Though some forms of data-dredging are lamentably common, it is important to note that often such problems arise from a lack of awareness rather than malfeasance. ‘false positives’) and is thus unreliable. Such an analysis can often generate statistically significant results in absence of a true effect (i.e. While many different choices might be defensible, a canonical case of p-hacking would involve trying out multiple different options and reporting the result which yields the lowest p-value (particularly when alternative choices generate values that do not yield a significant result). how to handle outliers, whether to combine groups, including/excluding covariates) which will produce a statistically significant p-value. In contrast, p-hacking occurs when an initial analysis produces results which are close to being statistically significant, then, in absence of a study protocol, researchers can make analytic choices (e.g. Ideally, these choices are guided by the principles of best practice and prespecified in a publicly available protocol. For example, in nearly any analysis of data there are several “researcher degrees of freedom”- i.e., choices that must be made in the process of analysis. fishing, p-hacking), but each essentially involves probing the data in unplanned ways, finding and reporting an “attractive” result, without accurately conveying the course of analysis. Data-dredging bias is a general category which includes a number of misuses of statistical inference (e.g.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |