Kurichi and Sonnad reported a table for selecting an appropriate statistical method to analyze clinical data. The authors are from the University of Pennsylvania.
Parameters:
(1) purpose of the study
(2) types of variable
Parameter |
Continuous Normal Distribution |
Nonnormal Distribution or Ordinal |
Dichotomous |
Survival Time |
descriptive |
mean, standard deviation |
mean, inter-quartile range |
proportion |
Kaplan-Meier survival |
compare 2 unpaired groups |
independent samples t-test |
Mann-Whitney test |
chi-square test or Fisher's exact |
log-rank test |
compare 2 paired groups |
paired t-test |
Wilcoxon test |
McNemar's test |
proportional hazards regression |
compare >=3 unmatched groups |
one-way ANOVA |
Kruskal-Wallis test |
chi-square test |
proportional hazards regression |
compare >= 3 matched groups |
repeated measures ANOVA |
Friedman test |
Cochrane Q |
proportional hazards regression |
determine if 2 variables are associated |
Pearson correlation |
Spearman correlation |
chi-square test |
NA |
predict outcomes from 1 independent variable |
simple linear regression |
nonparametric regression |
simple logistic regression |
proportional hazards regression |
predict outcomes from >= 2 independent variables |
multiple linear regression |
robust regression or regression on ranks |
multiple logistic regression |
proportional hazards regression |