Test Selections | Biostatistics

Independent variable	Dependent variable	Condition(s)	Test(s)
None	Continuous	One variable	One-sample T-confidence interval
None	Continuous	Two variables Parametric requirements^a Nonparametric requirements^b	Correlation Spearman p
None	Discrete	One variable	Chi-square goodness of fit
		Two variables	Chi-square test of independence
		Two variables, very small data sets	Fisher exact test
Discrete	Continuous	Independent variable (two levels)^c
		Unpaired, parametric requirements^a Unpaired, nonparametric requirements^b Paired, parametric requirements^a Paired, nonparametric requirements^b	Two-sample t test Mann-Whitney U Paired t test Wilcoxon matched-pair test
		Independent variable (two or more levels)^c
		Unpaired, parametric requirements^a Unpaired, nonparametric requirements^b Paired, parametric requirements^a	One-way analysis of variance Kruskal-Wallis Complete randomized block
		More than one independent variable^a	n-way analysis of variance
Discrete	Discrete	Two variables
		Unpaired Unpaired, very small data sets Paired	Chi-square test of independence Fisher exact test McNemar test
		Risk estimates
		Retrospective Prospective	Odds ratio Relative risk ratio
Continuous	Continuous	Two variables^a	Linear regression
Continuous	Continuous	More than one independent variable	Multiple regression

^aParametric elements are those in which a population is normally distributed and the variances are approximately equal.
^b Nonparametric requirements are those that do not require the data to be equally distributed and are appropriate for ordinal dependent variables.
^cLevels refer to the number of possibilities in a discrete variable (e.g., three treatment options for the independent variable would indicate three treatment levels).

Table borrowed from: DeMuth JE. Overview of biostatistics used in clinical research. AJHP. 2008; 66:70-81.

Click here for a detailed description.

Flow chart borrowed from: http://abacus.bates.edu/~ganderso/biology/resources/statistics.html