ARTICLE pubs.acs.org/jchemeduc
Examining the Relationships among Doctoral Completion Time, Gender, and Future Salary Prospects for Physical Scientists Geoff Potvin*,† and Robert H. Tai‡ †
Department of Engineering and Science Education, and Department of Mathematical Sciences, Clemson University, Clemson, South Carolina 29634, United States ‡ Curry School of Education, University of Virginia, Charlottesville, Virginia 22904, United States ABSTRACT: Using data from a national survey of Ph.D.-holding chemists and physicists, time-to-doctoral degree is found to be a strong predictor of salary: each additional year in graduate school corresponds to a significantly lower average salary. This is true even while controlling for standard measures of scientific merit (grant funding and publication rates) and several other factors expected to influence salaries (field of research, type of position and rank, type of employing institution, years of seniority, and age). This picture is complicated by the inclusion of gender in the analysis, which reveals that women earn significantly less than men overall and experience no effect of doctoral completion time on their salaries, while men do see a significant gain in salary stemming from earlier completion times. Further investigation indicates that doctoral completion time is largely unconnected to measures of prior academic success, research independence, and scientific merit. This suggests that doctoral completion time is, to a great extent, out of the control of individual graduate students. Nonetheless, it can be influential on an individual’s future career prospects, as can gender-related effects. KEYWORDS: Graduate Education/Research, Chemical Education Research, Testing/Assessment, Student/Career Counseling, Women in Chemistry
’ INTRODUCTION Many years ago, in their book Becoming Professional, Bucher and Stelling provided a moral compass for graduate educators, writing (ref 1, p 280):
uniqueness in their graduate program. So appropriate measures with which to evaluate individuals are less obvious than at other levels of education. Despite these essential differences between graduate school and other levels of education, there are certain canonical qualities and skills that a doctorate is supposed to signify: the ability to define a research problem, develop or learn a methodology to appropriately explore the problem, carry out the necessary experimentation and, ultimately, discover answers to the problem. An ideal graduate student develops into one of the “symbolic analytic workers” of science:2 those individuals who are responsible for “the problem-solving, -identifying, and strategic-brokering activities” (ref 3, p 177) of the scientific endeavor, including the requisite synthesizing, assessment, and systematizing of research. It is these people who set the agenda for future research: they form the core of the peer-review system, determining which problems have worth, which should be funded, who should be supported and encouraged to carry out certain research and, ultimately, who has merit in the scientific endeavor. How, then, are graduate students evaluated as they develop into full-fledged members of the scientific community? Possible indicators include grades, GPA, and GRE scores. An earlier metaanalysis4 found a small correlation5 (r ∼0.07) between Ph.D.-level
[I]f, as we think, it is important that individuals retain some control over their own destinies, then it behooves the trainers to inform potential trainees about the probable consequences of going through any particular training program and to give them as much information as possible about the nature of passage through that program. In order to meet this challenge and, concurrently, identify areas for reform in graduate education, it is necessary to understand the relationships between graduate school experiences, measures of scientific success, and career outcomes. In particular, one would like to know how success is measured for a doctoral student in the physical sciences. Unlike other levels of education where degree and diploma attainment relies on relatively structured assessments (tests and exams, rubrics, etc.), a doctoral degree relies more heavily on unstructured assessments of research. This ambiguity is inextricably linked to the unique goal of graduate-level science education: the training and development of scientific researchers who have specialized talents and research agendas. By definition, doctoral research should consist of an original investigation that explores a previously unanswered problem. Every doctoral student experiences a high degree of Copyright r 2011 American Chemical Society and Division of Chemical Education, Inc.
Published: November 28, 2011 21
dx.doi.org/10.1021/ed100555j | J. Chem. Educ. 2012, 89, 21–28
Journal of Chemical Education
ARTICLE
GPA and “job performance”, noting that “the work of many Ph.D.s and MDs is difficult to measure and admission to such programs is more selective than undergraduate programs.” (ref 4, p 552) Moreover, correlations between GPA and job performance were found to be lower in the sciences than in other fields. More recently, Kuncel and Hezlett completed a meta-analysis to find that standardized test scores are “valid predictors of many aspects of student success across academic and applied fields” (ref 6, p 1080); however, the correlation between such scores and the most scientifically relevant outcomes, including citation counts and research productivity, are notably smaller5 (∼0.1 0.2) than short-term success outcomes such as first-year graduate GPA, overall graduate GPA, and qualifying exams (∼0.4 0.5). Another possible measure of a graduate student’s scientific merit is doctoral completion time. However, there are a number of concerns about the appropriate interpretation of doctoral completion time. In the social sciences, Picciano et al,7 reported that graduates’ self-reported evaluation of program quality, quality of professional training, career goals, and satisfaction with their dissertation advisors all had significant impacts on time-to-degree. The authors concluded that time-to-degree could, to some extent, be interpreted as a measure of the quality of doctoral preparation, but cautioned that “[S]tudents with more ambitious projects and programs that support riskier endeavors might have longer average [times-to-degree], but should not be discouraged simply because of this” (ref 7, p 8). It has been noted that doctoral completion times in many fields have increased in recent decades. In the physical sciences, the median time-to-degree has increased from 5.9 to 6.8 years in the 25-year period between 1978 and 2003.8 Other research has indicated that a number of factors can influence time-to-doctoraldegree, including financial support,9 to some extent, gender,10 and disciplinary and departmental issues.11 14 In particular, doctoral completion time has a convoluted relationship with the quality of a doctoral program, as the former has often been used, in part or in whole, as a measure of the latter.15,16 The finding that many independent factors have a significant impact on time-to-degree suggests that doctoral completion time is a complex and possibly muddled measure of scientific merit. In a multidisciplinary study, Lovitts studied doctoral completion and attrition across several graduate departments.11 The author noted that some graduate educators have implicitly assumed that a student deficiency model could explain doctoral completion and attrition; that is, it has often been assumed that weaknesses or deficiencies found in students’ backgrounds lead to a failure of the timely completion of doctorates. Instead, Lovitts found that two particular departmental factors explained a significant amount of the variance in student attrition rates; namely, the effort and resources a department commits to the integration of graduate students and the efforts made toward students’ development and understanding of the formal and informal structures of their graduate programs and departmental research cultures. Fox has pointed out that “[C]ompared with nonscientific fields, sciences are fundamentally ‘social’ and ‘organizational’” (ref 7, p 658), emphasizing the importance of social interaction and culture to the workings of graduate science education. In particular, she noted their potential impacts on women scientists. The former finding is reinforced by the findings of Golde,12 14 who has identified several disciplinary and departmental themes that explain doctoral completion including: mismatched research practices and student strengths; mismatched expectations between students and departments; structural isolation of students;
student perceptions of a poor job market upon graduation; and problems of advisor advisee compatibility. These issues are particularly relevant to the physical sciences, which have suffered from a lack of growth for at least four decades and have shown consistently high rates of attrition.18,19 Furthermore, as mentioned earlier, doctoral education is becoming longer,8 which implies that pursuing a doctorate is an increasingly large commitment, both for individuals and the graduate programs and universities who need to support them. Thus, understanding doctoral completion time and the long-term impacts of graduate experiences could shed some light on these problems. The continued dearth of women in chemistry and, especially, physics at the graduate level and beyond is well-known.19 The National Research Council has long studied issues related to women in STEM, with mixed findings.20 22 In 2001, the NRC reported that gender differences in employment across various sectors in STEM had shrunk substantively in the period 1973 1995 noting that, in institutions ranked as R-I by the Carnegie foundation (a ranking system that was abandoned after 1994 and is no longer advocated by Carnegie, though still widely known), “[M]en and women have become increasingly similar in their distribution among types of institution” (ref 20, p 6). In terms of educational preparation, it also noted significant declines in differences between men and women on many characteristics of their educational background, including ranking of their Ph.D. institutions and their doctoral completion time. However, this report also found that substantial differences between men and women persisted, including gaps in representation and salary differences—salaries for women appeared to become stagnant after about 20 years of experience, while men’s salaries continued to rise. In 2010, again using survey data collected from faculty in a set of Carnegie-ranked R-I institutions, the NRC reported that women and men appeared to have “[e]njoyed comparable opportunities within the university, and gender does not appear to have been a factor in a number of important career transitions and outcomes” (ref 22, p 4). Despite this, substantial gaps in the representation of male and female faculty persisted. In their sample, only 24.1% of assistant professors and 7.6% of full professors of physics were women, a difference assigned to attrition at the critical transition points in career development. To summarize, an outstanding issue from the existing research on doctoral education is whether or not doctoral completion time should be thought of as a faithful measure of scientific merit. If more dedicated and capable students (“go-getters”) finish more quickly, then perhaps there is some value in employing doctoral completion time as a measure of potential and capability upon graduation. However, if students with more ambitious dissertation research projects tend to finish later, then completion time is a less optimal standard for assessing potential and pushing students to finish faster could diminish their opportunities to carry out truly influential research or lower graduate education standards. As such, this study first explores the impact that doctoral completion times have on graduate students’ future salary prospects, then examines whether doctoral completion times are actually predicted by measures of merit, and, third, whether there are genderspecific effects present in either doctoral completion times or career prospects. The research questions that are addressed are: • What are the long-term impacts, if any, of doctoral completion time on chemists’ and physicists’ employment prospects, as measured by their postgraduate salaries? 22
dx.doi.org/10.1021/ed100555j |J. Chem. Educ. 2012, 89, 21–28
Journal of Chemical Education
ARTICLE
• What factors influence doctoral completion time in chemistry and physics? • Are there gender-specific effects that influence doctoral completion time or physical scientists’ postgraduate salaries? To address these questions, a set of analyses was undertaken to explore them quantitatively. This study goes beyond previous work, which has often focused on R-I academic institutions, by considering physical scientists in several sectors, including industry, government, and across academia.
’ METHODOLOGY The data used in this paper were obtained as part of Project Crossover (NSF #0440002), a mixed methodological study designed to study the transition from graduate student to independent researcher in chemistry and physics. The first part of Crossover consisted of a set of qualitative interviews with chemists and physicists across the United States. Between 2005 and 2006, a total of 125 semistructured, open-ended interviews were conducted with various members of the chemistry and physics communities: graduate students, postdoctoral fellows, professors at all academic ranks (including two past Nobel Prize winners), industrial and other research scientists, and some individuals who had left scientific research entirely (before or after completing their doctorates). Following the interviews, a pair of surveys for members of two professional scientific societies for physical scientists were developed based on the interview results and a review of the prior literature in this area. One survey was designed for student members of these societies; one was designed for full members. In this paper, only the data collected from the full members are analyzed. The final version of this survey included 145 questions covering a wide range of topics: demographic questions, early science interests, academic history, undergraduate and graduate school experiences, and postgraduate experiences. Once the survey was developed, a randomized list of 13,000 physical scientists was obtained. These individuals were mailed paper versions of the survey in June 2007. They had two options for responding: by returning the paper survey in a provided envelope, or by filling out the survey online (which included entering an individualized serial number from the mailed materials to account for respondents). Four reminders were sent to those individuals who had not yet responded over a period of six months following the initial mailing. From the initial list, 557 were determined to have contained nondeliverable addresses (out of date, etc.), and 3100 were determined to have been sent to inapplicable individuals (individuals who had never participated in graduate education, student members who were incorrectly listed as full members, etc.). The number of qualified individuals with correct mailing addresses who comprised the original sample was therefore determined to be 9343. In total, 3220 completed surveys were obtained, consisting of 2136 filled out by chemists (64% online responses, 36% paper) and 1084 by physicists (60% online, 40% paper). The gender representation of the respondents was 29% female overall (29% of chemists, 31% of physicists). To assess the representativeness of the sample, respondents’ demographic backgrounds (race ethnicity and gender) and employing institution type (universities, federal agencies, nonprofit, for-profit, and other employment not fitting these categories) were compared with the National Science Foundation’s WebCASPAR database.23 The Crossover sample was found to have similar proportionate representation across these measures as the
Figure 1. Distribution of doctoral completion times. The mean is estimated at 5.26 ( 1.27 years. Respondents who indicated “