Overview

Chapter 7 described a range of evaluation goals and methods that are commonly used in usability engineering. However, while we discussed the major concepts of experimental design for example, the concepts of variables, hypotheses, and validity we did not present the methods or techniques of inferential statistics. These methods are required when evaluators wish to claim that the results they have obtained are due to the conditions of the experiment rather than chance variation.

There are many statistical tests that can be used to assess the significance of the patterns observed in usability outcomes. Some common tests are listed in Table A.1, along with sample questions they might be used to answer. For example, a t-test compares two independent data sets to see if their means are different, whereas analysis of variance (ANOVA) compares mean differences for the case of multiple independent variables (each of which may have two or more levels). In linear regression, a mathematical equation is constructed to show whether and to what extent each independent variable contributes to the prediction of a dependent variable.