The publication of the Swedish Medical Research Council's analysis of its grant selection scheme (Wenneras and Wold, 1997) came as a wake‐up call for many other funding agencies in the life sciences. As it showed an inherent discrimination against women applicants, many of those responsible for giving grants to young scientists became worried. It was indeed appalling to read that women in the Swedish system needed 2.5 times more publications to be successful, but the question was asked whether this was an isolated case? Furthermore, how would other funding programmes fare if subjected to the same scrutiny as the Swedish Medical Research Council?
When the success rate is calculated for the spring and autumn session for the years 1996–2001, the female applicants were, on average, 20% less successful than the males
Such questions triggered an analysis of the EMBO Long Term Fellowship granting scheme in order to find out whether there is any similar discrimination within this programme. The troubling outcome was that, at face value, there does seem to be a bias. Further analysis of the data, however, could also point to a slightly lower quality of the female applicants when judged by certain criteria. Figure 1 illustrates the potential problem. When the success rate is calculated for the spring and autumn session for the years 1996–2001, the female applicants were, on average, 20% less successful than the males. Such a consistent result suggests that there may be a systematic bias against women scientists in the decision‐making process.
To discover the source of this bias, each step in the selection process was thoroughly examined. Here, the cohort from spring 1998 was chosen for analysis since this group displayed one of the largest discrepancies in success rate between the sexes. Overall, 372 people—221 men and 151 women—applied for an EMBO fellowship, of which 74 were chosen. All applicants were of the same age group and there was no obvious difference in the quality of the laboratory they chose for their fellowship. Of the successful applicants 52 men obtained a fellowship and 22 women, thus the success rate of men was 23.5% and that of women 14.6%.
In all cases, the committee members gave a higher average score to the male cohort than to the female
The Swedish study pointed out that female scientists needed more papers of higher quality to be successful. This appeared to be the case for the EMBO scheme too. The average number of papers of all female applicants was 4.9 and that of all males 5.4. However, the successful women had published 7.1 papers and the men 5.8. Therefore, it could be concluded that for a woman to be successful, she needed to have published more papers than her male counterpart. But closer examination of the quality of the publications—judged by their total impact factor—gave a further perspective. The average male applicant had papers of a higher total impact factor (15) compared with females (13.2) and this was mirrored in the successful males (28.8) compared with the successful females (24.2). Equally, the males needed slightly more first author papers—3.4 versus 3.1—of a higher impact factor (26.9 compared with 24.0) to receive a fellowship. One could conclude, therefore, that by this criterion it was actually more difficult for men to receive an award.
Obviously, the bias is not in the number of publications needed but must be hidden in the selection process itself. Applicants are first interviewed by an EMBO member who is then asked to submit a report. Given that almost 90% of EMBO members are male, it might seem plausible that this was the source of the bias. But this was not the case with both sexes receiving equally positive assessments.
The fact that the successful females had a higher average number of papers may indicate that their results have been parcelled into a larger number of less consequential papers
These reports, together with all the information provided by the applicant, are then sent to a ten‐person committee. The members of the committee independently score each application and these figures are tabulated. At the selection meeting, these data are then discussed among the committee, which can result in minor changes made to the scores. Although the committee knows the personal details of each applicant, those discussions are never about criteria other than their scientific quality. In all cases—and there were two females, including the chair, on the committee—the committee members gave a higher average score to the male cohort than to the female (Figure 2).
The challenge is to try to understand why each member of the committee independently reached the same conclusion, i.e. that males were on average more worthy of a fellowship than females. There are two possible explanations here. The first is that the members of the committee all carry an inherent anti‐female bias. This cannot be discounted and, if it is true, it must be widespread as the same results have been found throughout the whole of the 1996–2001 period, despite a rapid turnover in the composition of the committee members. The second explanation is that the committee focuses on some subtle combination of qualities, which leads to the uniform response of its members. If a high sensitivity to the quality of the publications as opposed to simply their number predominates, then it could be concluded that the quality of the female applicants was slightly lower than that of their male colleagues. The fact that the successful females had a higher average number of papers may indicate that their results have been parcelled into a larger number of less consequential papers. Why this phenomenon is more prevalent in one sex is unclear unless there is a signal in the fact that the males had a slightly higher number of first‐author papers. Could it be that the females were more co‐operative in their home laboratory and then were placed in middle authorship positions with the consequent message to the panel that they were less focused on their work?
One could claim that women needed a greater number of publications in order to be successful, but the data equally illustrate a need for men to have publications with a higher impact factor
The conclusion of this study is therefore untidy; a tiny difference between the female and the male applicants appears to have been magnified in the selection process. One could claim that women needed a greater number of publications in order to be successful, but the data equally illustrate a need for men to have publications with a higher impact factor. Crass discrimination against women therefore does not take place, but there is something wrong in a system that consistently results in a lower success rate for female scientists.
The problem of the lower number of women in the scientific system as they move through their careers has been highlighted in many reports (ETAN report on Women and Science, 2000) and many measures to overcome male dominance in academia have been discussed. The biggest challenge for granting schemes, such as the EMBO fellowships, however, is to discover—if there are—sources of bias against women and the means to ensure that these are eliminated during the decision‐making process. Studies such as this may help to raise the consciousness of the decision‐makers on the problem and to find and dispel any deep‐rooted bias that may have an unfair influence on the judgement of fellowship applications.
- Copyright © 2001 European Molecular Biology Organization