O pogoni za wynikiem istotnym statystycznie. Konsekwencje rozpowszechnienia testowania istotności hipotezy zerowej w psychologii

Abstrakt

NHST (null hypothesis significance testing) is the most popular statistical paradigm in psychology. Mistakes in interpretation of its assumptions and their consequences are topic for methodological and statistical discussion for over fifty years. Article presents two problems associated with NHST that are particularly prevalent in psychology: identifying non-significant results with research failure and conducting underpowered research. They can contribute to increase in exploiting questionable research practices in order to obtain desirable, significant outcomes. Three practices: p-hacking, HARKing and selective publishing are described, along with analysis of their impact on replication crisis in psychological science.