Estimating the prevalence of negative attitudes towards people with disability: a comparison of direct questioning, projective questioning and randomised response

Abstract

Purpose. Despite being susceptible to social desirability bias, attitudes towards people with disabilities are traditionally assessed via self-report. We investigated two methods presumably providing more valid prevalence estimates of sensitive attitudes than direct questioning (DQ). Most people projective questioning (MPPQ) attempts to reduce bias by asking interviewees to estimate the number of other people holding a sensitive attribute, rather than confirming or denying the attribute for themselves. The randomised-response technique (RRT) tries to reduce bias by assuring confidentiality through a random scrambling of the respondent's answers.

Method. We assessed negative attitudes towards people with physical and mental disability via MPPQ, RRT and DQ to compare the resulting estimates.

Results. The MPPQ estimates exceeded the DQ estimates. Employing a cheating-detection extension of the RRT, we determined the proportion of respondents disregarding the RRT instructions and computed an upper bound for the prevalence of negative attitudes. MPPQ estimates exceeded this upper bound and were thus shown to overestimate the prevalence. Furthermore, we found more negative attitudes towards people with mental disabilities than those with physical disabilities in all three questioning conditions.

Conclusions. We recommend employing the cheating-detection variant of the RRT to gain additional insight in future studies on attitudes towards people with disabilities.

Keywords:

• Physical disability
• mental disability
• projective questioning
• randomised-response technique
• social desirability bias

Acknowledgements

We are grateful to Dennis Winter, Martin Grupe and Sven Keiner for their help in collecting the data, and to Ute J. Bayen and Morten Moshagen for helpful comments on an earlier version of the manuscript. This work was supported by the German Research Foundation (MU 2674/1–1).

Note

1. For reasons of consistency, we follow Clark and Desharnais [44] by referring to these respondents as ‘cheaters’. It is important to note, however, that in more neutral terms they might just as well be called ‘noncompliant’ or ‘disobedient’ participants.