hypoth.h provides facilities to test statistical hyptheses. For instance, could the differences between two sample means be explained away as sampling error?
Confidence limit functions
| Confidence Limit | UVS Function |
|---|---|
| xbar, sigma known | confidence_mean_known_sigma |
| xbar, sigma unknown | confidence_mean |
| mu1-mu2, sigma1 and sigma2 known and equal | confidence_delta_mean_known_sigma |
| mu1-mu2, sigma1 and sigma2 unknown | confidence_delta_mean |
| s squared | confidence_variance |
| variance ratio | confidence_variance_ratio |
Critical value functions
Probability of getting test statistic given null hypothesis functions
Generic hypothesis test functions
These last functions are the recommended approach to hypothesis testing. Make up your mind in advance how much alpha risk you are willing to take, do the test, and reject the null hypothesis or not based on the outcome. History has shown that looking at confidence limits or tail probabilities tends to lead one in the direction of chasing after random variation, precisely what hypothesis testing was designed to avoid!
Hypothesis objects for use by functions above
| Hypothesis Test | UVS Class |
|---|---|
| probability of success | HypBinomial |
| difference between two probabilities of success | HypBinomialDifference |
| mean of normal distribution with known sigma | HypMeanKnownSigma |
| difference between means (known sigmas) | HypMeanDifferenceKnownSigma |
| mean of normal distribution with unknown sigma | HypMean |
| mean of variable1 - variable2 (not paired) | HypMeanDifference |
| variance of a distribution | HypVariance |
| ratio of two variances | HypVarianceRatio |