May 23, 2012

Russians Back Protests, Political Freedoms

Appendix

A logistic regression analysis was conducted on the weighted full sample (n=1,000) to determine the independent impact of each of a series of factors on likelihood to favor Putin. The dependent variable was favor Putin (coded as 1=somewhat or very favorable, 0=somewhat or very unfavorable). The independent or predictor variables were recoded as dummy variables (0 or 1) based on the standard analytical groups used in the bivariate analysis.

The importance of democratic rights variable is an additive scale of the number of items a respondent names as very important among free speech, honest elections, fair judiciary, civilian-controlled military, uncensored media, and freedom of religion. The scale which ran from 0 to 6 was rescaled to run from 0 to 1, and reversed so that 1 = no items named as very important and 0 = all six items named as very important.

The graphic (see Chapter 4) shows the difference in predicted probabilities for the following groups:

The model also controlled for gender, age, and education, all recoded as dummy variables. The likelihood ratio chi-square for the model is χ2=94.3, p=.000.