No statistically significant difference in chi-square indicates that the more parsimonious model explains the data equally well compared to the more complex model with additional paths (Kline 1998). Additionally, the other fit indices were used to choose the final best fitting model. Results In

Table 1, descriptive statistics, reliabilities R788 molecular weight and inter-correlations among all study variables are presented. As can be seen from the table, the reliabilities were acceptable. Overall variables had test–retest reliabilities of at least .46 (see Fig. 1). The highest test–retest reliabilities resulted for emotional exhaustion and performance-based self-esteem. The internal consistencies for all constructs per measurement wave were satisfactory (α ≥ .85). In order to provide the basis for testing the relations of emotional exhaustion, work–family conflict and performance-based self-esteem over time, we performed a procedure recommended by Brown (2006) to test for longitudinal invariance. Neither of the steps tested and compared to each other resulted in a CFI difference that exceeded .01. Thus, we can assume that the constructs included in this study are invariant over time (Cheung and Rensvold 2002). In accordance

with recommendations from Little and Card (2013), the constraints of weak factorial invariance were maintained for the subsequent testing of our research questions. Table 1 Correlations and descriptive GSK-3 inhibitor statistics   M(SD) 1 2 3 4 5 6 7 8 9 10 1. Age 47.40 (10.05) –                   2. Gender (female) .53 (–)

.01 –                 3. University education .37 (–) −.05* .13* –               4. Having Ureohydrolase children .52 (–) −.30* −.02 .05* –             5. Work–family conflict T1 2.13 (1.04) −.10* .05* .15* .10* –           6. Emotional exhaustion T1 1.63 (1.47) .00 .12* .03 −.01 .49* .87         7. Performance-based self-esteem T1 3.59 (1.44) −.09* .05* .10* .01 .32* .32* .85       8. Work–family conflict T2 2.11 (1.05) −.13* .06* .17* .12* .54* .34* .27* –     9. Emotional exhaustion T2 1.71 (1.46) −.02 .13* .04* −.01 .37* .67* .26* .47* .87   10. Performance-based self-esteem T2 3.31 (1.40) −.11* .06* .13* .04* .30* .28* .66* .31* .28* .87 Listwise; n = 3,387. * p < .05; – not applicable. The scales ranged from 1 to 5 except gender (men = 0 and women = 1), age (in years), university education (which was coded 1 = university education, 0 = lower levels of education) and having children living at home (0 = no. 1 = yes). In the diagonal in italic: Cronbach’s alpha Fig. 1 Reciprocal model (Model 4): standardized coefficients. Notes *p < .05, dotted line for non-significant path, WFC work–family conflict, EE emotional exhaustion, PBS performance-based self-esteem In Table 2, the fit statistics for our four cross-lagged models are shown.