Social media advertising has become a common tool for participant recruitment, especially when targeting hard-to-reach or underrepresented populations. A common strategy to boost engagement is the use of social group cues – textual or visual references to religion, gender identity, or ethnicity. While these cues can enhance recruitment efficiency, their impact on response data quality remains poorly understood. This study investigates how social group cues affect problematic response behaviour in a Facebook-recruited survey on labour market experiences in Germany. Respondents were recruited via four distinct ad conditions: two referencing Muslim identity, and two with neutral framing. Seven indicators were used …

A procedure for examining group differences in predictability of latent constructs with social measurement instruments is outlined. The method is developed within the framework of latent variable modeling and is widely applicable with popular software. The approach is based on the notions of maximal reliability and optimal linear combination that have been receiving increased attention over the past several decades. The procedure is readily employed for point and interval estimation of the population discrepancy in construct predictability using the instrument components. The discussed method is related to the study of measurement invariance and is illustrated with numerical data.

Survey researchers recognize that total survey error consists of multiple components, broadly pertaining to error sources and biases along the measurement process on the one hand and representation on the other hand. However, the relationship between these different error sources remains less well understood. Drawing on 1,452 estimates of measurement error from large-scale MTMM experiments in the European Social Survey, covering 24 countries and 21 languages, we investigate the relationship between measurement error, representation bias, country, and language. Our findings indicate a positive association between measurement error and representation bias. Measurement error is very similar across languages within the same …

Numerous national and international survey programs, including general population surveys, have been shifting their mode of data collection from face-to-face to self-completion designs. Push-to-web designs are a particularly promising approach that, however, also introduces methodical challenges. In this paper, we address several of these challenges by providing in-depth empirical analyses of selection and measurement effects from a cross-sectional and longitudinal perspective (time trends across years and survey modes). We use data from the Social Survey Austria (SSA) 2018, 2023 and 2024, including questions from the Austrian implementation of the International Social Survey Programme (ISSP). The survey transitioned from Computer-Assisted Personal …