Using data from a 40-year longitudinal study [first 3 cohorts of SMPY], the authors examined 3 related hypotheses about the effects of grade skipping on future educational and occupational outcomes in science, technology, engineering, and mathematics (STEM). From a combined sample of 3,467 mathematically precocious students (top 1%), a combination of exact and propensity score matching was used to create balanced comparison groups of 363 grade skippers and 657 matched controls.
Results: suggest that grade skippers (1) were more likely to pursue advanced degrees in STEM and author peer-reviewed publications in STEM, (2) earned their degrees and authored their 1st publication earlier, and (3) accrued more total citations and highly cited publications by age 50 years.
These patterns were consistent among male participants but less so among female participants (who had a greater tendency to pursue advanced degrees in medicine or law). Findings suggest that grade skipping may enhance STEM accomplishments among the mathematically talented.
[Keywords: educational acceleration, gifted, math/science talent, longitudinal analysis, propensity score matching]
…Narrowing the scope of the analysis to only male participants, for greater clarity, shows a pattern consistent with this interpretation, as seen in the right column of Figure 5. Restricting the comparisons to male participants is reasonable due to the diversity of the paths of the female participants, with many publishing early but later slowing down or transitioning out of research positions into administration or teaching or into entirely different fields or motherhood (2011; Ceci et al 200915ya). Career development of talented women seems to follow a different path than that of their male counterparts in many instances.
The results from this phase, summarized in Figure 5, illustrate a pattern of increasing advantage as the cohorts increase in age, such that the grade skippers from the 1980 cohort have no observed advantage at age 42 years while the grade skippers from the 1972 cohort have a substantial advantage at age 50 years. Two potential explanations for the observed differences in effect sizes, aside from chance alone, are (1) cohort differences in accelerative opportunities and (2) cumulative effects from grade skipping. With respect to cohort effects, grade skipping was one of the few accelerative options for the 1972 cohort; the 1976 and especially the 1980 cohorts had many more accelerative opportunities available. The shrinking effect sizes between grade skippers and matched controls in progressively later cohorts may reflect the increased availability of alternative forms of acceleration, such as advanced placement (AP) courses, college courses in high school, summer programs, and research and writing opportunities (Wai et al 201014ya), which moderated the differences between the grade skippers and their matches. For example, in the 1972 cohort, the matched controls often reported no other accelerative opportunities, but the matched controls in the 1980 cohort experienced an average of ~3 other forms of acceleration (and on average, just one less opportunity than the grade skippers). In turn, the growth of alternative forms of acceleration over time may explain the progressively smaller effect of grade skipping on age of first STEM publication as well. The 1972 cohort grade skippers tended to author their first publication 3 years earlier than the controls, while the median age advantage in first publication among grade skippers in the 1980 cohort was only 0.3, or about 4 months. While the age of first publication of grade skippers was relatively constant across cohorts, the age of first publication by matched controls gradually decreased across cohorts. It could be that other accelerative opportunities used by the 1976 and 1980 cohorts were almost as effective in saving time as grade skipping. If the effect of grade skipping on these indices is mediated by its effect on age of first publication, then the observed differences across cohorts in Figure 5 are to be expected.
A second explanation is that the grade skipping has small effects that accumulate over time. Assuming that the indices are relatively good “snapshots” of a similar pool of STEM researchers at ages 42, 46, and 50 years, then the gradual increase in the differences between grade skippers and matched controls is the result of the grade skipping advantage. If researchers publish at a relatively constant rate and citation counts grow at an exponential rate (proportional to the amount of publications), then small differences in the time of the first publication will result in gradually widening differences in citation counts as time passes. An idealized example of the process is illustrated in the Appendix Figure C2 using an exponential function to generate accumulated citations from an individual’s publication count. The relationship between publications and citations will vary considerably across disciplines and individuals, but the key point is that for any given individual, a small amount of time saved could potentially translate into a large advantage later.