Spurious findings

It is well recognised that data will eventually “confess” if it is interrogated enough. Provided enough variables and tests are deployed, a “statistically significant” result can usually be obtained. Consequentially, many of the “discoveries” in clinical research later transpire to be spurious findi...

Descrizione completa

Dettagli Bibliografici
Autori principali: Cook, J, Ranstam, J
Natura: Journal article
Pubblicazione: Wiley 2016
Descrizione
Riassunto:It is well recognised that data will eventually “confess” if it is interrogated enough. Provided enough variables and tests are deployed, a “statistically significant” result can usually be obtained. Consequentially, many of the “discoveries” in clinical research later transpire to be spurious findings reflecting only chance occurrences and the idiosyncrasies of the dataset and analysis strategies used, and do not reflect a real, let alone clinical useful, relationship.