Institutional-Level Scholarly Impact in Academic Social Work: Results From a National Probability Sample

Background: Bibliometric methods such as citation analysis are commonly used to measure both research productivity and scholarly impact. Recently, the h-index, a popular bibliometric indicator, has been used to measure the impact of social work scholars. The h-index combines the number of articles published with the rate of citation; an academic with h-index=15 has published 15 articles that have been cited at least 15 times apiece. H-index is arguably superior to publication counts in that citation of articles reflects impact upon the field. At this point, descriptive norms for individual h-index (by rank and reputational ranking) have been disseminated. A substantial amount of scholarship has looked at institutional-level productivity rates using classic approaches (such as analyzing the number of publications per faculty member). However, no published research has  examined institutional-level h-index values.

Purpose: To estimate the 'collective h-index' values for a heterogeneous sample of institutions within academic social work.

Method: Using the 2008 U.S. News and World Report reputational rankings, a probability sample consisting of 26 Top-80 schools of social work was drawn, stratified by quartile. Complete publication records for each scholar were captured using Harzing's Publish or Perish bibliometrics software. A time-intensive coding and verification process was used to code publication records as accurately as possible. Publication records for all academics at the same institution were then combined to create a collective h-index for the institution as a whole.

Results: For institutions in the 1st quartile (n=6), the mean collective h-index was 58.83 (SD=21.94); in the 2nd quartile (n=5), mean collective h-index was 35.0 (SD=6.4); in the 3rd quartile (n=8), mean collective h-index was 34.25 (SD=11.85); and in the 4th quartile (n=7), mean collective h-index was 27.14 (SD=9.41). Median values were very similar. Mean collective h-index values fell with decreasing reputational ranking; there was a statistically significant relationship between quartile and collective h-index value (i.e., Kendall's Tau=0.513, p<.001). Performing the same analysis using the U.S. News and World Report 2012 rankings instead of the 2008 rankings generated the same general pattern of results. Exploratory data analyses demonstrated that ranking institutions by h-index resulted in some schools moving up or down substantially on the list; for instance, one state university ranked mid-list reputationally had a higher collective h-index than some elite schools ranked in the 1^stquartile. Comparisons were also made between these institutional h-index values and the h-index values of prolific individual social work scholars. The most prolific scholars in social work have individual h-index scores exceeding the collective h-index of some institutions.

Conclusion: Collective h-index values fell with decreasing reputational ranking. Institutions in the 1st quartile had a substantially higher collective h-index as compared to other quartiles, while differences between the 2nd, 3rd, and 4th quartiles were modest. The results were consistent with the existing social work literature arguing that reputational rankings and productivity/bibliometric impact indicators measure different concepts. This is the first published work on institutional h-index in social work, and contributes to the continuing debate regarding the validity of reputational rankings.