|
|
|
|
|
|
Common tests
On the preceding page, we described the concept of "test".
Below is a short list of some of the most popular tests. The distinction between "parametric" and "non-parametric" tests is adressed here.
|
SOME COMMON TESTS |
|||
|
|
TEST |
PURPOSE |
|
|
Parametric |
|
|
|
|
|
t tests |
Testing mean(s) of 1 or 2 normal populations. |
|
|
|
ANOVA |
Testing equality of means of more than 2 normal populations. |
|
|
|
Hotelling's T ² test |
Testing equality of means of 2 multinormal populations. |
|
|
|
Bartlett's test |
Testing equality of variances of more than 2 normal populations. |
|
|
|
Dunnett's test |
post hoc test for comparing groups to a reference group. |
|
|
|
Newman-Keuls test |
post hoc test for fair pairwise multiple comparisons. |
|
|
|
Fisher-Irwin test |
Testing the identity of 2 Bernoulli populations. |
|
|
|
|
|
|
|
Non parametric |
|
|
|
|
|
Anderson-Darling |
Goodness of fit for continuous distributions. |
|
|
|
Chi-square tests |
|
|
|
|
|
Goodness of fit. |
|
|
|
|
Identity of probability distributions. |
|
|
|
|
Independence of two categorical variables. |
|
|
|
|
Symmetry of the joint probability distribution of two categorical variables. |
|
|
|
|
Identity of marginal distributions. |
|
|
|
Cramér-vonMises |
Goodness of fit for continuous distributions. |
|
|
|
Fisher's exact test |
Independence of two dichotomous variables. |
|
|
|
Kolmogorov-Smirnov |
Goodness of fit for continuous distributions. |
|
|
|
Mann-Whitney |
Identity of 2 probability distributions. |
|
|
|
Kruskal-Wallis |
Identity of 3 or more probability distributions. |
|
|
|
Friedman |
Identity of 3 or more probability distributions (matched samples) |
|
|
|
McNemar |
Symmetry of the joint probability distribution of two dichotomous variables. |
|
See also two generic methods for building tests (with many examples) :
* The Neyman-Pearson lemma
.
* Likelihood ratio tests .
|
|
|
Download this Glossary |
|
|
|
|
|
