Sociology 405/805

Assignment 6 – Multiple Regression


Due Friday, March 24, 2000


Use the 703.sav data file to obtain a multiple regression with V4, “Tax laws and job benefits should recognize gay and lesbian couples as married,” as the dependent variable.  This question is coded from 1 (strongly disagree) to 5 (strongly agree).  Among the attitude variables in question 15 of the Survey of Student Attitudes and Experiences, this is the one that has the greatest variability.


In carrying out this regression, select at least three independent variables that you hypothesize might affect V4 and include these in the equation.  You might use the means procedure to identify variables that have somewhat different mean values of V4 for different categories of these independent variables.  However, the independent variables should also be selected because they might reasonably be expected to have an influence on V4.  For example, religious or spiritual values (question 41) might be expected to have a greater connection than might attitude toward personal safety on campus (question 19).  That is, use some sociologically reasonable ad hoc hypothesizing to select independent variables.


If some of the independent variables have only a nominal scale of measurement, you will have to construct dummy variables for the various categories of these, so that you may ultimately have more than three independent variables.  (Remember that if you do this, there must be one dummy variable, with values 0 and 1, for each category of the nominal scale, but with one less such variable than the number of categories).  If some of the variables are measured on an ordinal scale, they may be included as independent variables.  However, if they are really representing interval level measurements, but coded only at the ordinal level, convert them to interval level by recoding (e.g., code income as 1=10, 2=30, etc. so the new codes represent income in thousands of dollars yearly income). 


You may want to provide more than one regression, showing what you regard as less and more successful regressions.


For one of the regression equations outline and comment on the following in words:

·         The hypotheses.

·         The test of significance for each regression coefficient and for the equation as a whole.

·         How well the regression equation fits the data.

·         Your overall interpretation of the results.

·         Any violation of assumptions that may exist. 

·         Also describe any transformations of variables – e.g. how you obtained the dummy variables or how the variables were recoded.


If you have more than one regression equation, summarize the results of the other equations in shorter paragraphs.