Skip to main content

Table 10 Sensitivity analyses: model fit in separate groups regarding interview mode and language

From: Measurement invariance testing of the PHQ-9 in a multi-ethnic population in Europe: the HELIUS study

 

Free parametersa

χ2 (df)

RMSEA

CFI

Mode effect, total sample:

Interviewer (N = 6097)

33

668.076*

0.066 (0.062–0.071)

0.988

Internet (N = 9904)

33

944.605*

0.062 (0.059–0.066)

0.987

Paper (N = 7181)

33

723.689*

0.064 (0.060–0.068)

0.990

Turks onlya:

Dutch (N = 1874)

33

276.133*

0.075 (0.067–0.083)

0.988

Turkish (N = 894)

33

157.153*

0.079 (0.067–0.091)

0.989

Ghanaians onlya:

Dutch (N = 456)

33

53.541*

0.052 (0.033–0.071)

0.988

English (N = 1581)

33

130.249*

0.053 (0.044–0.062)

0.989

  1. aNumbers slightly differ from the sample size reported in Table 2 since we excluded those for which the questionnaire language was uncertain
  2. * Significant χ2 test or χ2 difference test (P < .001) (compared to the reference model)