What Makes A Productive I-O Faculty Member: A Predictive Validity Study January 2008

Industrial-organizational (I-O) psychologists are often interested in the predictors of job performance across a wide range of occupations (Schmitt & Chan, 1998). Ironically, few have examined the predictors of the scholarly performance (i.e., research productivity) within our own discipline. An exception is Judge, Kammeyer-Mueller, and Bretz (2004), who looked at predicting career success among I-O psychologists. Others have looked at predicting scholarly productivity in other fields (Buchmueller, Dominitz, & Hansen, 1999; Hansen, Weisbrod, & Strauss, 1978; Hogan, 1981, 1986; Long, Allison, & McGinnis, 1979; Long, Bowers, Barnett, & White, 1998; and Williamson & Cable, 2003). However, these studies either cumulated productivity over the entire career (e.g., Judge et al., 2004) or examined only early career productivity (e.g., Williamson & Cable, 2003). Yet, many academic institutions are interested in more senior-level searches or in predicting research productivity once tenure is granted. Also, many of the studies appeared to ignore the effect of lags (i.e., time spans) on the criterion. Specifically, criterion measures of scholarly performance usually involve publications, which are typically the result of years of development, review, and long publication queues for printing. Thus, work begun and carried out largely in graduate school may likely not “count” until one is well ensconced in their first job. These issues can lead to spurious conclusions if not considered.

In the current study we examined archival data of I-O psychologists who received PhDs from psychology departments and are employed in either psychology or business departments. Specifically, we recorded research productivity (number of publication; number of citations) over 12 years of their careers. To examine the potential lag effect, described in more detail below, we divided the observations into two time periods based on the assumption tenure occurs between the sixth and seventh year after graduation. Below we review the literature around predicting research productivity to assess our hypotheses.

Research Productivity

Educational institutions, like business institutions, obtain competitive advantage through the activities and performances of the members in that institution (Coff, 1997). For university faculty, the activities and performance considered important tend to fall into three categories: research, teaching, and service (Dunn & Zaremba, 1997). The importance of the domains varies across institutions, and the cost of obtaining quality information varies across domains. Faculty success in research has been found to increase the notoriety of the faculty’s department and is a major factor in the promotion of faculty members (deMeuse, 1987). Indeed, Rosenfeld and Jones (1987) observed a positive relationship between the number of publications a faculty member had and their academic rank 6 years after receipt of their doctorate within the field of psychology. For research-oriented institutions seeking faculty to train PhD students, publicly available information regarding academic publications (e.g., number and quality) has been the metric of choice at both the individual and program levels of analysis (Gibby, Reeve, Grauer, Mohr, & Zickar, 2002; Hansen et al., 1978; Hogan, 1981, 1986; Judge, et al., 2004; Levine, 1990; Long et al., 1998; Rosenfeld & Jones, 1987; Trieschmann, Dennis, Northcraft, & Niemi, 2000; Winter, Healy, & Svyantek, 1995; Zivney & Bertin, 1992). Thus, this seems to be an appropriate criterion with which to develop a selection model for a PhD program in I-O psychology. Indeed, it was concerns regarding contaminants in this criterion (not the criterion itself) in the literature that encouraged our own investigation.

Several studies, mostly conducted on non-I-O samples, have investigated numerous predictors for the research productivity criterion and have found a mixture of effects. Probably most ubiquitous has been the attention given to academic origin, graduate school research productivity, and academic affiliation as predictors of ones’ research productivity. Below we describe these predictors.

Academic origin. The first predictor we were interested in was the quality of the institution where an individual received their degree. Williamson and Cable (2003) suggested that highly productive graduate departments may provide students with more advantages than less productive graduate departments. Long, et al. (1998) explained that academic origin should be related to research productivity for two reasons: (a) high-status institutions should be able to successfully recruit doctoral students of perceived higher quality and potential, and (b) high-status institutions provide students with human capital advantages (e.g., knowledge that is conveyed, social ties that are formed with former graduates and faculty, and the value society places on the prestige of the institution) that should aid them in succeeding in future careers.

However, the findings here are mixed. Several studies found that graduate program quality was a positive predictor of future research productivity (Hogan, 1981, 1986; Williamson & Cable, 2003). In these studies, the criterion, research productivity, was cumulative over typically 6 or fewer years since graduation and included samples from graduates of management and economics programs. Additionally, several studies found no relationship between graduate program quality and future productivity (Judge, et al., 2004; Long, et al., 1979; Long, et al., 1998; Rodgers & Maranto, 1989). In these studies, research productivity was cumulative over as few as 12 years or as many as an individual’s entire career and included samples of graduates from programs in biochemistry, management, psychology, and industrial-organizational psychology. Finally, Hansen et al. (1978) observed a negative relationship between graduate program quality and research productivity among economists. Their measure of research productivity spanned an individual’s career. These researchers concluded that their negative finding was due to deficiency in construct validity. Specifically, their measure of research productivity was based on the number of publications an individual had authored with no measure of the quality of research. Rodgers and Maranto (1989) also found nonsignificant negative relationships between graduate program quality and productivity in two of their models. Given the above, we hypothesized that academic origin would positively relate to graduate school productivity, academic affiliation, and pretenure productivity but probably not post-tenure productivity.

Academic affiliation. Where an individual currently conducts their research has been found to be one of the strongest predictors of research productivity (Long, et. al., 1998; Williamson & Cable, 2003). Although this predictor would be unnecessary for any given selection system (those selected by a single academic department will share the same affiliation), it is interesting to note the impact of environmental influence on research productivity. Long, et al. (1998) hypothesized that the pressure to publish from one’s peers, or an environment that fostered and encouraged publishing, was potentially responsible for this positive finding. In their study the criterion was cumulative research productivity over a 12-year period. Williamson and Cable (2003) found that among faculty in management schools, initial job placement was positively related to productivity in the first 6 years post graduation. Likewise, we expected to find that academic affiliation would be positively related to pre-tenure productivity and positively related to post-tenure productivity.

Preappointment, pretenure, and post-tenure productivity. Behavioral consistency theory suggests that the best predictor of future performance is past performance (Wernimont & Campbell, 1968). A study by Long, et al. (1979) found that the future level of publications was strongly influenced by predoctoral publications. However, one particular concern is that the lag in the processes involved in creating publications produces a potential contaminate in the postgraduate school research productivity measure. Indeed, Williamson and Cable (2003) attempted to control this by breaking productivity into 1 through 3 and 4 through 6 years. They found that pre-appointment publications and presentations were positively related to post-appointment publications and presentations more so in the earlier time period than the later. Indeed, the length of the lag in productivity is unclear. Often a study can take 5 to 7 years from conception to publication. Thus, publications immerging even 4 to 6 years after graduation may reflect the ripe fruits of projects begun in graduate school!

In a study by Judge, et al. (2004), the researchers found support for the positive effects of graduate school productivity on career productivity. Additionally, Buchmueller, Dominitz, and Hansen (1999) found support for the positive effects of graduate school productivity on research productivity 6 years after receiving one’s degree. We expected that graduate school productivity would positively relate to pre- and post-tenure productivity.

Method

Participants
The 2002 SIOP directory was used to identify a sample of all members that graduated from a PhD program in industrial-organizational psychology, social-organizational, social-industrial, and organizational behavior programs between 1982 and 1987¹ and who were currently in academic positions with graduate programs (i.e., master’s or doctorate). The final sample included 94 individuals (39 in business departments, 55 in psychology departments). From each member listing we recorded their name, from where they graduated, and their most recent affiliation.

¹ These years were chosen in order to allow 13 years worth of data for each sample member as well as accommodate lags in citation counts.

Measures
Individual research productivity. To determine the research productivity of each member in the sample, the number of publications for each individual was found using PsychInfo. Additionally, the Social Sciences Citation Index was used to identify the number of citations for each publication found. The number of citations a published work received was used as a measure of the quality of the work. To this end, a composite containing both the sum of publications and citations was calculated to determine research productivity.² Both measures were negatively skewed, therefore a natural log transformation followed by a z-score transformation standardized the data. Finally, the z-scores were averaged to create the research productivity composite.

² Individual research productivity was also calculated using the average citation per publication rather than the sum of citations (Howard, Cole, & Maxwell, 1987). This method produced the same conclusions as the method reported.

Composites of publications and corresponding citations for each member of the sample were developed for three time frames: (a) all research productivity up to the members first year³ after receiving a graduate degree (identified as graduate school productivity), (b) all research productivity from 2 years postgraduate training through year 6 postgraduate training (identified as pretenure productivity), and (c) research productivity in years 7 through 12 post-graduation (identified as post-tenure productivity).

³ The first year was included in the graduate school productivity measure because these publications were likely “in press” while individuals were searching for their first job.

Academic origin. The quality of academic origin was based on two measures of departmental output. The first measure came from Levine (1990), where he identified the number of publications a given I-O department had in the Journal of Applied Psychology during the 1980s, which was the time period the sample was in graduate school. If a school was not listed, a score of five was given, which was halfway between the lowest score given (10) and the lowest score possible (0). The second measure described the number of SIOP presentations an I-O department had during the years 1986–2000 (Payne, Succa, Maxey, & Bolton, 2001). For the purpose of our study, only the years 1986–1990 were used for any given school to capture a graduate program’s research productivity while the individual was a student. If a school was not listed, a score of zero was given. To determine the overall score for an institution, the z-scores from each measure were averaged. Using this strategy, academic origin values were estimated for 23% of individuals in our sample. With zeros and fives added to missing programs, the internal consistency reliability of these two measures was 0.81. Although high, one concern is that the extrapolated data inflated the reliability; therefore, we recalculated reliability without the substituted values for missing data, which dropped the reliability to a still respectable 0.58 given two items.

Academic affiliation. To determine the quality of current academic affiliation, three different measures were used. For I-O psychology departments, two measures ranking doctoral programs were combined: North America’s Top I/O Psychology Doctoral Programs: U.S. News and World Report Revisited (Winter, Healy, & Svyantek, 1995) and The Top I-O Psychology Doctoral Programs of North America (Gibby, Reeve, Grauer, Mohr, & Zickar, 2002). Additionally, the rankings were reverse scored such that greater numbers meant higher rankings. To determine the overall score for an institution, the z-scores of each measure were averaged. If a program was not listed in either of the two measures, then a score of zero was given to the program. Because both of these measures focused on the research productivity of I-O programs granting PhDs, graduate programs awarding master’s degrees were given a score of zero. The internal consistency reliability of these two measures was 0.70 with missing data. The internal consistency reliability of these two measures was 0.82 with no missing data (zeros added to missing programs). For business school rankings Serving Multiple Constituencies in the Business School: MBA Program vs. Research Performance (Trieschmann, Dennis, Northcraft, & Niemi, 2000) was used. We also coded for the highest type of degree awarded (master’s = 0 and PhD = 1) given the typical differences in resources available to researchers in these two types of programs.

Results

Prior to the log transformations, the median number of publications found for graduate school, pretenure, and post-tenure was two (M = 2.70; SD = 2.14), five (M = 5.87; SD = 4.59), and five (M = 7.21; SD = 8.33), respectively. The median number of citations for graduate school, pretenure, and post-tenure was 11 (M = 40.97; SD = 76.70), 80 (M = 142.86; SD = 186.75), and 65 (M = 118.11; SD = 154.85), respectively. The relatively fewer citations in the post-tenure period likely reflect the shorter timeframe over which the publications were available to be cited. The differences between the median and mean values reflect the pre-log transformation skewness in the data.

Correlations between variables are listed in Table 1. Independent samples t-tests were performed on productivity measures (e.g., graduate school, pre-tenure, and post-tenure productivity) and department type (business school or psychology department). No significant differences were found. Thus, we collapsed across department type in subsequent analyses. The zero-order correlations supported the behavioral consistency hypothesis that the best predictor of future performance is past performance. By itself, academic origin was only positively related to graduate school productivity. However, academic affiliation and the type of degree granted by the institution were highly related to both postgraduation productivity measures as well as each other.

We were most interested in examining the joint effects of our predictors on the latter time period, post-tenure productivity (Table 2). Of interest were the effects of academic origin, graduate school productivity, and pretenure productivity on post-tenure productivity when controlling for current affiliation and type of degree program. Results of a hierarchical multiple regression showed that graduate school productivity no longer mattered, b = 0.01, p > .05, after pretenure productivity was added to the model, b = 0.56, p < .01. In addition, graduate school quality was negatively related to post-tenure productivity, b = -0.23, p < .01, whether we included pretenure productivity or not. This finding also held when we did not control for current affiliation and degree type.

Discussion

Our primary concern was developing a model for predicting future research productivity from information publicly available about new and not- so-new I-O psychologists. Many intuitions exist regarding the measures and predictors of productivity. Consistent with intuition, theory (Wernimont & Campbell, 1986), and research, we found that past research productivity was positively related to future research productivity. On the other hand, we found that the quality of one’s academic origin, measured in terms of the programs’ research productivity, was only positively related to research productivity while in graduate school. It did not translate into jobs at more productive affiliations or subsequent research productivity. Indeed, in later years, origin was negatively related to research productivity, once other factors were controlled. This implies that an individual’s graduate school productivity is contaminated by academic origin and that some of that productivity bleeds into pre-tenure productivity (Williamson & Cable, 2003). That is, students from “better” schools produce more publications while in graduate school, but that extra productivity should be discounted when predicting long-term productivity because it is a situational effect (i.e., not due to individual human capital).

Likewise, when considering a faculty member in another institution, the quality of that affiliation, particularly in terms of degree type, should be taken into account—though not in the way commonly portrayed (e.g., Williamson & Cable, 2003). That is, individuals in better schools and PhD programs are likely to be more productive than their brethren in lesser schools or master’s programs, but that higher productivity may be due more to qualities of the situation than the individuals.⁴ Indeed, many might agree that if we examined two individuals with the same level of productivity we might conclude that the individual from a lesser program was actually more productive given the lack of emphasis and resources needed for productivity from their department. We need to be careful, therefore, not to succumb to the fundamental attribution error (e.g., attributed individual behavior or outcomes to the individual rather than the situation).

⁴ More caution is required here than in the academic origin case because the affiliation quality is more likely substantially related to a current individual faculty member’s productively than it is to the productivity of one of its graduate students.

The findings also highlight the need to consider lags between behavior and outcomes when examining predictive relationships. The finding of a positive relationship between academic origin and early career productively seems to represent, to some extent, the time delay between work and evidence of that work (e.g., a published article; citations).

Of course, this research has its limitations. Our sample was relatively small and our criterion was relatively narrow. For instance, high-quality scholars often have other important pulls on their time, like administrative work, service work (e.g., private contracts), or other scholarly writing (e.g., books and chapters), all outcomes we did not measure. Nor did we measure teaching quality. Teaching is clearly an important part of the academics’ mission. Indeed, one implication of our findings may be demonstrating a disconnection between teaching and research. That is, prospective graduate students looking for the best graduate programs might not be served by examining quality ratings based solely on program research productivity (e.g., Gibby et al., 2002). At least they do not seem to translate into long-term (or career; Judge et al., 2004) research productivity. That seems more up to the individual and the place they can get a job.

References

     Buchmueller, T. C., Dominitz, J., & Hansen, W. L. (1999). Graduate training and the early career productivity of Ph.D. economists. Economics of Education Review, 14, 65–77.
     Coff, R.W. (1997). Human assets and management dilemmas: Coping with hazards on the road to resource-based theory. Academy of Management Review, 22, 374–402.
     deMeuse, K. P. (1987). The relationship between research productivity and perceptions of doctoral program quality. Professional Psychology: Research and Practice, 18, 81–83.
     Dunn, D. S., & Zaremba, S. B. (1997). Thriving at liberal arts colleges: The more complete academic. Teaching of Psychology, 24, 8–14.
     Gibby, R. E., Reeve, C. L., Grauer, E., Mohr, D., & Zickar, M. J. (2002). The top I-O psychology doctoral programs of North America. The Industrial-Organizational Psychologist, 39, 17–25.
     Hansen, L.W., Weisbrod, B. A., & Strauss, R. P. (1978). Modeling the earnings and research productivity of academic economists. Journal of Political Economy, 86, 729–741.
     Hogan, T. D. (1981). Faculty research activity and the quality of graduate training. Journal of Human Resources, 16, 400–415.
     Hogan, T. D. (1986). The publishing performance of U.S. Ph.D. programs in economics during the 1970s. Journal of Human Resources, 21, 216–229.
     Howard, G. S., Cole, D. A., & Maxwell, S. E. (1987). Research productivity in psychology based on publication in the journals of the American Psychological Association. American Psychologist, 42, 975–986.
     Judge, T. A., Kammeyer-Mueller, J., & Bretz, R. D. (2004). A longitudinal model of sponsorship and career success. Personnel Psychology, 57, 271–304.
     Levine, E. L. (1990). Institutional and individual research productivity in I/O psychology during the 1980’s. The Industrial-Organizational Psychologist, 27, 27–29.
     Long, J. S., Allison, P. D., & McGinnis, R. (1979). Entrance into the academic career. American Sociological Review, 44, 816–830.
     Long, R. G., Bowers, W. P., Barnett, T., & White, M. C. (1998). Research productivity of graduates in management: Effects of academic origin and academic affiliation. Academy of Management Journal, 41, 704–715.
     Payne, S. C., Succa, C. A., Maxey, T. D., & Bolton, K. R. (2001). Institutional representation in the SIOP conference program: 1986–2000. The Industrial-Organizational Psychologist, 39, 53–60.
     Rodgers, R. C. & Maranto, C. L. (1989). Causal models of publishing productivity in psychology. Journal of Applied Psychology, 74, 636–649.
     Rosenfeld, R. A., & Jones, J. A. (1987). Patterns and effects of geographic mobility for academic women and men. Journal of Higher Education, 58, 493–515.
     Schmitt, N., & Chan, D. (1998). Personnel selection: A theoretical approach. Thousand Oaks, CA: Sage.
     Trieschmann, J. S., Dennis, A. R., Northcraft, G. B., & Niemi, A. W. (2000). Serving multiple constituencies in the business school: MBA program vs. research performance. Academy of Management Journal, 43, 1130–1142.
     Wernimont, P. R., & Campbell, J. P. (1968) Signs, samples, and criteria. Journal of Applied Psychology, 52, 372–376.
     Williamson, I. O., & Cable, D. M. (2003). Predicting early career research productivity: The case of management faculty. Journal of Organizational Behavior, 24, 25–44.
     Winter, J. L., Healy, M. C., & Svyantek, D. J. (1995). North America’s top I/O psychology doctoral programs: U.S. news and world report revisited. The Industrial-Organizational Psychologist, 33, 54–58.
     Zivney, T., & Bertin, W. (1992). Publish or perish: What the competition is really doing. Journal of Finance, 47, 295–329.