Skip to main content

Table 2 Multigroup CFA/SEM measurement invariance configural model fit indices

From: Four-dimensional hierarchical structure of love constructs in a cross-cultural perspective

Model Description χ2 BIC df CFI RMSEA [90% CI] SRMR
1 Baseline 17133.907 281377.878 6606 0.817 . 088 [.086, .090] .064
2 Controlling for response setsa 18778.581 272117.871 7254 0.828 .087 [.086, .089] .062
3 Removing Turkey sample 16856.649 256963.258 6448 0.834 .088 [.086, .089] .060
4 Removing Turkey and Iran samples 14605.926 228251.704 5642 0.840 .088 [.086, .089] .057
5 Removing Turkey, Iran, and Midwest samples 12712.719 214411.701 4836 0.844 .086 [.084, .088] .057
6 Removing Turkey, Iran, Midwest, and Brazil samples 10653.304 182703.263 4030 0.845 .086 [.084, .087] .056
7 Removing items with ≥ 6 sig. differencesb 13040.926 225428.550 4923 0.849 .090 [.088, .091] .062
8 Removing items with ≥ 5 sig. differencesb 10374.024 203776.410 4059 0.866 .088 [.086, .090] .059
9 Model 2 including 2 higher-order latent factorsc 18662.204 272474.078 7209 0.828 .087 [.086, .089] .062
10 Model 7 including 2 higher-order latent factorsc 12921.780 225784.937 4878 0.849 .090 [.088, .092] .062
11 Model 8 including 2 higher-order latent factorsc 10259.014 204132.863 4014 0.866 .089 [.087, .091] .059
12 ESEM model not controlling for response sets 16983.637 278909.593 6966 0.822 .085 [.083, .086] 0.451
13 ESEM model controlling for response sets 19546.090 270829.735 7614 0.816 .088 [.086, .089] 0.346
  1. a Models 3–11 and model 13 control for response sets (acquiescence and extreme responding). b These significant differences were determined based on MIMIC model significant regression coefficients across dummy-coded grouping variables (see Table 1 of Supplementary Materials). c The two higher-order latent factors are (1) feelings towards the partner and (2) feelings about the relationship. χ2 = robust χ2 test of global fit (equivalent to Yuan-Bentler corrected test statistic). Measures of CFI and RMSEA are based on robust estimates for nonnormal data (Brosseau-Liard et al., 2012; Brosseau-Liard & Savalei, 2014)