Evaluating the Sampling Precision of Social Identity Related Published Research

Abstract

Social identity theory states that a person’s sense of who they are is based largely on their group membership(s). We categorize ourselves, identify with groups, and compare our groups with others, in the hopes that our self-esteem is maintained or boosted from this comparison. A prerequisite of scientific research, even regarding social identity, is that researchers need to be confident that the empirical facts really are factual; that is, that the sample statistics reported accurately estimate corresponding population parameters; this is known as sampling precision, or how precisely our sample corresponds to our population. By employing the recently invented a priori procedure, the present research assesses the sampling precision with which published experimental and correlational social identity research statistics, across three time periods, estimate corresponding population parameters. We hypothesized 1: The precision of research in the social identity area should be hopefully below the 0.10 level or at least the 0.20 level for true experimental designs and 2: Precision in the social identity area should be improving, with recent social identity research enjoying a precision advantage over less recent social identity research. A sample of 75 academic papers, across 46 different journals was collected for analysis. For experimental studies, the mean precision level was 0.51 and the median precision level was 0.50 (n = 39). For correlational studies, the mean precision level was 0.24 and the median precision level was 0.20 (n = 55). The main findings are pessimistic, but with the glimmer of light that precision is improving.

Keywords: precision, confidence, social identity theory, a priori procedure, estimation