College Quarterly - Winter 2005 Page 1of 22 College Quarterly Winter 2005 - Volume 8 Number 1 Home Doctoral Student Research as a Creative Endeavour Contents by Todd Dewett, Shung Jae Shin, Soo Min Toh, and Matthew Semadeni Abstract Doctoral student research is explored as a creative activity. Hypotheses concerning several factors associated with research activity are developed and tested using a sample of management doctoral students. Findings suggest that research-related goals provided by the dissertation chair and student information seeking efforts are both positively related to research activity and that information seeking partially mediates the effect of chair goals on research activity. In addition, the attainment of research products is influenced by the student's intrinsic motivation for research as well as his or her information seeking efforts. Further, research activity mediates the effect of information seeking on research products. The implications of these findings for doctoral students and faculty members are considered. Introduction What determines the success of a doctoral student with regards to his or her research efforts? To answer this question we must recognize that completing a doctoral program is a challenging endeavour for the student and also presents a challenge to departments in terms of how best to develop students. This is an important question given the typically small number of students in a program, the significant resources the department and university must commit to each student, and the need for departments to develop and offer quality students to the marketplace. In fact, as the demand for new business-related Ph.D.s continues to outpace supply (Merritt, 2004); it is becoming more imperative to examine the issues involved in producing quality Ph.D.s in the business school. Towards that end, this paper will explore research efforts among a sample of Management doctoral students. The operating assumption for most doctoral programs is that they are research-oriented (LaPidus, 1997). That is, they are designed to help junior scholars develop the ability to conduct sound, rigorous research (Golde & Dore, 2001) based on a system that involves coursework coupled with research under the guidance of an established researcher (LaPidus, 1997). Interestingly, little recent research in the organizational sciences has examined scholars in general. In fact, Newman and Cooper (1993) found that only 4 of the over 1800 articles published in the Journal of Applied Psychology, Academy of Management Journal, and Personnel Psychology during http://www.senecac.on.ca/quarterly/2005-vol08-num01-winter/dewett_shin_toh_semadeni.html College Quarterly - Winter 2005 Page 2of 22 the period of 1980-1989 addressed issues related to scholarly productivity and rewards. Nonetheless, at least one study has examined the functioning of doctoral students. Bauer and Green (1994) examined a sample of doctoral students across an array of sciences and found that individuals who had a realistic understanding of program expectations and who had past research experience and were more involved in their doctoral programs, were more engaged in doctoral research, and were more productive as measured by objective outcomes. Researchers have, however, begun to consider management scholars post graduation. For example, Gomez-Mejia and Balkin (1992) studied factors related to faculty pay among a sample of management professors and found that the main predictor of compensation is the number of publications appearing in top-tier journals an individual has authored. Cable and Murray (1999) examined the hiring practices by management departments in schools of business and found that hiring decisions were better characterized by tournaments in which job candidates were selected based on publication and presentation success instead of the eminence of their doctoral committee chairpersons or the prestige of their Ph.D. Researchers have also considered the determinants of scholarly productivity. For example, Long, Bowers, Barnett, and White (1998) examined a class of management Ph.D. graduates and found that academic affiliation, but not academic origin, was the primary determinant of scholarly productivity. Further, Taylor, Locke, Lee, and Gist (1984) explored the relationship between Type A behaviour and faculty productivity and found that Type A behaviour (e.g., aggressively working to achieve more in less time) was directly related to research productivity. Thus, while we know that publications heavily influence job attainment, to date researchers largely have not taken a step back in the process to consider what factors influence valued student outcomes while they are still in the doctoral program. Thus, this paper adds to the body of research examining scholars, doctoral students in particular, by presenting the results of an exploratory study designed to examine factors associated with management doctoral student research activity. Specifically, we frame the discussion in terms of creativity, arguing that research activity is inherently a creative process. Research as a Creative Endeavour We suggest that the discussion of student scholarly work presents two main considerations: research activity and research products. Research activity encompasses the work and effort a student has put forth on research tasks including writing research papers, collecting data, and analyzing data. Research products represent the much more rare, or sometimes nonexistent, outcomes of these efforts. These may include refereed journal articles, book chapters, conference presentations, and so on. This perspective on http://www.senecac.on.ca/quarterly/2005-vol08-num01-winter/dewett_shin_toh_semadeni.html College Quarterly - Winter 2005 Page 3of 22 research is guided by studies of employee creativity and our contention that scholarly research is an inherently creative endeavour. In fact, it has long been noted that creativity is an important measure by which organizational scholarship can be judged (Daft, 1984). Creativity is most often defined as the production of novel and useful ideas or products (e.g., Amabile, 1988; Oldham & Cummings, 1996). Thus our approach suggests that research products are the ultimate creative outcomes for students – and that they are predicated on creative efforts (i.e., research activity). In fact, in a different context, Bauer and Green (1994) empirically demonstrated this basic relationship, finding a positive association between research activity and research products. Theoretically, this finding builds on the work of Drazin, Glynn, and Kazanjian (1999) who defined creativity as the process of engagement in creative acts. Their work suggests that a strict focus on eventual creative outcomes overlooks the importance of creative efforts throughout the process. Similar ideas were expressed in the work of Albrecht & Hall (1991), Kahn (1990), Newell, Shaw, and Simon (1962), and Politz (1975). Thus, given the foundational importance of research activity for the attainment of research outcomes, the focus of the current paper will primarily be to explore factors that influence research activity. Influences on Research Activity Although doctoral education is generally geared towards research, evidence suggests that students are not always well informed about certain aspects of the research process. In fact, one broad based study of doctoral students suggests that fewer than half of all students feel prepared by their program to publish and only just over half of all students feel confident in their ability to do so (Golde & Dore, 2001). Given the paucity of research in this area, it is unclear which variables might influence research activity among the large array of variables that could be considered. Drawing primarily on the creativity literature, three variables that are particularly salient in the context of creative work will be considered: goals set by one's dissertation chair, intrinsic motivation for research, and information seeking. Figure 1 below displays the proposed relationships among the variables. Figure 1 A Model Of Management Doctoral Student Activity And Outcomes http://www.senecac.on.ca/quarterly/2005-vol08-num01-winter/dewett_shin_toh_semadeni.html College Quarterly - Winter 2005 Page 4of 22 Dissertation Chair Goals Research has established the importance of a student's dissertation chair (e.g., Cable & Murray, 1999; Crane, 1965) as well as the utility of goals in the context of creativity (Shalley, 1991, 1995) and faculty productivity (Taylor et al, 1984). Thus, goals set by one's dissertation chair will likely be related to student research activity. The importance of a student's dissertation committee chairperson in predicting reward distributions such as job attainments and outcomes such as research productivity has been demonstrated in the literature (Cable & Murray, 1999; Crane, 1965; Reskin, 1979). However, the importance of the chair's role for student research activity has not been considered from the perspective of goal setting. Building on volumes of research suggesting that goals can increase individual productivity, Shalley (1991, 1995) found that the presence of a creativity goal has a positive effect on creative behaviour. She suggests that this might be a result of enhanced effort and the ability of goals to direct attention to relevant information, given that effort and informational resources are key ingredients for creativity. Thus, building on research supporting the efficacy of specific and challenging goals (Locke & Latham, 1990), we expect that goals set by one's chair pertaining to research productivity will be influential in guiding the student's efforts. In particular, we suspect that when a chair sets specific and challenging research-related goals for a student – to obtain high quality publications, conference presentations, and employment at a research institution – this action will be related to the student's behaviour. Given the important role of the chair in the student's professional development, the ability of goals to direct attention and effort, and the creative nature of conducting research, we propose: Hypothesis 1: Research-related goals (concerning publications, conference presentations, and employment at a research-oriented institution) set by the student's chair will be positively related to research activity. Intrinsic Motivation http://www.senecac.on.ca/quarterly/2005-vol08-num01-winter/dewett_shin_toh_semadeni.html College Quarterly - Winter 2005 Page 5of 22 Intrinsically motivated behaviours are ones for which there is no apparent reward except the activity itself (Deci, 1975), thus it refers to an inherent enjoyment of one's tasks or work. Considerable research suggests that intrinsically motivated behaviours result in creativity, flexibility, and spontaneity (e.g., Amabile, 1983; Deci & Ryan, 1985). Empirically, the general finding that intrinsic motivation facilitates creativity has been repeatedly demonstrated (see Amabile, 1996 or Ryan & Deci, 2000 for a review). It has been suggested that strong motivation is related to productivity in science (e.g., Fox, 1983). Thus, it is likely that intrinsic motivation for conducting research will play an important role in a student's ability to create a body of research activity. This stems from the fact that intrinsic motivation is thought to encourage exploration, the search for many alternatives (Zhou, 1998), and persistence (Oldham & Cummings, 1996; Zhou, 1998) - which is a key requirement for attaining publications (Schneider, 1985). In the context of conducting research these should be particularly useful behaviours. The research process is a complex and often long-term endeavour that can be fraught with both theoretical questions as well as methodological ambiguities. In particular, regarding the dissertation process for doctoral students, sustained personal interest in the topic is a key influence on successful dissertation completion (Gordon, 2003). Given that many doctoral students perceive the research process as uncertain and painful (Graham & Stablein, 1985), intrinsic motivation seems essential. In short, research is a challenging process that should be facilitated by deep intrinsic interest leading to persistence in effort. Thus we hypothesize: Hypothesis 2: Intrinsic motivation for research activities will be positively related to research activity. Information Seeking The socialization literature describes newcomer information seeking as a way for individuals to compensate for the lack of perfect role-related information provided by an organization (Miller & Jablin, 1991). Information seeking is important because it reduces uncertainty, thus allowing the employee to more successfully understand, predict, and control the environment (Morrison, 1993). Of course, it stands to reason that this behaviour may continue to be instrumental for employees even after they become more integrated into the organization. For example, it is often the case that an employee's job duties will change or that the work is particularly complex. In the current context, we suggest that student research activities represent a case in which both of these are true – tasks are often changing and the work is complex. Amabile (1983, 1996) suggests that creativity is partially determined by domain skills – an individual's available mental data for solving a problem or completing a task that is comprised of all domain knowledge including facts, principles, technical skills, and opinions. http://www.senecac.on.ca/quarterly/2005-vol08-num01-winter/dewett_shin_toh_semadeni.html College Quarterly - Winter 2005 Page 6of 22 Thus we suggest that domain knowledge can be strengthened through information seeking, a contention that has not yet been explored in the literature. Morrison (1993) suggests that employees may seek five types of information: technical, referent, normative, performance, and social. However, given the focus on the research efforts of doctoral students, four aspects appear most salient. First, students are likely to seek normative information about expected behaviours and attitudes relative to research. Next, as they become involved in research projects, a student may seek technical information as he or she attempts to build research skills. In addition, performance feedback information is likely to be sought to enable the student to gauge how well research efforts are being perceived in general. Finally, more specific research product- related feedback is likely to be sought as students desire to have their research papers reviewed and critiqued by faculty members. In short, it is likely that the receipt of these types of information will support domain skills and facilitate the student's ability to generate a body of research activity. Thus, we propose: Hypothesis 3: Information seeking will be positively related to research activity. The Mediating Role of Research Activity Earlier, the fundamental positive relationship between research activity and research products was discussed based on the idea that finished products are the direct result of a body of work in progress over time (e.g., Albrecht & Hall, 1991; Bauer & Green, 1994; Drazin et al., 1999). Subsequently, we described the way in which three factors may influence a student's research activity. Thus, as shown in Figure 1, given that creative outcomes are predicated on significant creative efforts, it is theoretically plausible that research activity is the mechanism through which the antecedents discussed above, influence research products. Thus, we offer a final hypothesis: Hypothesis 4: Research activity will mediate the relationship between the indicators noted above (chair goals, intrinsic motivation, and information seeking) and research products. Methods Respondents and Procedures A sample of later stage management doctoral students was identified in October and November 2001 through the Academy of Management's placement website as well as several list serve electronic mailing addresses (e.g., Academy of Management Student List Serve). An electronic mail message was sent to these individuals informing them of the nature of our research and asking them to participate. Specifically, we invited only those individuals who were currently ABD (all but dissertation) status. This ensured that the respondents were advanced enough in their programs such that all of http://www.senecac.on.ca/quarterly/2005-vol08-num01-winter/dewett_shin_toh_semadeni.html College Quarterly - Winter 2005 Page 7of 22 our measures would be relevant to them. A hyperlink was included in the message allowing those choosing to participate to connect to a survey on the World Wide Web. Use of web-based surveys has become widespread as Internet use has grown because of cost and speed advantages (Simsek & Veiga, 2001). Further, data collected via web-based surveys rivals traditional mail survey data in several ways including the incidence of nonresponse errors and item variability (Stanton, 1998). The population of ABD candidates in Management during 2001 when the survey was administered was approximately 350. After one follow up electronic mail message, a total of 67 individuals responded, representing a 19% responding rate. After removing those providing incomplete information, the number was reduced to 59 – representing 40 different management departments. Our sample was diverse in terms of gender (38 males, 21 females), age (12 under 30, 29 between 30 and 39, 15 over 40, 3 not reporting), departments of management represented, and research activity/products. The list of institutions represented in the sample is shown in the Appendix. Measures Research activity. Bauer and Green's (1994) measure of research activity was adapted for use in the study. The seven item, five point, Likert-type scale contains items asking the respondent about the degree to which they have engaged in various research activities over the course of doctoral studies (1, not at all; 5, very often). These include such things as working with a faculty member on his or her research project, writing a research paper as a sole author, collecting data for a research project, and analyzing data for a research project. Cronbach's Alpha for this measure was .77. Research products. In order to measure research products we adopted an approach that would ensure adequate variance and consideration of both quantity and quality dimensions. First, for each respondent, we counted each of the following types of research products: refereed publications, revise and resubmit manuscripts, under review manuscripts, other publications (e.g., book chapters), and national conference presentations – given that conference presentations often represent a lower hurdle for students and because most students are supported in sharing their research at conferences (Golde & Dore, 2001). Next, weights were developed to reflect the quality of each research outcome. The weights were derived from a survey of six faculty members at a major research institution each of whom had served as an editorial board member for at least one leading management journal. The first weight concerned the quality of the research outlet. Each of the faculty members was asked to sort the list of journals represented in our sample into 'A', 'B', and 'C' tiers ('A' representing the highest quality) and to assign each journal a score from 1-100. Based on the relationship between the numerical ratings http://www.senecac.on.ca/quarterly/2005-vol08-num01-winter/dewett_shin_toh_semadeni.html College Quarterly - Winter 2005 Page 8of 22 across tiers, the following weights were assigned to the outlets: 'A' tier journals were assigned the value 10, 'B' tier journals 5, and 'C' tier journals 2. Book chapters and national conference presentations were rated 1.5 and 1 respectively to denote their relatively lower significance. The second weight addressed the type of research outcome. The judges were asked to rate several generic research outcomes such as 'A publication', 'A revise and resubmit', 'A under review', other publications, and national conference presentation on a scale from 1-100. Following the same procedure, the following values were assigned to different types of outcomes: publication, 2; revise and resubmit, 1.5; under review, 1; other publications, 1.25; and national conference presentations, 1. Finally, the two weights were applied to the research product counts and the results for each respondent's research outcome were then added to arrive at one research product score for each respondent. The measure was normally distributed with a mean of 21.6 and a standard deviation of 18.6. The research product scores were distributed as follows: 12 below 10, 19 between 11 and 20, 14 between 21 and 30, and 14 above 30. Goals set by the student's chair. To measure the extent to which research-related goals were set for the student by his or her chair, a three item, seven point, Likert-type scale (1, strongly disagree; 7, strongly agree) was developed. The items were designed to measure the extent to which the student's chair had assigned them specific and challenging goals (i.e., obtaining what are typically considered difficult and valued goals in a management Ph.D. program). The items include: 'My chair told me that one of my goals should be to attain as many quality publications as possible while I am a doctoral student', 'My chair told me that one of my goals should be to attain as many conference presentations as possible while I am a doctoral student', and 'My chair told me that one of my goals should be to gain employment at a research-oriented institution after I leave my doctoral program'. Cronbach's Alpha for this measure was .84. Intrinsic motivation. The scale used to measure intrinsic motivation was developed for this study. It consists of five items using a seven point Likert-type scale (1, strongly disagree; 7, strongly agree). The questions were designed to measure the respondent's interest and enjoyment of research as a part of the doctoral program. Sample items include: 'I enjoy the various tasks involved in conducting research projects' and 'I feel driven to do research because I genuinely like the tasks involved'. Cronbach's Alpha for this measure was .89. Information seeking. Information seeking was measured with an eight item, seven point (1, strongly agree; 7, strongly disagree), Likert-type scale adapted from Morrison (1993). Two items each measure four dimensions: normative information, technical information, performance information, and seeking advice or critiques from faculty of research work. Items include: 'When I am developing http://www.senecac.on.ca/quarterly/2005-vol08-num01-winter/dewett_shin_toh_semadeni.html College Quarterly - Winter 2005 Page 9of 22 my work I often seek advice and critiques from the faculty' and 'I have sought out information from the faculty concerning the use of statistical techniques or methodologies'. Cronbach's Alpha for this measure was .84. Prior research experience. Prior research experience was included in the study as a control variable. Bauer & Green (1994) found that prior research experience was positively related to doctoral students' current research activities, professional involvement, and publication success. We adapted Bauer & Green's (1994) measure to better fit a management student sample. The measure included eight items using a five point Likert-type scale (1, not at all; 5, very often) and asked the respondents' how often they had engaged in various research activities prior to starting their doctoral program. Items include such activities as working with a faculty member on his or her research project, conducting a research project alone, submitting a journal article, and having an article published. Cronbach's Alpha for this measure was .79. Department ranking. A ranking of the quality of the management department where each respondent was trained was included as a control variable. We included a department ranking so that we could control for other school-related variables that might influence the relationships we examined (e.g., the influence of faculty colleagues on student research). Trieschmann, Dennis, Northcraft, and Niemi (2000) provided rankings of research productivity for many management departments. This work did not include all schools in our sample, but a larger list of rankings provided by these authors did include all of the schools in our sample and was used in the study. Lower scores indicate higher departmental research productivity. For the forty schools represented in our sample, rankings ranged from 2 to 301. Nine schools received top 30 rankings, 19 schools were ranked between 31 and 100, and 12 schools were ranked above 100. Results Descriptive statistics and correlations among all of the study variables are shown in Table 1 below. The correlations suggested that information seeking was highly related to two other study variables – chair goals and research activity. In order to test the validity of this measure we conducted a series of confirmatory factor analyses (CFAs) on these three measures. We developed a series of 1-factor, 2-factor, and 3-factor models standing in nested sequence (depending on the hypothesized factor structure of the measures) and conducted chi-square difference tests to evaluate which model better fit the data (Anderson & Gerbing, 1988; Kelloway, 1998; Smidts, Pruyn, & Van Riel, 2001). We found that the hypothesized factor structures provided a better fit to the data than any reduced model. For example, the hypothesized 3-factor (research activity, chair goals, and information seeking) model provided a significantly better fit than a 1-factor model (_2 (3) = 20.17, p < .001), a 2-factor (combining research difference activity and information seeking) model (_2 (1) = 22.64, p difference http://www.senecac.on.ca/quarterly/2005-vol08-num01-winter/dewett_shin_toh_semadeni.html College Quarterly - Winter 2005 Page 10of 22 < .001), and another 2-factor (combining chair goals and information seeking) model (_2 (1) = 5.11, p < .05). With the difference exception of one item for the research activity measure, all of the items significantly loaded on their underlying common factors. Consequently, this analysis indicates that information seeking has discriminant validity. TABLE 1 Means, Standard Deviations, and Correlations Variables Mean s.d. 1 2 3 4 5 6 1. Prior Research 1.58 .76 - Experience 2. Department Rank 74.40 76.60 .22 - 3. Research Activity 3.36 .80 -.03 -.26* - 4. Research 21.60 18.66 -.16 -.14 .44** - Products 5. Chair Goals 4.74 1.73 -.09 -.06 .51** .18 - 6. Intrinsic 5.50 1.19 .00 .05 .11 .35** .08 - Motivation 7. Information 5.11 1.06 -.29* -.08 .45** .27* .45** .07 Seeking Note: N=59. *p < .05 **p < .01 Formal tests of the hypotheses were conducted using regression analyses. The results for Hypotheses 1-3 are shown in Table 2 below. TABLE 2 Regression Results – Hypotheses 1-3 Variables R2 _R2 F _ change Step 1 - Control Variables .06 .06 1.58 Prior Research Experience .11 Department Rank -.22 Step 2 - Independent .40** .34** 8.72** Variables Chair Goals .41** http://www.senecac.on.ca/quarterly/2005-vol08-num01-winter/dewett_shin_toh_semadeni.html