Many PhD students dream of getting a top job in academia. However, what determines whether one gets such a job is not always clear. Now, Lucas Carey et, al. principal investigator of the Single Cell Behavior Lab  in the Department of Experimental and Health Sciences (CEXS) at Pompeu Fabra University in Barcelona, together with  Ohad Manor, from Washington University in Seattle (USA) and David van Dijk, from the Weissman Institute of Rehovot (Israel), have mined a large database of scientific publications in order to develop a mathematical model capable of predicting who will become a professor. They have found that academic career changes are remarkably predictable.

Like in most competitive jobs, there are many more PhD students than there are positions as professors. This excess has several sources, among which are the current stagnation in scientific funding, and the fact that professors rarely voluntarily vacate their positions at a young age.

Thus, fewer than 10% of PhD students will go on to become a research professor, or principle investigator (PI) as a scientist that leads a group is usually called. The result is a highly competitive environment in which hundreds of applicants apply for every available position, and a committee decides which of these aspiring scientists will get to run their own research group.

In order to better understand the nature of this selection process and to determine which elements on an applicant's resume are predictive of success (or failure), the researchers collected the publication records for thousands of scientists and fed these data into a computer. By exploiting the fact that authorship position reveals academic status (professors are usually the last name on the list of authors for a publication) their program could extract which of these scientists became PIs, and which did not, and as a result it could "learn" what type of resume (i.e. publication record) is associated with academic success.

They found that, just based on publication record, they were able to predict for each scientist, with remarkable accuracy, who becomes a PI, and, to a lesser extent, how much time this process will take. This in itself is informative -- it suggests that other factors, such as personal connections and "soft skills", are either not important or are highly correlated with scientific output. Secondly, they found that the prestige of the journal in which a scientist publishes is far more important than the number of times the work is actually cited, suggesting that perceived quality trumps actual scientific quality.

Finally, by utilizing the power of their mathematical model they were able to measure the effect of both gender and the prestige of the university, and found that men coming from highly ranked universities not only had better resumes, but also, given the same resume, had a better chance at becoming a PI.

How "quality" of science and scientists should be measured is the topic of much current debate. These results should assist and stimulate further debate, to make science and academia more accessible and productive for the benefit of mankind and such that tax money is spent in the optimal way.

The question if one should become a professor and in specific what it takes to become one are questions that all junior scientists face at some point in their career. The researchers have made the predictive power of their model available online so that any scientist can predict their own future (www.pipredictor.com).

Reference:

David van Dijk, Ohad Manor, Lucas B. Carey (2014), " A quantitative analysis of publication metrics and success on the academic job market ", Current Biology, vol. 24, num. 11, 02 de juny.

Other relevant news about:

Richard Van Noorden (2014)," Computer model predicts academic success", Nature News, 2nd of June.

John Bohannon (2014), " Career Moneyball", Science Careers, 2 de juny.