AbstractScientific careers are conceived as one unique pathway which scientists must follow to succeed. We report the diversity of profiles scientists exhibit based on their contributorship and look into biases in their career trajectory. We use Bayesian networks to train a prediction model based on a dataset of 70,694 publications from PLoS journals representing 347,136 distinct authors and their associated contribution statements. This model is used to predict the contributions of 222,925 authors in 6,236,239 publications, and apply a robust archetypal analysis to profile scientists by career stage. We divide scientific careers into four stages: junior, early-career, mid-career and late-career. Three scientific archetypes are found throughout the four career stages: leader, specialized, and supporting. All three archetypes are encountered for the early- and mid-career stages, whereas for junior and late-career stages only two archetypes are found. Scientists assigned to the leader and specialized archetypes tend to have longer careers than researchers who belong to the supporting archetype. There is consistent gender bias at all stages: the majority of male scientists belong to the leader archetype, while the larger proportion of women belong to the specialized archetype, especially for early and mid-career researchers.
CRAFFT: an activity prediction model based on Bayesian networks
