Electronic statistics textbook banner

Statistical Advisor, How To Compare Groups

General (Nonparametric) Comparisons of Groups

NONPARAMETRICS AND DISTRIBUTIONS: This chapter describes numerous so-called nonparametric tests for comparing groups. Compared to parametric tests (t-test, ANOVA), nonparametric tests are based on less restrictive assumptions about the nature and distribution of the variables in the comparisons. For example, there are several tests for variables containing only rank order information. Such data may arise, if one asks respondents in a consumer survey to rank order their preferences for several competing brands of soap.

Other tests, such as the Kolmogorov-Smirnov two sample test, do not make any assumptions about underlying distributions, and will compare not only the central tendency (e.g., mean, median) of a variable in two groups, but also the general shapes of distributions.

GENERALIZED LINEAR MODELS (GLZ): This chapter describes analysis of variance (ANOVA) like designs, without assuming that the variables in the analysis follow the normal distribution, or that the effect of the categorical predictors is linear in nature. Also, the GLZ chapter refers to maximum likelihood methods instead of least squares estimation.

GRAPHICAL ANALYTIC TECHNIQUES: Graphical analytic techniques include numerous facilities for producing histograms, line graphs, scatterplots, etc. by groups. These graphs offer a wide variety of methods to visualize differences between groups.