Research Tips

Cross Tabulations and Analysis of Variance

When summarizing survey data, the first things most people look at are means (averages) and frequencies (a count of the number of people who gave each response). These are a good starting point, but important insights will normally be lost if the analysis does not extend beyond these basic descriptive statistics. Often the next logical step is to see whether responses differ according to particular characteristics (such as age, gender, or income). If the responses being examined are qualitative (e.g., where respondents live or their occupational categories), this can be accomplished using cross tabulations (cross tabs). If the responses being examined are quantitative (e.g., how many times a product was purchased or a service used), then the appropriate technique is analysis of variance (ANOVA). In either case, it’s important to use a test statistic to see whether any resulting differences are statistically significant or not. For cross-tabs, the appropriate test statistic is chi-square; for ANOVA, it’s an F-test.