Late blooming in STEM careers ‘almost never happens’

Only a tiny number of low-performing early career scientists ever rise to the top of their field, according to a longitudinal survey which suggests researchers’ lifetime productivity can be predicted at an early stage.

After tracking the output of 320,564 scientists with at least 25 years of publishing experience, researchers from Adam Mickiewicz University, Poznań examined whether a researcher’s productivity changed from early career (five to 14 years after first publication) to mid-career (15-24 years) and late career (more than 25 years).

Using a journal prestige measure as a signal of research quality, the study found slightly more than half of the researchers identified in the early career stage as being in the top 50 per cent for research productivity remained top half researchers in the next stage of their careers. In turn, most remained in the top half for productivity for the rest of their careers.

However, only a tiny number of early career researchers in the bottom decile of productivity (0.5 per cent, or 162 individuals in total) exhibited “extreme upward mobility” by reaching the top 10 per cent for research productivity by late career. In addition, only a “small fraction” moved up from the bottom three deciles to the top 10 per cent, with only 0.7 per cent in the second-lowest decile making this leap.

In the case of immunology, just one researcher made that jump in the 38 OECD countries studied, with just one economist also achieving the same feat between early career and late career, says the study.

Similarly, those early career researchers identified as being in the top 10 per cent for productivity almost never fell down to the bottom 10 per cent, suggesting “scientists are heavily locked-in early on their careers in productivity classes”.

Presenting the findings at a Centre for Global Higher Education seminar, the study’s lead Marek Kwiek, Unesco chair in institutional research and higher education policy, said “how you start and finish [your career] is more or less similar” in the majority of cases and publishing productivity is “largely settled in the first five to 10 years” of a career.

That is likely because those in postdoctoral research positions who contributed to highly cited or influential papers could use this success to acquire positions and resources that led to further success. “Papers are [turned] into grants, grants led to [access to] people [such as PhD students] and equipment, which leads to more grants,” he said.

Asked by Times Higher Education if the study’s conclusion that scientific success is fixed very early was a gloomy message for researchers, Kwiek, who conducted the study with data scientist Lukasz Szymula, said the “optimistic part of this data-driven story is that only slightly more than half of top performers remain top performers in the next stage of their careers”.

“That means that slightly less than half of top performers come from lower productivity deciles,” he continued, stating: “There is some mobility, not closure or total immobility.”

Furthermore, the study focused on mainly on STEM science (although the cohort did include 12,585 social scientists) so the patterns of productivity might not apply to other disciplines, added Kwiek, who is director of the Institute for Advanced Studies in Social Sciences and Humanities.

“Late blooming in humanities is pretty possible, perhaps widespread, but this is impossible to measure on a global scale,” he said, citing difficulties obtaining data.

“But this is the pattern: late blooming understood as high publishing productivity in STEM science almost does not exist,” he said.

Studying those few individuals who “jumped up” from the bottom decile to the top 10 per cent could provide clues for individual researchers on how to succeed, Kwiek added.

“Our data show that the best way to increase productivity so rapidly is to increase internationalisation in team formation, increase prestige of journals in which we routinely publish and, sometimes, change affiliation to more affluent countries. The combination of internationalisation, journal prestige and country change…should help,” he said.

That said, it was important not to overinterpret the study’s findings, such as forcing low-performing researchers into teaching-only roles based on their early track record in science. “There is no reason to give up research early,” said Kwiek, who noted that while “publication indicators work well for huge numbers of observations, [they work] much worse at the individual level”.

jack.grove@timeshighereducation.com