Null hypothesis significance testing (NHST) remains the most frequently used approach to statistical inference in communication research, in spite of serious concerns about the ability of researchers to properly interpret its results. Drawing on data from a survey of the ICA membership (N = 221), we assess the degree to which communication scholars endorse false statements about the interpretation of inferential statistics, looking not only at NHST but also at confidence intervals as its most prominent alternative. The vast majority of communication researchers committed at least one error in interpreting both p-values (91%) and confidence intervals (96%), while being confident, on average, about their performance. The performance of pre- and postdocs hardly differed, and statistical experience (training, teaching, and applied) did not entail substantially better interpretation outcomes. In sum, the findings point to major problems regarding the quality of conclusions communication researchers draw from their data.
