Peeking at a test daily inflates your false-positive rate from 5% to ~30%.

Peeking at a test daily inflates your false-positive rate from 5% to ~30%.
The mistake: stopping the moment p < 0.05 appears. With one look you control error at 5%; with daily looks over two weeks you've taken ~14 shots at a 5% target.
— Fix: fix sample size in advance, or use sequential testing (mSPRT, group-sequential boundaries) that budgets the looks.
— Caveat: "reached significance" without a pre-set stopping rule is not significance.
Read the number, not the story. [error rate 5%→30% · 14 looks]