learn.howToCalculate
learn.whatIsHeading
Determines if A/B test results are statistically significant or due to chance. Critical for making confident product and marketing decisions.
Trin-for-trin guide
- 1Define null hypothesis and significance level (typically 0.05)
- 2Calculate sample size needed for power
- 3Run A/B test long enough
- 4Calculate p-value using chi-square or t-test
- 5Accept or reject hypothesis
Løste eksempler
Input
A 10%, B 12%
Resultat
95% confidence
Almindelige fejl at undgå
- ✕Stopping test too early when results favor variant
- ✕Multiple testing inflating false positive rate
Ofte stillede spørgsmål
What sample size do I need?
Depends on baseline rate and expected lift; typically 500-2,000 per variant for 80-90% statistical power.
What's statistical significance?
95% confidence the result wasn't due to chance; p-value < 0.05 typically required.