Statistical Methods for Psychology

From Exhibit 17.4 you can see that the program begins with the saturated model. It

then considers what would happen if each of the two-variable interactions were removed.

We see that if the Verdict 3 Fault interaction were removed the change in the log likeli-

hood chi-square would be 36.990 on 1 df. That would be a significant decrement in the fit

of the model, so we won’t want to drop that. Similarly dropping Verdict 3 Moral would

also lead to a significant decrement. However dropping Fault 3 Moral would only produce

Section 17.7 Deriving Models 651

Exhibit 17.4 Stepwise solution

* * * * * * * *HIERARCHICAL LOG LINEAR* * * * * * * *

DESIGN 1 has generating class

VerdictFaultMoral
Note: For saturated models .500 has been added to all observed cells. This value may be changed
by using the CRITERIA = DELTA subcommand.

Backward Elimination (p = .050) for DESIGN 1 with generating class

Verdict*Fault*Moral

Likelihood ratio chi square = .00000 DF = 0 P =.

If Deleted Simple Effect is DF L.R. Chisq Change Prob Iter

Verdict*Fault*Moral 2 .255 .8801 3

Step 1
The best model has generating class
VerdictFault
VerdictMoral
Fault*Moral
Likelihood ratio chi square = .25546 DF = 2 P = .880

If Deleted Simple Effect is DF L.R. Chisq Change Prob Iter

Verdict*Fault 1 36.990 .0000 2

Verdict*Moral 2 8.406 .0149 2

Fault*Moral 2 2.556 .2786 2

Step 2
The best model has generating class
VerdictFault
VerdictMoral
Likelihood ratio chi square = 2.81175 DF = 4 P = .590

If Deleted Simple Effect is DF L.R. Chisq Change Prob Iter

Verdict*Fault 1 37.351 .0000 2

Verdict*Moral 2 8.768 .0125 2

* * * * * * * *HIERARCHICAL LOG LINEAR* * * * * * * *

The final model has generating class

Verdict*Fault

Verdict*Moral