logical trauma exhibits a statistically significant negative effect. Model II
includes control variables and accounts for 8.9 percent of the variance. Once
controls are entered into the equation, the effects for poverty and the urban
poor are no longer evident. However, trauma and attendance demonstrate a
negative relationship net of the effect of covariates. That is, respondents who
experience greater psychological trauma attend religious services at lower
rates. The analysis also shows that older and more educated respondents
report more frequent attendance. Southern residents and African American
respondents are more likely to attend religious services than their counter-
parts. Married respondents, respondents with children under the age of eigh-
teen living at home, and females are also more likely to attend. The coefficient
for the urban scale shows that respondents living in more urban areas attend
services less than those in small communities.
Table 5 shows that Model I for 1988 is not statistically significant. Hence,
there are no main effects in this model. Once controls are included in Model
II, psychological trauma has a statistically significant effect on attendance.
The economic factors show no significant effect. The effect of control vari-
ables varies from the 1978 model. Age and education remain significant pre-
dictors of attendance while an increase in community size leads to a decrease.
Married and female respondents are also similar to the 1978 model. However,
southern residents are no more likely than their nonsouthern counterparts to
attend. Finally, having children at home is not a significant predictor.
The third set of models in Table five exhibits the main effects model and
the model with controls for 1998. Model I is statistically significant but explains
a small percentage of the overall variance. Like the 1978 model, the main
effects for psychological trauma exhibit a negative relationship with atten-
dance. The coefficients for the economic variables are not statistically significant.
Model II includes covariates. This model shows that the relationship between
trauma and attendance remains statistically significant net the effect of co-
variates in the model. In addition, the economic effect is evident as well.
Respondents with incomes below the poverty line are less likely to attend.
Conversely, the more urban poor are more likely to attend religious services.
Discussion and Conclusion
The analysis generates a few somewhat limited conclusions and raises a num-
ber of theoretical and empirical questions.
Operationalizing the Critical Theory of Religion • 349