Understanding institutional collaboration networks computer science vs. psychology

Institutions assume that if they are more productive (i.e., publish more papers), they will produce more high quality research. They also assume that if they collaborate more, they will be more productive. We test these causal assumptions using nearly 30 years of worldwide publication and citation data in Computer Science and Psychology. Four quality metrics, three collaboration metrics and one productivity metric were used. Spearman’s Rank Order non-parametric correlation shows that these three groups of variables are highly inter-correlated. Regression analysis was used to partial out the effect of the third variable and reveal the independent correlation between each pair of the variables. \ud In Computer Science, the more productive institutions publish higher quality research as measured by citation counts (including citation counts recursively weighted by the citation counts of the citing institution); the effect is the same, but not as strong, in Psychology. Higher average paper quality in both Computer Science and Psychology are more likely to be a result of greater institutional collaboration than of higher institutional productivity. The proportion of the institutional collaboration is closely linked to institutional quality and productivity. The more proportionally collaborated institutions in fact are less qualitative as well as less productiv