A statistical
manipulation where instead of tweaking data to achieve p-values below 0.05, the researcher sets an
artificially high significance threshold (e.g., p < 0.01) and then selectively reports only results that meet that stricter cut‑off, ignoring equally valid findings that fall just above. The goal is to appear more rigorous while actually discarding meaningful results. Reverse p‑hacking can also involve running multiple analyses and reporting only those that fail to reach significance, to support a null
hypothesis. It’s a form of hidden selective reporting that distorts the evidence base, often used to dismiss real effects.
Reverse P-Hacking *Example: “He ran five
different models, then
reported only the one where p = 0.08 as ‘
insignificant’ – reverse p‑hacking, using a stricter threshold to bury a genuine trend.”*