FIGURE 8.

Effect of information treatments on pro-women attitude (Mturk sample). The gray bars show the average pro-women attitude in a given treatment. Each bar in the control group shows the average pro-women attitude for pro-women, neutral and pro-men respondents, respectively. The left-most bar is based on 2,144 judgments made by 268 respondents who were pro-women in the base scenarios and were randomly assigned to the control group, judging four pairs of scenarios involving two judgments each. Each of the next four bars is based on 458 judgments made by 229 respondents who were pro-women in the base scenarios and were randomly assigned to any of the three treatment groups. Similarly, the left-most bar in the middle section is based on 340 judgments by 170 respondents who were neutral in the base scenario, and each of the next four bars is based on 366 judgments made by 183 respondents. The left-most bar in the right-most section is based on 340 judgments made by 170 respondents who were pro-men in the base scenario, and the next four are based on 298 judgments by 149 respondents. Vertical lines show 95 percent confidence intervals based on standard errors clustered at the individual level. Statistical significance of the difference between control group and scenarios with different information treatments is denoted by ***p-value < 0.01, **p-value < 0.05, and *p-value < 0.10.