Do employees speak their mind about sexism and diversity policies at work?

In some countries, like France, gender quotas for executives are implemented by the government. In other countries, like the United States, firms have implemented diversity training programs that inform workers about sexism and behaviour. So far, the effectiveness of these policies appears limited. A possible explanation is that employees covertly resist these policies. Therefore, we wanted to study whether workers openly voice their opinions about sexism in the workplace and DEI policies, or whether their reported views contain social desirability bias (SDB).

‘Our findings suggest that a sizeable fraction of managers do not actually support DEI policies’

For our study, we compared data from the United States to data from France, as those two countries have different approaches to DEI policies. We used a list experiment to obtain the relevant information. Respondents were randomly assigned to a veiled condition or a direct condition. In the veiled condition, respondents receive four lists of five statements and must say with how many they agree, not disclosing which ones in particular. One of the five statements in each list is potentially sensitive and relates to sexism or DEI policies. In the direct condition, respondents receive lists of the same four non-sensitive statements but must provide an answer to the potentially sensitive statement separately. The reasoning behind this is that people in the veiled group can express their real opinions without exposing them, whereas people in the direct group could be more inclined to give socially desirable answers. Comparison between the groups gives an estimate of the true attitude and an estimate of the degree of SDB regarding the sensitive issue. 

Substantial differences 

The survey was distributed by an online panel company to employed workers in five male-dominated industries: information technology, finance, law, business consultancy, and engineering. Eventually, our sample contained 703 respondents for France and 921 for the United States. In our research, we found out that there are substantial differences between the two conditions for all four sensitive statements in France and for two sensitive statements in the United States. For instance, among French respondents in the direct condition, 75 percent agree with ‘I dislike sexist jokes in the workplace.’ However, the estimated difference between the direct and the veiled condition is 24 percentage points, implying that only about half of the respondents truly agree with this statement. For the statement ‘Diversity, equity, and inclusion policies at the workplace do more good than harm’, the difference in the (estimated) fraction of respondents that agree between the direct and the veiled condition is about 18 percentage points in both France and the US. 

Policy (in)effectiveness 

Given the design of the list experiment, we can interpret the differences between the veiled and direct conditions as the extent of SDB. We conclude that there is widespread SDB in France and in the United States, which use fairly different DEI policies. Remarkably, we document SDB among both men and women. Our findings imply that it is important to disentangle changes in actual attitudes from changes in stated views in evaluating the effectiveness of DEI policies. Also, our findings suggest that a sizeable fraction of managers do not actually support DEI policies. This lack of genuine managerial support further hampers policy effectiveness.