Chapter 22.

Example data

Throughout the chapter, a major data set is used to illustrate the procedures for evaluating ANOVA assumptions and setting up contrast coding. The data are based on a study investigating the effect of background sound on learning. It is predicted that students will learn most effectively with a constant background sound, as opposed to an unpredictable sound or no sound at all. In order to test this, participants were asked to study a passage of text for 30 minutes. Then they were given 20 multiple choice questions to answer based on the material. Each participant was randomly assigned to one of the three studying groups. One group had background sound at a constant volume. A second group studied with noise that changes volume periodically in the background. A third group had no sound at all.

The data are stored in an SPSS data file dist.sav. A description of each variable follows. numright shows the number of correct responses over the 20 items. condit is the grouping variable with the three experimental conditions. Other quantitative variables, normal, skewed, uniform, and hetervar are variables simulated from distributions that are normal, positively skewed, uniform, and heterogeneous in variances, respectively. Further, independ and depend are variables simulated from distributions with and without independent observations.

Practice data

Another data set is provided for those who want to practice or experiment with the techniques described in this chapter. The data are revised from the High School and Beyond Study (Glass & Hopkins, 1996) and are stored in an SPSS data file HSB.sav. Achievement (reading, math and science scores) and demographic (sex, SES) information was recorded for a national representative sample of 600 high school seniors. hetervar and nonnormal are variables simulated from distributions with unequal variances and non-normal distributions respectively. A description of each variable is as follows.