Statistical significance is conventionally assessed at a p-value below zero point zero five, meaning the observed result would occur by chance less than five percent of the time if no true association existed.
It does not measure the size or importance of an effect.
Alternative Names:
P-Value, Significance Testing
Why it Matters?
Significance and causation are frequently conflated in expert testimony, though a significant result establishes only that chance is an unlikely explanation, not that the association is causal or that confounding was excluded. The multiple comparisons problem is the related issue, since testing many associations guarantees some significant results by chance alone. Studies reporting one significant finding among dozens tested warrant that scrutiny.
Frequently Confused with
Related terms
Frequently asked questions
Does statistical significance establish causation?
What is the multiple comparisons problem?
