What Is a P-Value?
A p-value quantifies how surprising your observed statistic is under the null hypothesis. If the null hypothesis is true, the p-value is the probability of seeing a result as extreme or more extreme than the one observed. Small p-values suggest the data are less compatible with the null model and may justify rejection at a chosen significance level (alpha).
Tail Direction and Formula
For z-tests, p-values are derived from the standard normal cumulative distribution function Φ(z).
- Left-tailed: p = Φ(z)
- Right-tailed: p = 1 − Φ(z)
- Two-tailed: p = 2 × (1 − Φ(|z|))
Tail choice must match your hypothesis statement. If your alternative is directional, use one-tailed. If your alternative is simply “different,” use two-tailed.
Worked Example
Suppose your test yields z = 2.10 with a two-tailed hypothesis.
- Φ(2.10) ≈ 0.9821
- Upper tail = 1 − 0.9821 = 0.0179
- Two-tailed p = 2 × 0.0179 = 0.0358
At alpha = 0.05, p = 0.0358 < 0.05, so you reject the null hypothesis.
How to Interpret P-Values Responsibly
- P-value is not effect size: significance does not imply practical impact.
- P-value is not certainty: it does not give the probability that the null is true.
- Context matters: sample design, assumptions, and multiple testing can shift interpretation.
- Use companion metrics: confidence intervals and effect sizes should be reported together.
Common Alpha Levels
| Alpha (α) | Confidence Equivalent | Typical Use |
|---|---|---|
| 0.10 | 90% | Exploratory analysis |
| 0.05 | 95% | General scientific reporting |
| 0.01 | 99% | High-stakes or strict inference |
Sources and References
- Fisher, R. A. Statistical Methods for Research Workers.
- Wasserstein, R. L., & Lazar, N. A. (2016). The ASA Statement on p-values.
- NIST/SEMATECH e-Handbook of Statistical Methods — Hypothesis Testing.