Exercise 7.15 used regression to examine the effect of three interventions on prisoner MMPI scores. The response variable in the regression was trt.effect, the change in MMPI score (pre-intervention - post-intervention).

Instead of estimating the intervention effect through the change in scores, suppose one is interested in predicting a post-intervention score based on the pre-intervention score for an individual and a particular intervention.

(a) The table below shows an alternative regression model that can be fit to the data. In this model, the response variable is the post-intervention MMPI value (post, not shown explicitly in the table) and the predictors are the pre-intervention score (pre) and the treatment, coded as in problem 7.15. Write the estimated equation for this model.

(b) In this model, describe in general terms the association of the pre-intervention and post-intervention scores.

(c) Does the pre-intervention score appear to be an important predictor of a post intervention score?

(d) What is the predicted post-intervention score for an individual with a pre-intervention score of 73 and receiving no tape after the isolation?

(e) Explain the interpretation of the coefficients for coefficient of treatment Neutral. Is there strong statistical evidence that it is an important predictor?

Data From Exercise 7.15

Data From Exercise 5.35

Data From Exercise 5.47

Transcribed Image Text:

Estimate Std. Error t value Pr(>|t) (Intercept) 28.4053 12.2949 2.31 0.0264 pre 0.6593 0.1628 4.05 0.0002 treatmentNeutral -5.7307 3.5545 -1.61 0.1152 treatment Therapeutic -9.7450 3.5540 -2.74 0.0093