CRA Taulbee Survey: Ph.D.s Holding Steady

By Gregory R. Andrews
Chair, CRA Surveys Committee

For 24 years the Computing Research Association and its predecessor--the Computer Science Board--have been charting the growth of Ph.D. production and employment of computer scientists and computer engineers in North America.

The accompanying tables present the results of this year's CRA Taulbee Survey of Ph.D.-granting departments of computer science (CS) and computer engineering (CE)--and combinations thereof--in the United States and Canada. Each September, the survey is mailed to all organizations included on the CRA Forsythe List of departments that offer a Ph.D. in computer science or computer engineering.* The tables include all responses received by the end of January.

Information on degree production and enrollment applies to the last academic year (1993-94). Information on faculty applies to the current fiscal year (1994-95). Faculty salaries reflect those in effect as of Jan. 1, 1995.

The response rate--especially from computer science departments--once again was quite high. This greatly enhances the utility of the data. However, one should keep in mind that the results are from Ph.D.-granting departments only. There are hundreds more departments that award only bachelor's or master's degrees.

The survey was revised and expanded this year. I will describe the changes and rationale for them. The remainder of the article comments briefly on the most important results and trends.

Additions and changes

This year's survey questionnaire appeared to be much longer than in the past (10 pages versus four), much to the dismay of some department chairs. Part of the increased length was due merely to using a more graphical layout and including general instructions. We did seek some additional information.

The 1994 CRA Conference at Snowbird pointed out the need for more detailed information on Ph.D. production and demand such as subfields in which dissertations are done and the need for a more complete view of the "pipeline" of bachelor's and master's students. This year's information is interesting in its own right; it should prove especially useful in years to come, because it will give us more detailed information on trends.

The other additions to the survey grew out of our desire to have data that is comparable in scope, detail and reporting periods to that used by federal agencies in the United States and Canada. Consequently, we clarified the reporting periods, added enrollment information and changed categories for ethnic origin and residence status.

Many questions were reworded to make them more consistent and precise. Finally, questions about students were expanded to distinguish between students in computer science and computer engineering degree programs, and to distinguish between full-time and part-time students.

Perhaps as a result of the increased length and complexity of the survey, the response rate was down slightly from last year (from 94% to 92% of CS/CE departments).

Results and trends

To a first approximation, academic computing has reached a steady state, at least within Ph.D.-granting departments. The number of Ph.D.s awarded has leveled out, and the number of Ph.D.-granting departments was up only slightly, after dramatic growth in the 1980s. Enrollment in Ph.D. programs was fairly stable. Anecdotal evidence indicates that enrollment also was fairly stable in undergraduate and master's programs, after a dramatic fall in undergraduate enrollment in the late 1980s. Faculty sizes in individual departments also were virtually unchanged. Moreover, departments no longer expect the number of faculty positions to grow much over the next five years, despite previous predictions of growth of 10--20% over five years. (Such predictions always have been overly optimistic; this year's numbers are probably much more realistic.)

The number of Ph.D.s reported this year (1,005 for both computer science and computer engineering) appears to be down somewhat from the past three years. However, ever since CE programs have been included in the survey, it has been difficult to get a high rate of response from departments offering CE degrees. And those departments sometimes have had difficulty separating out CE from EE degrees. (The response rate from CS also is down slightly this year.) Still, we believe this year's figures on Ph.D. production are accurate, and that those from the previous few years were slightly higher than they should have been. Hence, Ph.D. production has for all practical purposes been steady during the 1990s.

Table 5 presents the employment status of last year's Ph.D. recipients and includes new information on degree areas (specialties). Not surprisingly, there are significant differences between the numbers of Ph.D.s in various specialties, and the specialty and employer mix. Despite horror stories and student fears, it appears that most of last year's graduates found jobs. However, the data could be somewhat misleading, because some students might have deferred graduation. Postdoctorates were not listed as a separate category. The list of specialties may need to be refined. (It looks like ÒotherÓ was a good category to be in last year.)

The percentage of female Ph.D. recipients in computer science continued to increase (from 12% in 1992, to 14% in 1993, and 17% in 1994). Moreover, 23% of the new hires for tenure-track faculty positions were women, so they were hired into such positions in a higher proportion than their presence in last year's graduating class.

However, there is no clear trend in the numbers of female professors at various ranks: the number of assistant professors was up slightly (133 to 137), and the number of associate professors is up significantly (87 to 102), but the number of female full professors decreased (66 to 59). (This year's information combines CS and CE faculty, so it is possible some EE faculty erroneously have been included.)

The percentage of degrees awarded to women last year was almost the same for bachelor's (18%), master's (19%) and Ph.D. degrees (17%). Tables 18-26 report on faculty salaries. Most readers most likely have already studied these tables in detail and made their own interpretations.

For these tables, each department was asked for the minimum, mean and maximum salary for each category of professor. Because tables show the minimums and maximums of the minimums and maximums reported by each department, these figures reflect salaries of individual professors. Also shown are the means of the minimums and maximums reported by each department. Finally, the average of all salaries is the average of the means reported by each department. If a department gave only a partial answer for a category of professor, it was discounted. All Canadian salaries are in Canadian dollars.

Rankings

For Tables 1, 12 and 18-26, which group computer science departments by the rank of 1-12, 13-24 and 25-36, we based our ranking on information from a 1980 assessment of research--doctorate programs in the United States done under the auspices of the National Research Council. We modified our ranking to include top Canadian universities.

Our top 12 schools are Stanford University, Massachusetts Institute of Technology, Carnegie Mellon University, University of California at Berkeley, Cornell University, University of Illinois at Urbana-Champaign, University of California at Los Angeles, University of Toronto, University of Washington, University of Texas at Austin, University of Wisconsin at Madison and the University of Southern California.

The departments ranked 13-24 are the University of Maryland, Princeton University, Brown University, University of Utah, New York University, University of Massachusetts at Amherst, the State University of New York at Stony Brook, University of North Carolina at Chapel Hill, University of Pennsylvania, Yale University, Pennsylvania State University and the Georgia Institute of Technology.

The departments ranked 25-36 are the University of California at San Diego, the California Institute of Technology, Columbia University, Ohio State University, Rice University, Duke University, Northwestern University, Syracuse University, Rutgers-the State University of New Jersey, University of California at Irvine, University of Minnesota and the University of Rochester.

Acknowledgments

The staff at CRA headquarters--notably Juan Osuna and Phillip Louis--were responsible for drafting the survey, collecting information and preparing the accompanying tables. Jeffrey Ullman of Stanford University, Duncan Lawrie of the University of Illinois at Urbana-Champaign, and John Savage of Brown University helped refine the survey. Mary Jane Irwin of the Pennsylvania State University provided the interpretation of female statistics noted in this article. Lawrie and Robert Schnabel of the University of Colorado provided useful feedback on this article.