Junior scientists who collaborate with a leading scholar early in their career are more likely to become top researchers themselves, particularly if they are based at a less prestigious university, according to a paper.

A study based on more than 22,000 published academics found that 32.5?per cent of those who co-authored research with a leading scientist in the first three years of their career went on to become a highly cited scientist themselves by the 20th year of their career, compared with 19.1?per cent of academics who did not co-author a paper with a prominent figure early on. The figure rises to 45.3?per cent among academics who co-authored papers with more than one top scientist early on.

The research, published in , found that co-authorship with a top scientist did not make a difference to the professional trajectory of those scholars who, early in their careers, were in the top 10?per cent among their peers in the following three categories: institutional prestige, research productivity and citations received.

However, co-authorship with a top scientist ¡°truly has potential career-altering consequences for junior researchers who are not in the top 10?per cent of any of the categories we considered¡±, according to the study. Such academics who co-author work with top scientists keep attracting citations at a higher rate than their peers even for papers that are not co-authored with other top scientists throughout their entire career.

ÖйúAƬ

The research was based on all academics who had published articles in journals on cell biology, chemistry, physics and neuroscience, whose career started between 1980 and 1998 and lasted at least 20 years, who have at least 10 publications, and who have published at least one paper every five years ¨C a total of 22,601 scholars.

Giacomo Livan, senior research fellow in computer science at UCL and co-author of the study, said that although the findings were not surprising in qualitative terms, he had not expected the co-authorship effect to be so ¡°consistent and robust¡±.

ÖйúAƬ

¡°We tried to check the robustness of the results with respect to a number of potential confounding factors, and we always found them to be there,¡± he said.

ellie.bothwell@timeshighereducation.com