TABLE 4

CFA for Factorial Invariance Testing

ConstraintsConstruct(s)∆χ2∆dfCFIRMSEA
Configural model ANoneAll28 332.77a2520a0.9050.025
Construct testing (single construct fixed; remaining constructs vary across time)Metric invariance: factor loadings fixed equal across timeCON36.87b6b0.906b0.025b
HYP151.47b8b0.905b0.025b
EMO77.95b10b0.908b0.024b
PEER96.32b10b0.909b0.024b
PRO137.51b10b0.907b0.025b
Scalar: factor thresholdsCON149.06b10b0.905b0.025b
HYP933.41140.9030.025
EMO2226.11180.9000.025
PEER2549.87180.8990.025
PRO297.15b18b0.904b0.025b
Strong: factor loadings and thresholdsCON108.83b16b0.906b0.025b
HYP551.46220.9030.025
EMO734.84280.9030.025
PEER826.81280.9030.025
PRO338.95b28b0.906b0.025b
Model BAll factor loadings; CON and PRO thresholdsAll699.18b72b0.912b0.023b
Model CcAll factor loadings; CON, PRO, and HYP thresholdsAll940.66b86b0.911b0.024b
Model DAll factor loadings; all thresholdsAll22876.521420.7710.037
  • χ2 Estimates were obtained by using DIFFTEST (Mplus, Muthén & Muthén, Los Angeles, CA). N = 16 659. CFA, confirmatory factor analysis; CON, conduct problems; EMO, emotional problems; HYP, hyperactivity; PEER, peer problems; PRO, prosocial.

  • a Configural (baseline) model.

  • b Acceptable change in fit indices (compared with configural model).

  • c Indicates preferred model (model C).