Type I and Type II Errors¶
Formula¶
\[
\alpha = P(\text{reject } H_0 \mid H_0 \text{ true}),\qquad
\beta = P(\text{fail to reject } H_0 \mid H_1 \text{ true})
\]
Parameters¶
- \(\alpha\): Type I error rate (false positive rate under \(H_0\))
- \(\beta\): Type II error rate (false negative rate under \(H_1\))
What it means¶
Type I error is a false positive; Type II error is a false negative.
What it's used for¶
- Designing and evaluating hypothesis tests.
- Understanding tradeoffs between sensitivity and false alarms.
Key properties¶
- Power equals \(1-\beta\).
- For fixed sample size, lowering \(\alpha\) often increases \(\beta\) (tradeoff).
Common gotchas¶
- Error rates depend on the test procedure and assumptions.
- Confusing \(\alpha\) with the p-value from one experiment.
Example¶
In medical screening, a Type I error may wrongly flag a healthy patient, while Type II misses a true condition.
How to Compute (Pseudocode)¶
Input: hypothesis test procedure and null/alternative scenarios
Output: Type I/Type II error rates (alpha, beta) and power
specify the rejection rule of the test
compute alpha under the null model (false-positive probability)
compute beta under the alternative model (false-negative probability)
compute power <- 1 - beta
return alpha, beta, power
Complexity¶
- Time: Depends on whether error rates are derived analytically or estimated by simulation; simulation cost scales with the number of simulated trials and test-evaluation cost
- Space: Depends on whether simulated outcomes are stored or streamed into summary counts
- Assumptions: Error rates are properties of a test procedure under specified data-generating models, not one-off experimental outcomes