Ridderprint

Paradoxes in global talent pipelines 127 Table 6.1. Assymetrical correlation matrix for Shell (above diagonal) and Consumer Goods (under diagonal). 1 2 3 4 5 6 7 1. Voluntary turnover a .00 -.14 *** -.06 *** -.12 *** -.02 -.02 2. Gender: Male .00 .05 *** .08 *** .03 ** -.02 .03 * 3. Tenure b -.06 ** -.05 * .04 ** .36 *** .07 *** .26 *** 4. Age b .06 ** .02 .03 -.04 ** -.06 *** -.06 *** 5. Performance c -.09 *** .05 * .21 *** -.03 .44 *** .18 *** 6. Potential c -.07 *** -.02 .14 *** .04' .28 *** .14 *** 7. International exp. a -.06 ** .03 .23 *** -.01 .16 *** .22 *** Note: Consumer Goods ( N = 2532) under diagonal, Shell ( N = 6437) above diagonal. a at final observation. b band 18-29 = 0, band 30-39 = 1, at Consumer Goods; continuous at Shell. c average score during observational timeframe. * p < .05 ** p < .01 *** p < .001 6.5.1 Survival and Hazard Curves Figure 6.1 displays the estimated survival over the course of the observational timeframes. Graduate trainees left Consumer Goods significantly more often and quickly than Shell ( χ 2 = 374.13, p < .001). Five years in, an estimated 45.8% of graduate recruits would have resigned at Consumer Goods compared to 14.7% at Shell. Figure 6.1. Estimated survival curves including 95% confidence interval. 6.5.2 Cox Proportional Hazard Regression Tables 6.2 and 6.3 display the nested Cox proportional hazard regression models for respectively Consumer Goods and Shell. In line with Hypothesis 1, favorable performance evaluation had a strong negative influence on the voluntary turnover risk in the full