Table 2 Model fit for multiple group models and measurement invariance comparisons across countries.

From: Psychometric functioning, measurement invariance, and external associations of the Relationship Assessment Scale in a sample of Polish Adults

Model RAS (7 items)

chi2 (df)

RMSEA [90% CI]

SRMR

CFI

TLI

Δ chi2 S-B

Δ CFI

Δ RMSEA

Decision

One factor pooled sample (n = 1635)

187.656*** (14)

0.052 [0.046, 0.059]

0.034

0.998

0.997

Pass

Country

Baseline Poland sample (n = 733)

109.252*** (14)

0.058 [0.048, 0.069]

0.032

0.998

0.998

Pass

Baseline Hungary sample (n = 703)

119.282*** (14)

0.063 [0.053, 0.074]

0.037

0.998

0.996

Pass

Baseline U.S. sample (n = 200)

57.521*** (14)

0.083 [0.061, 0.106]

0.049

0.997

0.995

Pass

Configural invariance

279.611*** (42)

0.064 [0.057, 0.071]

0.036

0.998

0.997

Pass

Metric invariance

357.430*** (54)

0.081 [0.073, 0.089]

0.045

0.996

0.995

79.586**(12)

− 0.002

0.017

Pass*

Scalar invariance

437.500*** (94)

0.063 [0.057, 0.069]

0.057

0.995

0.997

57.142*(40)

− 0.001

− 0.018

Pass

Country, Monte Carlo sampling (n = 200 for each group, number of samples = 100)

Configural invariance

240.057*** (42)

0.077 [0.068, 0.087]

0.041

0.997

0.996

Pass

Metric invariance

330.785*** (54)

0.100 [0.090, 0.111]

0.054

0.994

0.993

90.72***(12)

− 0.003

0.023

Pass*

Scalar invariance

245.001*** (94)

0.077 [0.068, 0.086]

0.116

0.994

0.996

5.91 (40)

0

− 0.023

Pass

  1. Differences between groups are on the loadings of item RAS1: λ PL = 0.887 ± 0.09, λ HU = 0.889 ± 0.09, λ US = 0.865 ± 0.09. After relax this parameter constraint invariance was achieved: Δ RMSEA = 0.009 < 0.015 and Δ CFI = -0.001 < 0.010. * based on SRMR < 0.06, because CFI and RMSEA indices are inconsistent.
  2. CFI comparative fit index, TLI Tucker–Lewis index, RMSEA root mean square error of approximation, CI confidence interval, df degrees of freedom, SRMR squared root mean residuals, S-B Satorra–Bentler correction, Δ comparisons of nested model: configural-metric, metric-scalar.