared: -0.0047
## F-statistic: 0.628 on 2 and 157 DF, p-value: 0.535
res4 = lm(continuous ~ condition, data = su)
summary(res4) #no main effect of condition
##
## Call:
## lm(formula = continuous ~ condition, data = su)
##
## Residuals:
## Min 1Q Median 3Q Max
## -3.205 -0.205 -0.073 0.795 0.927
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 4.073 0.121 33.79 <2e-16 ***
## conditionmoral 0.132 0.173 0.76 0.45
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 1.09 on 158 degrees of freedom
## Multiple R-squared: 0.00368, Adjusted R-squared: -0.00262
## F-statistic: 0.584 on 1 and 158 DF, p-value: 0.446
Quick plot of the data
xyplot(continuous ~ chance | condition, data = su)
(nonsignificant main effect of chance on continuous; multiply the bottom by 10)
plot(effect(term = "chance", mod = res2, xlevels = list(condition = c(0, 1))),
multiline = TRUE)
Main analyses
res5 = glm(call ~ chance * condition, family = binomial, data = su)
summary(res5) #no interaction on dichotomous variable
##
## Call:
## glm(formula = call ~ chance * condition, family = binomial, data = su)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -2.045 0.514 0.531 0.671 0.708
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 1.23559 0.55395 2.23 0.026 *
## chance 0.02101 0.09693 0.22 0.828
## conditionmoral 0.59520 0.99667 0.60 0.550
## chance:conditionmoral -0.00678 0.15877 -0.04 0.966
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 145.25 on 159 degrees of freedom
## Residual deviance: 143.38 on 156 degrees of freedom
## AIC: 151.4
##
## Number of Fisher Scoring iterations: 4
res6 = lm(continuous ~ chance * condition, data = su)
summary(res6) #no interaction on continuous variable
##
## Call:
## lm(formula = continuous ~ chance * condition, data = su)
##
## Residuals:
## Min 1Q Median 3Q Max
## -3.232 -0.261 -0.004 0.860 1.005
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 3.9754 0.2493 15.95 <2e-16 ***
## chance 0.0193 0.0430 0.45 0.65
## conditionmoral 0.0504 0.3960 0.13 0.90
## chance:conditionmoral 0.0102 0.0631 0.16 0.87
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 1.1 on 156 degrees of freedom
## Multiple R-squared: 0.00755, Adjusted R-squared: -0.0115
## F-statistic: 0.395 on 3 and 156 DF, p-value: 0.756
res6.1 = lm(continuous ~ poly(chance, 2) * condition, data = su)
summary(res6.1) #no interaction including quadratic effect of continuous variable
##
## Call:
## lm(formula = continuous ~ poly(chance, 2) * condition, data = su)
##
## Residuals:
## Min 1Q Median 3Q Max
## -3.298 -0.255 0.093 0.861 1.120
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 4.083 0.124 33.05 <2e-16 ***
## poly(chance, 2)1 0.585 1.554 0.38 0.71
## poly(chance, 2)2 -0.548 1.559 -0.35 0.73
## conditionmoral 0.100 0.177 0.57 0.57
## poly(chance, 2)1:conditionmoral 0.689 2.292 0.30 0.76
## poly(chance, 2)2:conditionmoral -0.492 2.251 -0.22 0.83
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 1.1 on 154 degrees of freedom
## Multiple R-squared: 0.011, Adjusted R-squared: -0.0211
## F-statistic: 0.342 on 5 and 154 DF, p-value: 0.887
Plotting the nonsignificant interaction just to take a look. Slight main effect of condition in the right direction at least.
plot(effect(term = "chance*condition", mod = res6, xlevels = list(condition = c(0,
1))), multiline = TRUE)
