Are the previous projections, for 1995–2005, consistent with current workforce data? The answer to this question is limited by the following:
To make the comparison, the earlier projections are rerun excluding the foreign-trained researchers but not changing any other assumptions.
The projections, shown in Table D-A, up to 2001 for male and female biomedical researchers were 2-3 percent too low. By comparison with actual growth, this is not a large error. For males the error is smaller than growth in the workforce between 2000 and 2001; and, for females only a sixth of 2000-2001 growth. The error is also small relative to the uncertain number of foreign-trained researchers in the workforce. Nevertheless as small as this error is, it is not easy to explain.
Looking at the number of graduates, the most important component of the projections, it can be seen that the medium projection
TABLE D-A Comparison of Projected with Reported 2001 Workforce in Biomedical Research
from 1995 did not capture the fluctuations in reported numbers but is reasonably close. The estimates of male graduates are slightly too high rather than too low.
Another possible explanation for projection error has to do with stay rates, the proportions of graduates assumed to stay and work in the U.S., based on their stated intentions. The 1995 projections assumed constant stay rates, whereas rates actually rose, so that by the year 2001 males were 93 percent instead of the projected 90 percent. This still would not account for more than a quarter of the error. Other possibilities may include retirement or death rates being slightly too high, relatively too much assumed movement toward jobs outside science, or even errors in the data (such as errors in age distribution) that could affect projections. Because the projections seem generally accurate, many of these errors, if they exist, probably cancel each other out.
doctoral degrees. It is decremented by retirements and deaths. Movements within the workforce are also modeled, such as those between employment and unemployment. The most common movements are those in both directions between science employment and nonscience employment.
The biomedical and behavioral fields include only Ph.D.s. The clinical field includes Ph.D.s and M.D.s. The distinction between Ph.D.s and M.D.s is relevant partly for methodological reasons as the kinds of data available for each are quite different. Therefore, these groups are discussed separately. Similarly, a distinction between U.S.-trained and foreign-trained researchers is important because the types of data differ.
The components of the current potential workforce, as illustrated in Figure D-1, are given in the following order: U.S.-trained Ph.D. researchers; M.D. researchers; foreign-trained Ph.D. researchers; entrants into the workforce, meaning migrants and graduates; and movements within the workforce and exits from it. Current levels as well as recent trends that suggest possible approaches to projection are