UPDATE (3/19/2013): I have updated the tables and figures below to incorporate the newly available GSS 2012 data here. While the analysis below is generally still applicable, the models have changed.
--
In the last post, I laid out the data. Here, I'll present some results. The analysis is based on multinomial logit regression. The dependent variable is Supernatural Disposition (both/religious not spiritual/spiritual not religious/neither). The independent variables (IV) of primary interest, which I will take up in a later post, are measures of attitudes regarding whether other people can be trusted, take advantage, look out for themselves, and a measure of whether the respondent has been unemployed in the last ten years. I control for year, gender, race, education, and income. Here are the various models with the IV p-values:
Income is insignificant at the .05-level and is dropped from subsequent models. While the "people take advantage" measure loses significance when the other measures of alienation are included, it is significant when the other measures are not included, and the alienation measures are jointly significant so all four are retained in the final model. All of the results I present below and in future posts (unless otherwise noted) are from model 8 above and hold all other variables at their means. Here are the overall probabilities for each Supernatural Disposition category:
Respondents are 54% likely to answer that they are both religious and spiritual, 24% likely to answer that they are neither religious nor spiritual, 15% likely to answer that they are spiritual but not religious, and 7% likely to answer that they are religious but not spiritual. The probability of each Supernatural Disposition category has changed over time:
The probability of identifying as both religious and spiritual has remained remarkably consistent between 1998 and 2010. The probabilities of identifying as neither religious nor spiritual and religious but not spiritual have both declined over the same period, dropping about 3 and 2 percentage points respectively. The probability of identifying as spiritual but not religious, however, has increased over the period by nearly 5 percentage points.
More analysis to come.
--
In the last post, I laid out the data. Here, I'll present some results. The analysis is based on multinomial logit regression. The dependent variable is Supernatural Disposition (both/religious not spiritual/spiritual not religious/neither). The independent variables (IV) of primary interest, which I will take up in a later post, are measures of attitudes regarding whether other people can be trusted, take advantage, look out for themselves, and a measure of whether the respondent has been unemployed in the last ten years. I control for year, gender, race, education, and income. Here are the various models with the IV p-values:
Income is insignificant at the .05-level and is dropped from subsequent models. While the "people take advantage" measure loses significance when the other measures of alienation are included, it is significant when the other measures are not included, and the alienation measures are jointly significant so all four are retained in the final model. All of the results I present below and in future posts (unless otherwise noted) are from model 8 above and hold all other variables at their means. Here are the overall probabilities for each Supernatural Disposition category:
Respondents are 54% likely to answer that they are both religious and spiritual, 24% likely to answer that they are neither religious nor spiritual, 15% likely to answer that they are spiritual but not religious, and 7% likely to answer that they are religious but not spiritual. The probability of each Supernatural Disposition category has changed over time:
The probability of identifying as both religious and spiritual has remained remarkably consistent between 1998 and 2010. The probabilities of identifying as neither religious nor spiritual and religious but not spiritual have both declined over the same period, dropping about 3 and 2 percentage points respectively. The probability of identifying as spiritual but not religious, however, has increased over the period by nearly 5 percentage points.
More analysis to come.
No comments:
Post a Comment