Table 2.
Model | Panel A: TNRI Warmth
|
Panel B: TNRI Conflict
|
||||||||
---|---|---|---|---|---|---|---|---|---|---|
χ2(df) |
χ2 Difference Test
|
RMSEA | χ2(df) |
χ2 Difference Test
|
RMSEA | |||||
Compare | χ2(df) | p | Compare | χ2(df) | p | |||||
|
|
|||||||||
(A) Invariance Across Time (Years 1 to 4) | ||||||||||
Configural | 2045.56(1171) | 0.032 | 419.55(198) | 0.039 | ||||||
Metric | 2131.29(1207) | Configural | 85.734(36) | < 0.001 | 0.032 | 462.95(213) | Configural | 43.40(15) | < 0.001 | 0.040 |
Scalar | 2229.69(1246) | Metric | 98.401(39) | < 0.001 | 0.033 | 568.86(231) | Metric | 105.90(18) | < 0.001 | 0.044 |
|
|
|||||||||
(B) Invariance Across Students' Gender | ||||||||||
Year 1 | ||||||||||
Configural | 349.81(118) | 0.077 | 35.54 (12) | 0.076 | ||||||
Metric | 370.36(130) | Configural | 20.55(12) | 0.06 | 0.075 | 39.48(17) | Configural | 3.94(5) | 0.56 | 0.062 |
Scalar | 376.61(142) | Metric | 6.24(12) | 0.90 | 0.071 | 47.37(22) | Metric | 7.89(5) | 0.16 | 0.058 |
Year 2 | ||||||||||
Configural | 325.29(118) | 0.078 | 25.15(12) | 0.060 | ||||||
Metric | 342.47(130) | Configural | 17.18(12) | 0.14 | 0.075 | 29.96(17) | Configural | 4.82(5) | 0.44 | 0.050 |
Scalar | 365.22(142) | Metric | 22.74(12) | 0.03 | 0.074 | 43.08(22) | Metric | 13.11(5) | 0.02 | 0.056 |
Year 3 | ||||||||||
Configural | 248.28(118) | 0.064 | 21.93(12) | 0.055 | ||||||
Metric | 267.46(130) | Configural | 19.18(12) | 0.08 | 0.063 | 22.95(17) | Configural | 1.02(5) | 0.96 | 0.036 |
Scalar | 285.39(142) | Metric | 17.92(12) | 0.12 | 0.061 | 31.38(22) | Metric | 8.42(5) | 0.13 | 0.040 |
Year 4 | ||||||||||
Configural | 314.85(118) | 0.080 | 25.62(12) | 0.066 | ||||||
Metric | 334.45(130) | Configural | 19.59(12) | 0.08 | 0.078 | 32.62(17) | Configural | 6.99(5) | 0.22 | 0.059 |
Scalar | 355.31(142) | Metric | 20.86(12) | 0.05 | 0.076 | 43.32(22) | Metric | 10.70(5) | 0.06 | 0.061 |
Note. RMSEA is root-mean-square error of approximation. Following the sequence of tests of measurement invariance (Meredith, 1993; Millsap & Cham, 2012), we tested configural, metric, and scalar invariance models using χ2 difference test (at α ≤ 0.05) and the examination of significant differences of the RMSEA (change ≤ 0.015; Chen, 2007) between the two nested invariance models.