Group differences broadly refer to statistics which quantify the differences between two or more subpopulations.
Group differences broadly refer to statistics which quantify the differences between two or more subpopulations. Examples include
- medical trials where one group receives a drug and the others does not
- the 90-10 earnings differential within certain education brackets
- crime rates across social groups and districts within a city
A large array of techniques is available for assessing such differences. Common between-group tests are the Chow test, matching methods, difference-in-differences, $\chi^2$ and t-tests or Wilks' lambda. Sometimes also the within-group properties are of interest or because we may want to compare the variability within groups across groups. For instance, ANOVA can be used to separate the within- and between-variability of group data.
With observational data a common concern is that there may be unobservable factors which have led individuals to self-select into certain groups. Therefore group differences are mostly descriptive unless the data comes from a randomized experiment or methods for causal inference like matching of difference-in-differences are applied.