Limits...
Research groups: How big should they be?

Cook I, Grange S, Eyre-Walker A - PeerJ (2015)

Bottom Line: The relationships for the impact factor and the number of citations are extremely weak.Our analyses suggest that an increase in productivity will be achieved by funding more PIs with small research groups, unless the cost of employing post-docs and PhD students is less than 20% the cost of a PI.We also provide evidence that post-docs are more productive than PhD students both in terms of the number of papers they produce and where those papers are published.

View Article: PubMed Central - HTML - PubMed

Affiliation: School of Life Sciences, University of Sussex , Brighton , United Kingdom.

ABSTRACT
Understanding the relationship between scientific productivity and research group size is important for deciding how science should be funded. We have investigated the relationship between these variables in the life sciences in the United Kingdom using data from 398 principle investigators (PIs). We show that three measures of productivity, the number of publications, the impact factor of the journals in which papers are published and the number of citations, are all positively correlated to group size, although they all show a pattern of diminishing returns-doubling group size leads to less than a doubling in productivity. The relationships for the impact factor and the number of citations are extremely weak. Our analyses suggest that an increase in productivity will be achieved by funding more PIs with small research groups, unless the cost of employing post-docs and PhD students is less than 20% the cost of a PI. We also provide evidence that post-docs are more productive than PhD students both in terms of the number of papers they produce and where those papers are published.

No MeSH data available.


Paers per group member versus group size.The number of publications per group member versus group size.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4465944&req=5

fig-3: Paers per group member versus group size.The number of publications per group member versus group size.

Mentions: The average number of publications published by each group in the previous five years was 22.0 papers (SD = 18.8) but varies considerably between PIs, from 0 to 177. The number of publications over the preceding 5 years is significantly correlated to the total group size (r = 0.43, p < 0.001) (Fig. 2). However, group size explains less than 20% of the variance in the number of papers, and at all levels of group size there is substantial variance in the number of papers produced (Fig. 2). The slope of the relationship between the number of papers and group size is significantly less than one (b = 0.57 (SE = 0.06), p < 0.001) indicating a diminishing returns relationship; i.e., the number of papers increases with group size but less than proportionally. This is not simply due to adding one to the number of papers before log transforming, because if we do not add one and drop the one group with no research papers, we get qualitatively the same relationship: b = 0.62, p < 0.001). The diminishing returns relationship can be illustrated simply by dividing the (untransformed) number of papers by the (untransformed) group size; the number of papers per group member decreases as group size increases (Spearman’s rank correlation = − 0.20, p < 0.001) (Fig. 3).


Research groups: How big should they be?

Cook I, Grange S, Eyre-Walker A - PeerJ (2015)

Paers per group member versus group size.The number of publications per group member versus group size.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4465944&req=5

fig-3: Paers per group member versus group size.The number of publications per group member versus group size.
Mentions: The average number of publications published by each group in the previous five years was 22.0 papers (SD = 18.8) but varies considerably between PIs, from 0 to 177. The number of publications over the preceding 5 years is significantly correlated to the total group size (r = 0.43, p < 0.001) (Fig. 2). However, group size explains less than 20% of the variance in the number of papers, and at all levels of group size there is substantial variance in the number of papers produced (Fig. 2). The slope of the relationship between the number of papers and group size is significantly less than one (b = 0.57 (SE = 0.06), p < 0.001) indicating a diminishing returns relationship; i.e., the number of papers increases with group size but less than proportionally. This is not simply due to adding one to the number of papers before log transforming, because if we do not add one and drop the one group with no research papers, we get qualitatively the same relationship: b = 0.62, p < 0.001). The diminishing returns relationship can be illustrated simply by dividing the (untransformed) number of papers by the (untransformed) group size; the number of papers per group member decreases as group size increases (Spearman’s rank correlation = − 0.20, p < 0.001) (Fig. 3).

Bottom Line: The relationships for the impact factor and the number of citations are extremely weak.Our analyses suggest that an increase in productivity will be achieved by funding more PIs with small research groups, unless the cost of employing post-docs and PhD students is less than 20% the cost of a PI.We also provide evidence that post-docs are more productive than PhD students both in terms of the number of papers they produce and where those papers are published.

View Article: PubMed Central - HTML - PubMed

Affiliation: School of Life Sciences, University of Sussex , Brighton , United Kingdom.

ABSTRACT
Understanding the relationship between scientific productivity and research group size is important for deciding how science should be funded. We have investigated the relationship between these variables in the life sciences in the United Kingdom using data from 398 principle investigators (PIs). We show that three measures of productivity, the number of publications, the impact factor of the journals in which papers are published and the number of citations, are all positively correlated to group size, although they all show a pattern of diminishing returns-doubling group size leads to less than a doubling in productivity. The relationships for the impact factor and the number of citations are extremely weak. Our analyses suggest that an increase in productivity will be achieved by funding more PIs with small research groups, unless the cost of employing post-docs and PhD students is less than 20% the cost of a PI. We also provide evidence that post-docs are more productive than PhD students both in terms of the number of papers they produce and where those papers are published.

No MeSH data available.