Differences in measurement quality depending on recall: results for a question about trust in the parliament

Abstract

A large body of studies estimate the measurement quality of survey questions using multitrait-multimethod (MTMM) experiments. In these experiments, respondents are asked (nearly) identical questions at least twice. Most commonly, this is done within the same survey by, for instance, changing the response scale characteristics. However, it has frequently been argued that the estimates from these experiments might be biased when respondents recall their first answer and base their second answer on the initial one. So far, only little is known about the impact of memory effects on the estimates of measurement quality. In this study, we conducted a MTMM experiment in the probability-based German Internet Panel (N = 4,401) to investigate whether and to what extent measurement quality estimates differ across respondent groups varying in terms of recall. For this purpose, we use a survey question on trust in the German parliament that was asked with three different response scales. The results reveal that the recall of answers and memory effects vary significantly depending on the scales used for the trust in the parliament question. We also found significant differences in measurement quality across different recall groups. However, more refined research is necessary to better understand the association between answer recall and measurement quality estimates.