Figure 4 - available via license: Creative Commons Zero 1.0
Content may be subject to copyright.
Results of coarsened exact matching regressions. a-c. Logistic regression of the odd of surviving in academia. The green (pink) bars indicate the positive (negative) regression coefficients for the corresponding variables on the y-axis. The numbers next to the bars indicate the value of the regression coefficient. The statistical significance of the variables are presented at the top of each value (* p < 0.05; **p < 0.01; ***p < 0.001). The error bars indicate the standard error of each regression coefficient. d-f. Linear regression of the fecundity of G 1 mentees. g-i. Logistic regression of the odds of being in the top 5% in the AAC rank. Note that we show only the statistically significant variables of the regression models after coarsened exact matching the data.
Source publication
Mentoring is a key component of scientific achievements, contributing to overall measures of career success for mentees and mentors. A common success metric in the scientific enterprise is acquiring a large research group, which is believed to indicate excellent mentorship and high-quality research. However, large, competitive groups might also amp...
Contexts in source publication
Context 1
... The group size of a given mentee is defined as the total number of nodes in G 1 , that is the number of mentees that were supervised by the same mentor 5 years before and after the mentee's graduation. For example, in Fig. 1a, the node p 1 is mentored with two other mentees during the five years before and after p 1 's graduation, whereas in Fig. 1b, p 4 is mentored with four other mentees five years before and after p 4 's graduation. The group size is thus 3 for p 1 and 5 for p 4 . Notice that the group size associated with a mentee is fixed, but a mentor can lead a group whose size can change over time and is equal to the number of mentees mentored in any 10 years window. The usage ...
Context 2
... findings are not a trivial consequence of dividing the data into the small groups and big groups, as shown by a null model where we randomize the mentor-mentee relationships while keeping the group size constant. In this null model, we do not see significant differences between mentees from big groups and small groups ( Supplementary Fig. S4). The findings about survival suggest that being mentored in a big scientific group can have long-term career competitive advantages in academic performance, but these are conditional to the lower odds of surviving. ...
Context 3
... (CEM) coupled with regression models to study the relation between scientific group size and predictors of future academic performance. CEM regression consists in running a separate regression model on matched groups of mentees, resulting in a more stringent way of controlling for confounding factors than regression alone (Methods) 50, 51 . In Fig. 4a-c, the logistic regression applied to CEM datasets shows that the most significant variable, with a negative weight, to predict survival is MenteeFromBigGroup variable, confirming the finding that being trained in a big group lowers the odds of future survival in academia. The positive regression coefficient of the variable ...
Context 4
... negative effect on mentee survival in Physics. Also, the regression coefficient of the YearlyPubsOfMentor variable indicates that the mentor's yearly productivity, i.e. the average number of papers published in a year, has a negative effect on the mentee's survival probability. Taken together, a possible explanation for the results observed in Fig. 4a-c is that busy mentors, such as those from big groups and with a high publishing rate, have typically little time to spend on supervising each mentee, affecting their future academic career. The negative association between mentor productivity during the mentee training and mentee success is further confirmed when we study the ...
Context 5
... datasets also to control for confounding factors in the prediction of fecundity and citation performance, and confirm our previous observations: Group size is a significant factor, being positively associated with future fecundity and citation performance, captured by being among the top 5% scientists for yearly citations (Top5%YearlyCitations, Fig. 4d-i). The only exception is Neuroscience, where group size is not significant to predict a top-cited scientist (Fig. 4j). Overall, the regression analysis confirms that, if surviving, a mentee from a big group has long-term competitive advantages compared to small ...
Context 6
... our previous observations: Group size is a significant factor, being positively associated with future fecundity and citation performance, captured by being among the top 5% scientists for yearly citations (Top5%YearlyCitations, Fig. 4d-i). The only exception is Neuroscience, where group size is not significant to predict a top-cited scientist (Fig. 4j). Overall, the regression analysis confirms that, if surviving, a mentee from a big group has long-term competitive advantages compared to small ...
Context 7
... difference between survived mentees and dropepd out mentees (Fig. 5e-g Insets). Mentees working in small groups can receive more even attention from the mentor in the same period because there are fewer trainees. Finally, our regression models have between 66% and 73% prediction accuracy in mentee survival and reveal another main factor (Fig. 4 and Supplementary Table S5, Fig. S15): Surprisingly, the more productive a mentor is, the smaller the probability that their mentee will stay in academia.Taken together, our findings quantitatively support the hypothesis that the attention received from the mentor plays a key role for the higher survival rate and success of mentees in ...
