October 2008 Poll - Discussion
The Effects of Type and Demographic Variables
The analyses on the preceding pages looked only at the effects of type on the political orientation measures. As seen in the results from the correlation analysis , however, the various type, demographic, and political orientation measures show multiple correlations. For example, sex is correlated with the T-F preferences (as might be expected) as well as many of the political orientation measures. To examine the effect of type on political orientations while controlling for these multiple correlations, we conducted a multiple regression analysis.
The dependent variable in the multiple regression analysis consisted of a total score constructed from the five political orientation measures. The independent variables included the type dichotomies (E-I, S-N, T-F, and J-P), demographic variables (age and sex) as well as a variable indicating APTi membership (yes or no). All variables were entered into the analysis at the same time.
Results indicated the following significant results:
- Sensing types scored significantly less liberal than did Intuition types.
- Thinking types scored significantly less liberal than did the Feeling types.
- Males scored significantly less liberal than did females.
- The variables explained about 9% of the variance in the liberal-conservative scores.
These results generally support results from previous versions of the Type & Politics Poll. In particular, the S-N preferences consistently have been related to liberal-conservative orientations. The T-F preferences also have shown some relationship, although not as strongly as have the S-N results. A look at the bivariate correlations in this study for the S-N, T-F, and political orientation variables suggests one hypothesis as to why the T-F preferences may have had such a significant effect in this analysis. In particular, the S-N preferences seem to relate more strongly to the social measures while the T-F preferences seem to relate more strongly to the economic measures. The explicit distinction between economic and social measures of liberal-conservative orientations in this study thus may have allowed the T-F effects to play a more significant role. (Note: Separate regression analyses using only the Social Attitudes measure and the Economic Attitudes measure seem to support this conclusion. Results for social are here and results for economic are here ).
The above analysis did not include party identification (Republican, Democrat, and Independent) in the analysis. We thus conducted as second regression analysis which included these variables (results here ). As we have found in the analysis of previous poll responses, the relationships between type preferences and liberal-conservative orientations were not significant when party identification was included. And, as might be expected, Republicans scored significantly more conservative than did Independents and Others whereas Democrats scored more liberal than did Independents and Others. (We did not included the Independents due to linear depedency.)
Note: We also tested for the effect of interaction terms in various regression analyses both with and without demographics included as main effects. No significant interaction results were found.