Details about the factor analysis

The statistical correlation of the seven variables is highly significant (p < .001, two-tailed). The p-value of .001 can be interpreted to mean that there is only one chance in a thousand that the tendency of the variables to increase or decrease together occurred by chance. Factor analysis further evaluates the underlying relational pattern in the data to determine whether the items ‘load’ significantly onto a shared common factor, and it allows use of maximum likelihood estimations. Factor analysis also allows the model to take into account significant correlations between error terms. Overall fit statistics for our analysis demonstrate that the model fits the underlying pattern in the data. The Normed Fit Index (NFI) value is .993 and the Tucker Lewis Index (TLI) value is .961, both indicating a strong fit (values between .90 and .95 are generally acceptable, and above .95 are good); the Root Mean Square of Error Approximation (RMSEA) value is .045, which also indicates a good model fit (values of .050 or less are generally acceptable and scores above .100 indicate a poor fit). The amount of variance explained by the RCN factor for each question ranged between .27 and .53 (R-square values); the seven items each load onto the factor at .50 or stronger. We also analyzed other models that included additional items from the 2006 GAP survey characteristics battery, but found that the model that fit the data was one where the seven items describe in this article loaded onto a single common factor. The characteristics that were included on the 2006 GAP survey that did not load were: devout, tolerant, fanatical, and respectful of women.