Statistical Forensics

Topics


Statistical forensics refer to statistical techniques that can assess the credibility or likely replicability of scientific studies.


Questionable Methods


A p value of < 0.05 is widely regarded as the necessary threshold in order for a paper to be publishable. However, many researchers have started engaging in p-hacking in order for their results to be publishable. P-hacking is the process of manipulating data to make non-significant results statistically significant. P-hacking is usually not outright fraud but falls within a researcher’s freedom to make choices about the study design. This could include decided what factors to control for and analyzing many measures but reporting only those with a p-value of less than .05.

Resources about p-hacking:

Science isn’t broken. Fivethirtyeight