Grey literature and publication bias in MetaLab meta-analyses
Date created: | Last Updated:
: DOI | ARK
Creating DOI. Please wait...
Description: Meta-analyses are an indispensable research synthesis tool for characterizing bodies of literature and advancing theories. In typical meta-analyses, noisy measurements from multiple independent samples are normalized onto a single scale (typically a measure of effect size) and combined statistically to produce a more accurate measurement. Effects for meta-analysis can come from the published literature, unpublished data, or even the author’s own work, but different strategies for identifying datapoints for inclusion can have major consequences for the interpretation of the meta-analytic estimate. In particular, the exclusion of unpublished work can lead to a bias for positive findings and compromise validity. Thus, it is important to assess the utility – and impact – of strategies for including unpublished datas. In the present article, we describe our successes and failures with gathering unpublished data for meta-analyses within developmental psychology, and assess how the addition of these datapoints changes the conclusions drawn from meta-analyses. PREPRINT CAN BE FOUND IN MetaArXiv LINK BELOW