RELATED STORY: Faculty-Productivity Index Offers Surprises (November 16, 2007)
How The Index Works
The index examines faculty members who are listed on a Ph.D. program's Web sites, and includes a total of 217,254 names. A professor listed in both history and American studies would be counted twice. But at the next level of aggregation (the humanities in this case), the professor would be counted only once. The index creators call this "de-duplication." The total number of actual faculty members rated by the index is 164,843.
The productivity of each faculty member is measured, although the data are aggregated before being published. Faculty members can be judged on as many as five factors, depending on the most important variables in the given discipline: books published; journal publications; citations of journal articles; federal-grant dollars awarded; and honors and awards.
For each discipline, Academic Analytics assigns a weight to each variable. Publications, which include journal articles, citations of those articles, and in many cases, books, count as 60 points out of 100. Books are included in six of the eleven broad fields: Business; Education; Family, Consumer and Human Sciences; Health Professions Sciences; Humanities; and Social and Behavioral Sciences but not in Agricultural Sciences; Biological and Biomedical Sciences; Engineering; Natural Resources and Conservation; and Physical and Mathematical Sciences Books that were published from 2002 to 2006 were recorded using Baker and Taylor's database. When books are included, their weight is five times that of journal articles for the Humanities and three times that of a journal article in other broad fields. Journal articles are counted for the years 2004, 2005, and 2006. Citation counts cover a four year span so refer to citations to articles published for the years 2003, 2004, 2005 and 2006. The index uses Scopus, an abstract-and-citation database that covers more than 15,000 peer-reviewed journals.
Grants count as 30 points out of the 100, if they meet a threshold of importance in a particular discipline — that more than 10 percent of the programs in that discipline have received a federal grant. Grant data from 2004, 2005, and 2006 were collected from the National Institutes of Health, the National Science Foundation, the U.S. Department of Education, the National Endowment for the Humanities, and the U.S. Department of Agriculture, NOAA, and from three programs in the Department of Energy.
Awards and honors count as 10 points out of 100, as long as more than 10 percent of the programs in the discipline have received awards. Data are collected from the Web sites of 357 organizations that grant awards and honors and are matched to names and programs.
Awards considered more prestigious are given more weight than others. For example, most awards, like Fulbrights, are counted only if they were awarded between 2002 and 2006. But a Nobel Prize can be counted in the 2006-07 index if it was awarded within the past 50 years.
If one or more variables are not used in the calculation of faculty productivity, that part of the equation is removed and the point scale reduced accordingly. So if honors are not included, the total possible score is reduced to 90 from 100. Institutions that pay for the data have the ability to reweight the variables in any category, according to their preferences. Starting with FSP 2006-07, subscribers to Academic Analytics will also have the option to obtain the complete dataset for disciplines of interest to them, so they can use the raw data as they please. For more information about the data, contact Academic Analytics.
The faculty's scholarly productivity in each program is expressed as a z-score, a statistical measure (in standard deviation units) that reveals how far and in what direction a value is from the mean. The z-score allows the performance of programs to be compared across disciplines. A z-score of zero indicates that the program is at the national mean for the discipline; a z-score of 1 indicates that the program is one standard deviation unit higher than the national mean.