P-hacking, also known by "data dredging" and a few other fun names, is an attempt to achieve a result by altering the conditions in statistical [[hypothesis testing]], for example by changing the [[level of significance]] after seeing the [[p-value]]. This [xkcd](https://www.explainxkcd.com/wiki/index.php/882:_Significant) comic illustrates the issue.  Problematic by itself, when combined with news reporting that favors positive results (and often is not interested in printing corrections following subsequent studies), p-hacking causes confusion in the public and erodes confidence in science. See the Washington Post expose on [how and why a journalist tricked news outlets into thinking chocolate makes you thin](https://www.washingtonpost.com/news/morning-mix/wp/2015/05/28/how-and-why-a-journalist-tricked-news-outlets-into-thinking-chocolate-makes-you-thin/) for a canonical example. If you must perform simultaneous tests with multiple predictors, adjust for the [[multiple comparison problem]].