Correlation Matrix Input to Proc Reg: Homework, Form C

Moore, C. H., Wuensch, K. L., Hedges, R. M., & Castellow, W. A. (1994: The effects of physical attractiveness and social desirability on judgments regarding a sexual harassment case. Journal of Social Behavior and Personality, 9, 715-730) conducted a mock jury trial in which the social desirability of both the male defendant and the female plaintiff were experimentally manipulated. The defendant was accused of sexually harassing the plaintiff. Jurors who decided that the defendant was guilty also decided how much money to award the plaintiff. The award variable will serve as our dependent variable in this assignment. Because this variable had considerable positive skewness it has been subjected to a square root transformation, so the unit of measure for our dependent variable is $**.5. We shall use the gender ('1' coded male, '2' coded female) of the juror as one of our predictor variables (it is perfectly acceptable to use a dichotomous variable as a predictor in multiple regression analysis). Our second predictor will be the rating the juror gave the defendant when asked to rate him on a 9-point scale ranging from cruel (1) to kind (9). Our third will be the rating of his sensitivity from insensitive (1) to sensitive (9). Here are the summary statistics.

Variable / N / Mean / Std Dev
sr-AWARD / 130 / 302.5242 / 156.2059
GENDER / 130 / 1.5615 / 0.4981
KIND / 130 / 3.9462 / 1.8978
SENSITIV / 130 / 3.3385 / 2.0252

Pearson Correlation Coefficients : N = 130

sr-AWARD / GENDER / KIND / SENSITIV
sr-AWARD / 1.00000 / 0.09878 / -0.39079 / -0.31718
GENDER / 0.09878 / 1.00000 / -0.06617 / 0.00225
KIND / -0.39079 / -0.06617 / 1.00000 / 0.63808
SENSITIV / -0.31718 / 0.00225 / 0.63808 / 1.00000

Construct a TYPE=CORR data set for use with PROC REG. Employ PROC REG to obtain a complete analysis, including beta weights, semipartial correlation coefficients, and tests of the significance of the unique contribution of each predictor. As before, the (transformed) award recommended for the plaintiff is your dependent variable.

In the table below, enter the appropriate values in each cell which I have shaded. For each zero-order correlation coefficient and beta weight, add an asterisk if it is significant at the .05 level.

Table 1
Awards Related to Gender and Jurors’ Ratings of the Sensitivity and Kindness of the Defendant (N = 130)
Zero-Order r /  / sr2 / b / VIF
Variable / Sensitivity / Kindness / Gender / sr-Award
Sensitivity
Kindness
Gender
Intercept =
Mean
SD / R2 =

Note. Gender was coded 1 for male, 2 for female. A square root transformation was applied to the recommend awards to reduce positive skewness.

*p < .05

You should report a 90% confidence interval for the R2. You can use myConf-Interval-R2-Regr program to obtain that CI. To test the significance of the zero-order correlations, use an Internet calculator, such as that at Vassar.

Write, in a Word document, an APA-style summary/interpretation of your results, after reading my document Presenting the Results of a Multiple Regression Analysis. Grade your own paper (or get a classmate to critique it for you) -- I'll grade one later. Do be prepared to go over your solutions in class.