In a follow up to my previous post, I decided to use the GSS’s logit regression feature to probe the relationship between a set of variables and attitudes toward homosexuality. The columns are the dependent variables, while the rows are the independent ones. I’ve omitted all variables where the beta coefficient is not statistically significant at p-value = 0.05. I’ve also bolded the largest beta in each column.
|Allow Homo. Speak||Allow Homo. Teach||Is Homo. Wrong||Homo. Right To Marry||Allow Homo. Books|
One immediate comment here is that age is not a very strong independent variable when you control for others. In other words, the young are not gay-friendly because they are young, but because other variables (e.g., religiosity) are correlated in a particular manner with youth. Secondly, attitudes toward the Bible (Word of God, Book of Fables, etc.) are excellent predictors of attitudes toward homosexuality. What does that tell us? I think this is one place where the old over-used chestnut that “correlation does not equal causation” holds. What the Bible variable is getting at in my opinion is more the particular subculture or Zeitgeist in which an individual is embedded. This is interpretation is conditional on my assessment that most people who hold the Bible to be the Word of God do not systematically read and memorize the Bible and transform themselves into inference machines whereby all actions and beliefs are reflectively munged through Biblical criteria. They may say that this is what they are doing, but I do not believe that humans are so conscious, reflective or deductive in their thinking.