Ridderprint

Chapter 6 128 models (M3). Consumer Goods’ employees became 1.52 times less likely 14 to leave with every point increase in performance ( B = -0.42, p < .001) whereas Shell’s employees became 1.45 times less likely to leave with every decimal point increase in performance ( B = -3.73, p < .001). Hypotheses 2a and 2b were rejected. Potential ratings did not affect graduate turnover at Consumer Goods ( B = 0.01, p = .933) or Shell ( B = -0.02, p = .817). Although potential related to decreased turnover in the second model at Shell ( B = -0.22, p < .01), this effect disappeared when the other predictors were accounted for. 15 In line with Hypothesis 3a, employees on STIA were 2.47 times less likely to leave Consumer Goods ( B = -0.90, p < .05) and 3.77 times less likely to leave Shell ( B = -1.33, p < .01), as compared to unassigned peers. However, this effect only lasted for the duration of the STIA as the accumulated amount of international experience had no effect at Consumer Goods ( B = 0.02, p = .48) or Shell ( B = -0.05, p = .057). In line with Hypothesis 3b, turnover was 2.29 times more likely among repatriates at Shell ( B = 0.83, p < .01), but this was not found at Consumer Goods, where repatriates were not more or less likely to leave ( B = -0.19, p = .71). 6.6 Discussion We analyzed the voluntary turnover of 9013 graduate recruits for traineeships at two large multinational organizations – Consumer Goods (anonymized) and Shell – using survival analysis. Our results demonstrate that turnover patterns can vary considerably between organizational contexts – both in terms of the absolute numbers, the development over time, and the predictors. At both organizations, Cox proportional hazard regression models with time-varying covariates confirmed that employees’ turnover decreased with the favorability of their performance evaluations. Although voluntary turnover among trainees was less likely during STIA, the amount of international experience accumulated had no effect. Paradoxically, employees at Shell became more likely to leave after repatriating from STIA. Finally, leadership potential assessments did not affect voluntary turnover after controlling for performance evaluations and international experiences. The next section discusses these findings in more detail, along with their implications, limitations, and avenues for future research. 14 1 – log(β) 15 Inherently, unassessed Shell employees demonstrated increased hazard ( B = 1.44, p < .001) because they had to have left early in the timeframe–before their first assessment, within 2 to 4 years after hiring.