The Tip of the Iceberg? From Data Quality to Exploring Publication Bias in the Social Science

How can we ensure trustworthy knowledge in the social sciences? Two central challenges are data quality and publication bias. Data quality determines whether our empirical evidence is reliable, comparable, and valid, while publication bias threatens the integrity of the scientific record by privileging significant and “positive” findings over null results.
In this talk, I will connect these two perspectives. First, I introduce core dimensions of data quality and discuss how the KODAQS initiative (Competence Center Data Quality in the Social Sciences) supports researchers in addressing quality challenges across traditional surveys and emerging digital data sources. Second, I turn to publication bias as a systemic quality problem of knowledge production. Drawing on the PubBias project, I present empirical evidence from 178 studies in the SOEP-IS and GESIS Panel.pop. Our findings show that a large share of study proposals never reach publication, that most published hypotheses are new rather than preregistered, and that supported hypotheses remain disproportionately represented.
By linking data quality with publication practices, I argue that trustworthy social science requires quality assurance at both ends: in how we collect and curate data, and in how we decide which findings enter the published record.