Statistical Methods for Psychology

Exercises 655

Exercises

All of the problems in this chapter will require solution by one or more computer programs.

My answers in the back of the book are based on SPSS GENLOG, and they may differ from

answers you receive if you use a different program.

17.1 Allison (1991) offers an interesting example from a study by Morgan and Techman (1988)

looking at race, gender, and the sexual intercourse for a sample of 15- and 16-year olds. The

data follow.

Intercourse

Race Gender Ye s N o

White Male 43 134

Female 26 149

Black Male 29 23

Female 22 36

What are the possible models that could be hypothesized to underlie the data matrix?

17.2 In Exercise 17.1 Intercourse is the obvious dependent variable. What is the difference

between the roles played by the Gender 3 Race interaction and the Gender 3 Intercourse

interaction?

17.3 Use SPSS HILOGLINEAR to derive the optimal model for the data in Exercise 17.1 using

backward elimination. (Hint: useLogLinear/Model Selectionfrom the menus.) Then re-

produce the results using specific models found in Exercise 17.1. Compute and interpret

coefficients for the most appropriate model.

17.4 Maimaris, Summer, Browning, and Palmer (1994) reported on a study of head injuries in

children and adults involved in bicycle accidents. They broke down the data on the basis of

Age, whether a motor vehicle was involved, whether the rider was wearing a helmet, and

whether there was a head injury. The data appear below.

Motor

Vehicle Helmet Injury Count

Young Ye s

Ye s

Ye s 0

No 9

No

Ye s 8

No 36

No

Ye s

Ye s 0

No 41

l

Key Terms

Log-linear models (Introduction)

Symmetric relationships (Introduction)

Asymmetric relationship (Introduction)

Saturated model (17.1)

Geometric mean (17.2)

(lambda) (17.2)

Saturated model (17.2)

Additive model (17.2)

Conditional odds (17.4)

Odds ratio ( ) (17.4)

Sparse matrices (17.6)

Hierarchical model (17.6)

Defining set (17.6)

Generating class (17.6)

Æ

l

(continues)