STATISTICS CASE: INTERPRETATION OF MULTIPLE REGRESSION
One hypothesis was tested in this case. This hypothesis was as follows: The value of a subject's CFINCONT score (dependent variable) will vary in relation to both a subject's CCOMMUN score (independent variable) and a subject's GENDER (independent variable). The null hypothesis was that variations in a subject's CFINCONT score would be unrelated to either variations in a subject's CCOMMUN score or a subject's GENDER. The rejection criterion for the null was a p<.05 level of probability.
Two primary assumptions were made in relation to this hypothesis test. The first assumption was that a normal distribution characterized the population. The second assumption was that the sample was representative of the population.
The purpose of this test was to determine the strength of the independent variables (CCOMMUN scores and GENDER) as predictors of CFINCONT scores. The hypothesis posited that a combination of subjects' CFINCONT scores and GENDER would permit the prediction of subjects' CFINCONT scores.
Definitions of the Results Categories
Regression, correlation, and analysis of variance (ANOVA) results were included in the output of the multiple regression applied to the data to test the hypothesis. The regression coefficient for the independent variable GENDER was negative, which indicated that the value of a subject's CFINCONT score varied inversely with a subject's GENDER. As the "male" classification was valued as "1" and the "female" classification was valued as "2" in this analysis, the regression coefficient for GENDER indicated that lower CFINCONT scores were associated with the "female" classification and higher scores were associated with the "male" classification. The t value of the GENDER variable indicated that the regression coefficient was statistically significant at p.002
...