The aim of the project was to provide experimental evidence on the nature and extent of pressure to conform, self-censorship, and the concealment of discriminatory attitudes in customer-facing public authorities and to explain these with reference to specific authority structures and cultures. We implemented two list experiments in our own public authority survey to measure the refusal of public authority employees to openly express a) racist views and b) observed racism. This indirect survey method avoids the ubiquitous problem of social desirability bias in sensitive questions. In our analysis, we used list experiments to determine the extent of self-censorship as a substantive variable of interest. The core idea is a systematic comparison of indirect and direct questionnaire items. This enabled us to determine self-censorship behaviour as a central result of the analysis: Self-censoring = true attitude - reported attitude. We refer to the racism shown in the list experiments as the "true" attitude of the public authority employees.
In the direct survey, 7.8 percent of the employees of all four public authorities surveyed—the Federal Office for Migration and Refugees (Bundesamt für Migration und Flüchtlinge, BAMF), the Federal Employment Agency (Bundesagentur für Arbeit, BA), the Federal Police (Bundespolizei, BP), and the German customs—agree with the statement that some ethnic groups are inherently less intelligent. In the indirect survey using a list experiment, this proportion rises to 11.6 percent. This results in a difference of 3.8 percentage points, which points to self-censorship behaviour that is not very pronounced in terms of substance but nevertheless statistically verifiable. When asked about the perception of racism in their own authority, 13.4 percent of authority employees affirm the existence of racist discrimination in a direct survey. In the list experiment, this figure rises to 15.7 percent, which corresponds to a difference of 2.7 percentage points. Self-censorship behaviour can also be detected here, although it is not very pronounced overall. However, there are major differences between the various types of authorities. For example, both racist attitudes and their concealment are more widespread in the federal police than in the other three public authorities. For the federal police, the list experiment yields around twice as high a proportion of racist attitudes as a direct survey. This indicates considerable self-censorship behavior.
