**4.2 Predictors of employees' return**

Previous research shows that mentioned seven characteristics differentiate home-centred women from the rest employees. However, can we rely on them to predict the return of home-centred women to the same employer? The binary logistic regression model was statistically significant, χ2 = 184,695, p < 0,00. Hosmer and Lemeshow test also confirmed model significance (χ2 = 1,799, p = 0,97). For the regression, we narrowed the sample to 281 responses, which referred to this cluster.

The model explained 69,1% (Nagelkerke R2) of the variance in employees' intention to return to the same employer, and correctly classified 84,7% of cases (**Tables 7** and **8**). In line with the expectations, staying in touch with employees with childcare commitments during their leave (increased chances of return by 2,202 times) and showing care about their return (expected increase of returns by 5,339 times) are good predictors of actual return by employees to work. Not mother's age, neither city or children age could predict employees' return of home-centred mothers.
