Skip to main content
. Author manuscript; available in PMC: 2016 Jan 31.
Published in final edited form as: Arch Sex Behav. 2014 Mar 27;44(2):399–409. doi: 10.1007/s10508-014-0279-1

Table 3.

Model Comparison and Fit Indices

χ2 Δ in χ2 χ2/df RMSEA CFI TLI
Value (df) p-value Value (df) p-value
Model 1 6.22 (4) 0.18 -- -- 1.56 0.03 1.00 0.99
Model 2 7.67 (8) 0.47 1.46 (4) .80<p<.90 0.96 0.00 1.00 1.00
Model 3 12.38 (11) 0.34 4.71 (3) .10<p<.20 1.13 0.01 1.00 1.00
Model 4 139.10 (69) 0.00 -- -- 2.02 0.04 0.95 0.92
Model 5 136.60 (76) 0.00 5.27 (7) 0.628 1.80 0.03 0.96 0.94

Notes: Model 1 is a completely unconstrained multi-group CFA model. Model 2 adds the loadings of the indicators of the latent factor constrained across groups and intercepts freely estimated. Model 3 restricted the intercepts and allowed the erection function and orgasmic function intercepts to be freely estimated. Model 4 used Model 3 to create a structural model, in which parameters were unconstrained across groups. Model 5 was the final structural equation model, with all structural regression coefficients constrained across groups, except for stimulant use and CD4 cell count, which were set to zero in the HIV-negative group. RMSEA = root mean square error of approximation; CFI = Comparative Fit Index (CFI); TLI = Tucker-Lewis Index.