candidates interviewed out of the total number of candidates (the cooperation rate) is 29 and 24 percent respectively.
The committee recognizes that the lower than desired response rate for top-ranked and competitive candidates might affect the validity of the evaluation. Also, the committee is especially concerned about the very low number of candidates in the initial cycles of Markey funding. These very low numbers preclude potential analyses by funding cycle.
The committee recognizes the possibility of a selection bias in the two comparison groups in that some top-ranked and competitive candidates were not available to participate in the study. Unfortunately, the committee did not have the resources to contact a sample of these nonrespondents to determine if their characteristics were similar to the characteristics of those comparison group members who did respond. However, the committee reasoned that professionally successful comparison group members would be less likely to self-select out of the study.
The committee also worried that the lower than expected number of female Scholars and comparison group members may have had some impact on the study. The percentage of females at all stages of the Markey Pathway (Table 2-4) was lower than expected. In 1988 (the middle of the Markey award period), for example, 37 percent of persons graduating with doctorates in the biological sciences were female, while only 24 percent of candidates for the Markey award were female (National Science Foundation, 1997). The committee was less concerned about the higher number of candidates with Ph.D. degrees than those with M.D. or M.D./Ph.D. degrees. In 1988, there were over 4,000 Ph.D. graduates in biological sciences (National Science Foundation, 1997). However, of the nearly 16,000 M.D.s that graduated in 1988, only 214 were in M.D./Ph.D. programs and only 540 anticipated a research fellowship (American Association of Medical Colleges, 1988).
The committee is aware that the outcome measures used in this study may have been affected by both the low response rates of persons in the two comparison groups and the lower than expected number of females among Markey Scholar candidates. The latter factor, however, would not have influenced the analytical comparisons since the percent of Markey Scholars who are women (19.5%) is similar to the percent of women in each of the two comparison groups (18.8%), as shown in Table 2-4.