Article Review (Obj. 4.2 and 4.4)Select and read one of the following articles, located in the Topic 4 materials:1. The Role of Faculty Mentors in the Research Training of Counseling Psychology Doc

The Role of Faculty Mentors in the Research Training of Counseling Psychology Doctoral Students Merris A. Hollingsworth and Ruth E. Fassinger University of Maryland This study investigated research mentoring experiences of counseling psychology doctoral students as predictors of students’ research productivity. The authors also assessed the research training environment and research self-efficacy as influences on research productivity. Participants were 194 third- and fourth-year counseling psychology doctoral students. Results indicated that the research training envi- ronment predicted students’ research mentoring experiences and their research self-efficacy. Both research mentoring experiences and research self-efficacy mediated the effect of the research training environment on research productivity. Analyses showed no significant differences in these relationships by student gender or scientific stature of training programs. Research training of counseling psychology doctoral students has received increased scrutiny in the last 2 decades. This scrutiny stems, in part, from the observation that few counseling psychol- ogists conduct research after completing their doctoral require- ments despite training in a scientist–practitioner model (Brems, Johnson, & Gallucci, 1996). Although research suggests that in- dividual factors, such as personality and interests, play a major role in research attitudes and productivity (e.g., Kahn & Scott, 1997; Krebs, Smither, & Hurley, 1991; Mallinckrodt, Gelso, & Royalty, 1990), theorists have also proposed that the research training environment plays an influential role in shaping counseling psy- chologists’ perceptions of research (Gelso, 1997).

To describe the role of the research training environment, Gelso (1997) has proposed and empirically tested a model. The research training environment model hypothesizes nine themes central to research training, which include (a) teaching students that all research is flawed, (b) teaching students to look inward for re- search ideas, (c) helping students understand the connection be- tween science and practice, (d) teaching varied methodologies, (e) teaching statistics in ways that are relevant to research applica- tions, (f) faculty modeling of appropriate scientific behavior and attitudes, (g) providing positive reinforcement of scientific activ- ity, (h) involving students in research activities early in graduate training, and (i) viewing participation in science as a partially social activity. A number of empirical studies have indicated that the research training environment model describes critical ele- ments that differentiate between research training programs (Gelso, Mallinckrodt, & Judge, 1996; Kahn & Scott, 1997; Roy- alty, Gelso, Mallinckrodt, & Garrett, 1986).

Studies also consistently show positive relationships between the research training environment, students’ research self-efficacy,and students’ research productivity. For example, Krebs et al.

(1991) found a positive relationship between students’ perceptions of the research training environment and subsequent research productivity. Investigations also supported positive relationships between the research training environment and students’ research self-efficacy (Bishop & Bieschke, 1998; Phillips & Russell, 1994).

Further analyses suggested that research self-efficacy mediates the relationship between the research training environment and stu- dents’ research productivity; that is, the training environment affects productivity indirectly, through the influence of the training environment on students’ research self-efficacy (Brown, Lent, Ryan, & McPartland, 1996; Kahn & Scott, 1997).

In addition, previous research suggests that student gender mod- erates the research training environment, self-efficacy, and pro- ductivity relationship, with significantly different relationships existing between these variables for men and for women. Specif- ically, Brown et al. (1996) found that research self-efficacy had a significantly stronger effect on research productivity for male students than for female students; in contrast, the research training environment had a greater direct effect on productivity for female than for male students. Evidence from Kahn and Scott (1997) also supports student gender as a possible moderator, with males re- porting higher research self-efficacy than females.

The literature also supports expressed interest in research as a predictor of research productivity (Kahn & Scott, 1997; Parker & Detterman, 1988; Royalty & Magoon, 1985). Although the re- search training environment and research self-efficacy appear to influence students’ interest in research (Bishop & Bieschke, 1998; Kahn & Scott, 1997), students also have had interest and experi- ences with scientific inquiry that they developed prior to their doctoral program training (Gelso, 1997). Varied levels of prior research interest reflect individual differences that students bring to their doctoral programs. Although data have supported the effects of the research training environment in changing students’ level of research interest (Mallinckrodt et al., 1990; Royalty et al., 1986), the extent to which prior levels of research interest may influence students’ later research productivity is unclear.

Although mentoring is not a specific focus of the research training environment literature, faculty mentoring emerges as a consistently important undercurrent in the research training envi- Merris A. Hollingsworth and Ruth E. Fassinger, Counseling and Per- sonnel Services, University of Maryland.

The data were collected by Merris A. Hollingsworth as part of her doctoral dissertation.

Correspondence concerning this article should be addressed to Merris Hollingsworth, who is now at the Center for Counseling and Student Development, 261 Perkins Student Center, University of Delaware, New- ark, Delaware 19716. E-mail: [email protected] Journal of Counseling PsychologyCopyright 2002 by the American Psychological Association, Inc.

2002, Vol. 49, No. 3, 324 –3300022-0167/02/$5.00 DOI: 10.1037//0022-0167.49.3.330 324 ronment.For example, Gelso (1993) outlined specific faculty behaviors associated with good research-related mentoring. In this description, faculty offer interpersonal reinforcement for research activity, express enthusiasm for science and research, acknowl- edge the inevitability of flaws in research, expose students to a variety of research methods, model a balance of science and practice, and use relationship skills that communicate empathy, positive regard, and genuineness to students.

However, some researchers have critiqued research training environment theory in regard to mentoring, suggesting that faculty mentoring should be a more explicit element of the research training environment. For example, Hill (1997) compared the role of the faculty–student mentoring relationship in research training to that of the working alliance between the counselor and the client, which prompted her suggestion that the faculty–student mentoring relationship itself may be an essential ingredient in the research training environment. Similarly, Mallinckrodt (1997) rec- ommended a systemic perspective, in which each advisor–student relationship is considered as a“micro-environment”that exists within the larger contexts of a department and institution.

Other researchers have voiced similar support for the important role of mentors in research training, although not within the specific context of the research training environment. For example, Royalty and Reising’s (1986) data indicated that research activities involving interaction with role models or an advisor were among the strongest positive influences on interest in research. O’Brien (1995) and Gelso (1997) both noted that student responses to open-ended questions about critical incidents in their research training often focused on their relationships with faculty members.

Several studies suggest that faculty modeling or mentoring in research activities corresponds with higher rates of research in- volvement and productivity among psychology students and recent graduates (Cronan-Hillix, Gensheimer, Cronan-Hillix, & David- son, 1986; Galassi, Brooks, Stoltz, & Trexler, 1986; Krebs et al., 1991).

Despite these associations, no studies could be identified that focused specifically on research-related mentoring in counseling psychology. The literature suggests that graduate students believe that having a mentor is a critical component of graduate training (Atkinson, Neville, & Casas, 1991; Lark & Croteau, 1998; Luna & Cullen, 1998). Many psychology graduate students also appear to have mentors during their training; two studies among this popu- lation found that more than half of respondents reported having a mentor during their graduate work (Cronan-Hillix et al., 1986; Mintz, Bartels, & Rideout, 1995). Furthermore, Atkinson et al.

(1991) surveyed ethnic minority psychologists and found that respondents recalled their faculty mentors’encouragement related to research involvement as important and useful.

A few studies have investigated outcomes associated with research-related mentoring in other academic disciplines. Green and Bauer’s (1995) study of doctoral students in the physical sciences showed little relationship between research mentoring and students’research productivity after controlling for partici- pants’research interest prior to graduate school. In contrast, Cronan-Hillix et al. (1986) found a significant relationship be- tween receipt of mentoring and several measures of research productivity. The lack of additional studies in academic settings that explore outcomes associated with mentoring contrasts sharply with studies of mentoring in business settings, where measures such as rate of promotion, salary increases, and job satisfaction areconsistently correlated with receipt of mentoring (e.g., Bahniuk, Dobos, & Kogler Hill, 1990; Bowen, 1985; Turban & Dougherty, 1994; for a more complete review of this literature, see Noe, 1988a).

The current study extended the investigation of research training in counseling psychology by exploring the role that faculty re- search mentoring plays in predicting student research productivity, above and beyond the contributions of the research training envi- ronment, students’research self-efficacy, and students’past re- search attitudes. Five research questions guided our work:

1. Does the research training environment predict students’ research mentoring experiences, their research self-efficacy, or their research productivity?

2. Do students’research mentoring experiences mediate the relationship between the research training environment and productivity?

3. Do students’self-efficacy beliefs mediate the influence of the research training environment on research productivity?

4. Does controlling for students’past attitudes toward research significantly change the relationships between research training environment, self-efficacy, research mentoring, and research productivity?

5. Are relationships between these variables moderated by stu- dents’gender or by the scientific stature of their training program? Method Participants Participants were 194 (135 women and 59 men) third- or fourth-year students enrolled in 25 APA-approved counseling psychology programs.

Only students working toward a PhD participated in the study, and the response rate was 70%. The majority of the participants identified them- selves as European American (71%), and 12% identified as African Amer- ican/Black, 5% as Hispanic/Latino/Latina, 4% biracial, 3.5% Asian Amer- ican, 2% international students, and 1.5% unspecified. Ninety-five percent of the respondents categorized themselves as third- or fourth-year doctoral students, whereas the remaining 5% included second-, fifth-, and sixth-year students. The ages of the participants ranged from 23 to 58 years (M 31.08 years,SD 6.36 years). Participants from high and medium research productivity programs comprised the majority of the sample (38% and 36%, respectively), with 26% coming from low research productivity programs. More than half of the respondents (57.5%) indicated that they currently participated in an active research team, and 72% considered themselves as currently having a research mentor. Students who did not have a research mentor were instructed to“consider the faculty relationship that has beenmostimportant in your research training while in your current doctoral program”when answering questions.

Instruments Independent variables.The research training environment was as- sessed by a modified version of the Research Training Environment Scale—Revised (RTES–R; Gelso et al., 1996). The original instrument contains nine subscales measuring the following: teaching relevant statis- tics, facilitating students“looking inward”for research ideas, teaching that all experiments are flawed and limited, focusing on varied investigative styles, wedding science and clinical practice, faculty modeling of appro- priate scientific behavior, faculty reinforcement of student research, stu- dents’early involvement in research, and science as a partly social expe- rience. Items ask students to rate their doctoral program in each of these areas. Test–retest reliabilities for each subscale range from .74 to .94, and the subscales consistently correlate with changes in research attitudes 325 RESEARCH MENTORS during graduate training and with research self-efficacy (Gelso et al., 1996). The current study used a modified 16-item version of the RTES–R.

First, the three items with the highest factor loading on each subscale were selected for this study on the basis of factor analyses conducted by Kahn and Gelso (1997), in which all items were forced to load on one of the nine subscales. This step yielded an abbreviated version with 27 items. To avoid a potential problem of item overlap between this instrument and the Research Mentoring scale (described below), we omitted 11 additional items from the RTE measure because they addressed the role of the faculty advisor (e.g.,“I feel that my faculty advisor expects too much from my research projects”). Participants responded on a 5-point Likert scale rang- ing from 1 to 5, with higher numbers indicating a greater level of agree- ment with each statement. Responses were added to yield a total score, with potential scores ranging from 16 to 80. Cronbach’s alpha for the RTE measure was .87 in the current study.

We measuredresearch mentoring experienceswith the Research Men- toring Experiences Scale (RMES), a measure created for this study that is based on comparable instruments developed for business settings (e.g., Noe, 1988b; Ragins & McFarlin, 1990). The RME included two subscales.

The first subscale, Psychosocial Mentoring, includes 13 items that explored the affective aspects of research training, focusing on the personal elements of the relationship between faculty member and student. Participants indi- cated the extent to which a specific faculty member expressed emotional support, communicated respect and personal regard, and modeled positive attitudes toward research. The second subscale, Career Mentoring, inves- tigated faculty members’efforts to help students acquire specific informa- tion necessary to complete research tasks successfully. The 16 items on this subscale explored faculty members’teaching of research skills, giving advice, and providing research opportunities. For both psychosocial and career mentoring, instructions asked respondents to rate their relationship with the faculty member whom they considered most important in their current doctoral research training. Possible responses ranged from 1 (fac- ulty member pays very little attention to...)to5(faculty member pays a great deal of attention to...).Responses to items were added and divided by the number of items to generate a total score. Possible scores ranged from 1 to 5. The RMES was initially tested and revised in a pilot study (n 25); Cronbach’s alpha in the current study was .74.

Research self-efficacy, hypothesized as a second mediating variable, was measured by a shortened version of the Self-Efficacy in Research Measure (SERM; Phillips & Russell, 1994). As previously adapted by Kahn and Scott (1997), the shortened version of Phillips and Russell’s measure includes 12 items asking doctoral students to describe their confidence in applying four types of research-related skills: research design, practical research skills, quantitative and computer skills, and writing skills. In this study, participants indicated their responses on a 5-point Likert scale, ranging from 1 (no confidence)to5(total confidence). Each response was added to yield a total score, with potential scores ranging from 12 to 60.

This instrument yielded high internal consistency (.90) in previous research (Kahn & Scott, 1997) and in the current study ( .87).

Past attitudes toward researchwas measured by the four items con- structed by Royalty et al. (1986). These items measured counseling psy- chology students’recalled interest in conducting research prior to their enrollment in the doctoral program. The items included the following: (a) “I would have preferred to have the option of completing my doctoral training without being required to complete research projects”(Prefer- ence), (b)“I had a strong interest indoingresearch”(Interest), (c)“I placed a high value on the place of research in my future career”(Value), and (d) “Participating in research activities after graduation wasnota major priority for me”(Priority). Participants rated their level of agreement with each item, using a 5-point Likert scale, which ranged from 1 (strongly disagree)to5(strongly agree), and the first and last items were reverse- scored. Responses were added, then divided by the number of items to produce a final score, with a potential range from 1 to 5. Previous research shows good internal consistency for the scale, with alpha ranging from .87 to .90 (Gelso et al., 1996; Kahn & Scott, 1997; Royalty et al., 1986), andthe test–retest correlation for this measure was .93 (Royalty et al., 1986).

Cronbach’s alpha in the current study was .89.

Dependent variable.The dependent variable,research productivity, was assessed using Kahn and Scott’s (1997) 8-item measure. These items provided a broad measure of students’involvement in research-related activities, including collection and analysis of data, development of manu- scripts, participation in public presentations, and attendance at research conventions (Kahn & Scott, 1997). Students responded to each item by providing a number indicating the number of projects for which they are currently collecting or analyzing data, the number of manuscripts they have completed or are now working on, and so forth. Responses were summed to obtain a total number, with potential scores ranging from zero to infinity.

The current study yielded responses ranging from zero to 40, with a modal score of 6. Internal consistency coefficients (K-R 20) for this scale ranged from .59 to .72 (Kahn & Scott, 1997), and in the current study Cronbach’s alpha was .75 for this measure.

A demographic data form requested information about participants’ gender, race, year in doctoral program, and age. Additional items asked if students were currently part of an active research team and if they currently had a research mentor. Procedure The instruments for this study included a demographic form and a survey booklet with the scales administered in the following sequence:

Attitudes Toward Research, Research Productivity, Research Training Environment, Research Self-Efficacy, and Research Mentoring Experi- ences. To incorporate participants from training programs with varying levels of emphasis on research, the 63 currently active, APA-accredited counseling psychology programs were stratified into three groups on the basis of their scientific stature: high, medium, or low. An evaluation of counseling psychology doctoral programs based on faculty research pro- duction (Hanish, et al., 1995) served as a guideline for each training program’s designation within a category. Nine programs were selected from each category (high, medium, or low scientific stature), with attention to geographical and institutional diversity. The researchers’home univer- sity was omitted from the participant pool to avoid potential bias. Three program directors declined to participate, and only one replacement pro- gram could be identified on short notice, resulting in a sample representing 25 programs: 9 high, 9 medium, and 7 low scientific stature training programs.

To identify participants, counseling psychology program directors were asked to provide names and mailing addresses of third- and fourth-year doctoral students in the spring of 1998. Potential participants (N 278) received an advance postcard inviting their participation, followed by a mailed survey packet and personalized cover letter 1 week later, accom- panied by a stamped return envelope. Each survey packet was numerically coded to permit accurate follow-up. As an incentive to participate, we included a pencil with each test packet. To maximize the return rate, potential participants who had not yet returned their survey received a reminder postcard 10 days after the mailing of the original survey, and a follow-up letter and additional survey packet 2 weeks after the mailing of the reminder postcard. Nonrespondents received one final request for participation by mail 2 weeks after the mailing of the second survey packet.

In addition, survey packets were distributed anonymously by two program directors who were unable to release student names. No follow-up mailings were sent to these two sites. Overall, 200 students responded. Data from 6 surveys were incomplete and could not be used; thus, the final number of participants was 194. Results In preliminary analyses, we noticed strong positive skew on the Research Productivity scale. To manage this difficulty, we used a logarithmic transformation to adjust each score, which yielded a 326 HOLLINGSWORTH AND FASSINGER more normal distribution. The data in Table 1 and all subsequent analyses used the transformed productivity scores. Means, stan- dard deviations, and zero-order correlations for the five variables are indicated in Table 1.

Multiple regression analyses were used to investigate the re- search questions. We used hierarchical regression to investigate research mentoring experiences as a mediator of the research training environment’s influence on students’research productiv- ity. According to Baron and Kenny (1986), several conditions are needed to support mediation: (a) the independent variable (training environment) must be related to the mediator (mentoring), and (b) the independent variable (training environment) must be related to the dependent variable (productivity). In addition, when the me- diator (mentoring) and the independent variable (training environ- ment) are added, in subsequent steps, to a regression equation predicting the dependent variable (research productivity), the re- gression coefficient for the independent variable (training environ- ment) should decrease when the effects of the mediator (mentor- ing) are partialed out. Research mentoring would be considered a “perfect”mediator if the training environment had no effect on productivity when mentoring is controlled (Baron & Kenny, 1986). The first two conditions were supported by correlation coefficients (see Table 1), so we proceeded with a regression equation in which research mentoring experiences was added in the first step, followed by research training environment in the second step. This analysis supported research mentoring experi- ences as a mediator ( .45,p .001) because the research training environment became a nonsignificant predictor of re- search productivity ( .03,p .60).

Similarly, hierarchical regression was used to analyze research self-efficacy for a mediating effect between the research training environment and research productivity. As above, the two prelim- inary conditions were satisfied by correlation coefficients (i.e., training environment and self-efficacy were related, and training environment and productivity were related). Regression supported the mediational hypothesis: Research self-efficacy predicted re- search productivity ( .36,p .001), whereas the research training environment coefficient decreased ( .07,p .30) after self-efficacy was partialed out.

In a third analysis, we explored the role of research mentoring experiences and research self-efficacy when we controlled for students’past attitudes toward research. Past research attitudes was entered in the first step of the hierarchical regression. Past research attitudes emerged and remained a significant predictor of researchproductivity ( .38,p .001) despite the addition of research training environment, research mentoring experiences, and re- search self-efficacy to the regression equation. Although past research attitudes explained an additional 10% of students’re- search productivity, research mentoring experiences and research self-efficacy remained significant predictors ( .38,p .001, and .28,p .001, respectively) of research productivity.

We tested student gender as a potential moderating variable of the relationships noted earlier with additional regression equations.

In the first hierarchical regression, we entered gender as a dummy variable, followed by entry of the research training environment and research mentoring experiences in subsequent steps. We then analyzed the two- and three-way interactions among these vari- ables. A significant interaction term would suggest that student gender acts as a moderator, affecting the strength and/or direction of the relationship between the independent variables (Baron & Kenny, 1986). As Table 2 indicates, student gender was not a significant predictor of research productivity, and the two- and three-way interaction terms with mentoring and the training envi- ronment also were not significant. We completed a similar regres- sion analysis with student gender, research self-efficacy, and the research training environment and found no significant interaction effects. Similar analyses showed no significant differences based on scientific stature of students’programs (high, medium, or low) in the relationships among research training environment, research mentoring, and research self-efficacy. Results are reported in Table 2. Discussion This study built on previous research describing the research training environment and its effects on counseling psychology doctoral students by addressing fundamental questions about stu- dents’research mentoring experiences as a potentially important addition to the research training environment. Previous research supports the research training environment, research self-efficacy, and students’past research attitudes as predictors of students’ research productivity. The current study incorporated these estab- lished variables and investigated research mentoring experiences as an additional influence on productivity. Consistent with previ- ous research, analyses supported the role of the research training environment, research self-efficacy, and past research attitudes as direct predictors of productivity.

The data also suggested that students’mentoring experiences serve as an important predictor of research productivity, mediating the relationship between the research training environment and research productivity. This finding supports recent assertions that faculty mentoring is a critical component within the research training environment as a whole (e.g., Gelso & Lent, 2000; Hill, 1997) and provides additional evidence that students’experiences with faculty research mentors are important to students’develop- ment as researchers. The strong correlation between the research training environment and research mentoring experiences supports the logical proposition that a strong research training environment is most likely to promote strong research mentoring relationships.

However, the mediating role of research mentoring in the predic- tion of research productivity suggests that a research mentoring relationship is the vehicle through which the training environment has greatest impact on individual students’research production. If this is the case, then working to improve student–faculty research Table 1 Zero-Order Correlations, Means, and Standard Deviations of Variables Variable 1 2 3 4 5 1. Research training environment— 2. Research monitoring experience .46**— 3. Research self-efficacy .34** .22**— 4. Attitudes toward research .04 .15* .31**— 5. Research productivity .19** .45** .36** .37**— M56.27 3.18 45.85 3.28 2.20 SD11.94 0.82 7.46 1.17 0.76 Note. N 194.

*p .05. **p .01. 327 RESEARCH MENTORS mentoring may be an important step toward promoting greater research productivity among counseling psychology doctoral students.

Students’research self-efficacy served as another mediator be- tween the research training environment and research productivity.

This result supports earlier findings by Brown et al. (1996) that the training environment influences productivity indirectly through its effects on students’self-efficacy.

When we controlled for students’past attitudes toward research, students’mentoring experiences and research self-efficacy re- mained as influential predictors of research productivity. Although this finding supports the importance of individual differences that students bring to their doctoral programs, it also underscores the effects of environmental and interpersonal factors, such as men- toring relationships and research self-efficacy, during doctoral training. Although one may argue that students who enter a train- ing program with high research interest are more likely to seek research mentoring and to develop research self-efficacy, these data suggest that environmental interventions to support research mentoring and development of students’research self-efficacy have effects exceeding those associated solely with students’past level of interest. This result supports earlier suggestions (e.g., Gelso, 1997) that environmental interventions may play a decisive part in promoting student research activity.

Participants’gender did not significantly influence these rela- tionships among the variables of interest. This outcome diverges from results of previous research, which showed male studentsexperiencing research self-efficacy as a greater influence on re- search productivity than their female peers (e.g., Brown et al., 1996). The absence of significant findings by gender for research self-efficacy in the current study may reflect restricted range in participants’responses because we used a 5-point scale for re- search self-efficacy responses, compared with the 10-point scale used in the earlier studies (Brown et al., 1996; Kahn & Scott, 1997). Our results also showed no difference by gender in the effects of the research training environment on productivity, con- trasting with a previous study (Brown et al., 1996). Although the abbreviated version of the research training environment measure used in our study was based on earlier work (Gelso et al., 1996; Kahn & Gelso, 1997) and showed internal consistency comparable to the full scale, perhaps the smaller number of items limited the range of participant responses, leading to no effects by gender.

However, the lack of significant effects by gender in our study also may reflect changes that have occurred in many counseling psy- chology training programs because the data for Brown et al.’s analyses were collected in the early 1990s. As many training programs have experienced an influx of female students, perhaps the research training environments have shifted, leading to fewer differences in the research training experiences of men and women. Because not many published studies have analyzed par- ticipant gender and our findings conflict with earlier work, these results highlight the critical need for investigation of potential differences and similarities by gender in future research.

Our analyses also showed no differences between male and female students in the effects of research mentoring on productiv- ity. This result suggests that mentoring plays an equally important role for students, regardless of gender; however, this finding should be interpreted cautiously given the absence of prior work in this area. As above, we note the importance of further research, with attention to possible differences in the relationships between variables as a function of student gender.

We also investigated potential differences in the effects of the training environment, mentoring, and research self-efficacy on research productivity by the scientific stature of participants’train- ing programs. The absence of significant differences suggests that these variables function similarly to support research productivity in all counseling psychology doctoral programs, regardless of the scientific stature of the program. This result should be considered preliminary because other studies have not used a comparable stratification system in selecting programs to sample. However, if supported by further study, this finding lends support to the uni- versality of the research training environment, research mentoring, and research self-efficacy as constructs that are fundamental com- ponents of the research training process. Strengths and Limitations This study has several strengths that underscore its value to counseling psychology. This work provides a conceptual bridge linking two areas—the research training environment and mentor- ing relationships—that had not yet been combined empirically.

The research demonstrates that linking these two areas is critically important in understanding the research training process. The extent of the sampling and the high return rate support the gener- alizability of these findings to upper level doctoral students in counseling psychology. Our study also generates numerous ques- tions for further research. Table 2 Summary of Hierarchical Regression Analyses Predicting Research Productivity VariableFp First equation Gender 0.10 .75 Research mentoring experience (RME) 9.66 .00** Research training environment (RTE) 3.14 .08 Gender RME 0.00 .95 Gender RTE 0.06 .81 Gender RTE RME 2.18 .12 Second equation Gender 0.27 .60 Research self-efficacy (RSE) 4.75 .03* RTE 2.46 .12 Gender RSE 0.09 .76 Gender RTE 0.15 .70 Gender RSE RTE 1.13 .33 Third equation Scientific stature 2.08 .13 RME 8.29 .00** RTE 1.72 .19 Scientific Stature RME 1.98 .14 Scientific Stature RTE 1.77 .17 Scientific Stature RTE RME 1.72 .17 Fourth equation Scientific stature 2.27 .11 RSE 2.82 .10 RTE 1.03 .31 Scientific Stature RSE 1.92 .15 Scientific Stature RTE 2.46 .09 Scientific Stature RTE RSE 1.77 .15 Note. N 194.

*p .05. **p .01. 328 HOLLINGSWORTH AND FASSINGER Several limitations also must be considered in interpreting these results. First, the data collected were cross-sectional and correla- tional. Thus, perceptions of past research interest were based on participants’recollections, which easily may be blurred by current experiences. In addition, mentoring relationships, like most rela- tionships, are likely to change over time. For example, a student in the early stages of developing a mentoring relationship is likely to have a different perspective than when the relationship is well established. Similarly, a student who has a long history with a specific mentor may find their relationship becoming more colle- gial as the student approaches the status of a professional peer.

Relying on cross-sectional data provides only a brief snapshot of students’experiences, which may result in omission of important information. A future study that incorporated a longitudinal design could address some of these concerns.

Second, the measures relied solely on self-report by student participants. The data did not corroborate students’perceptions of their research training environment, research productivity, or their mentoring relationships from additional sources. Additional re- search would greatly benefit from study of paired observations regarding these variables, for example, comparisons of student and faculty mentor responses. Furthermore, the study design and ac- companying analyses assume independence among respondents.

Despite random sampling of training programs, clusters of respon- dents were enrolled in the same doctoral program and shared the same research training environment. Consequently, one might find some homogeneity within clusters, based on students having met similar admission criteria and selecting the same research training program environment (Kish, 1965). Lack of independence may magnify the relationships between variables. This problem could be corrected by conducting analyses at the program level; how- ever, the sample size in this study was insufficient to support this level of analysis.

Finally, the absence of established measures to explore mentor- ing relationships prompted use of a new, unproven instrument to assess psychosocial mentoring in this study. Although the measure was revised on the basis of pilot study data and achieved an adequate measure of reliability in this study, the data should be regarded with some caution given the lack of established reliability and validity evidence for this instrument. This limitation points clearly to the need for more instrument development in the study of academic mentoring relationships. In addition, the research training environment measure was greatly abbreviated from Gelso et al.’s (1996) original version. Although an adequate reliability coefficient was obtained for the measure used in this study, results should be interpreted with the awareness that this abbreviated version is not an established use of this scale.

Implications for Research and Practice This study suggests a number of avenues for further research.

Additional research is warranted to explore the role of faculty mentoring within the research training environment and as a con- tributor to students’research productivity. Consistent with this suggestion, further efforts to develop and refine instruments to assess faculty mentoring are particularly needed. Several well- established instruments to describe mentoring exist in business settings, but few attempts have been made to create comparable measures for academic settings. This study focused exclusively on dyadic mentoring relationships between faculty and students. Ad-ditional research could investigate the effects of group mentoring, such as that received through research team experiences or peer mentoring by research-oriented classmates. In addition, specific models of mentoring, such as feminist mentoring (Fassinger, 1997), could be explored to investigate aspects of mentoring approaches, such as traditional, hierarchical approaches versus more collaborative styles. It is also important to note that meth- odological diversity is needed in both the research training envi- ronment and mentoring literatures, incorporating methods that permit longitudinal assessment of change. Methodological diver- sity should include a wider range of perspectives than self-report by students and use both quantitative and qualitative approaches.

Further exploration of potential similarities and differences in the research training environment and students’research self-efficacy by gender are also needed, particularly as training programs begin to reflect greater numbers of female faculty and female students.

Because this study applies specifically to the interactions of doctoral students and faculty, the implications for practical appli- cation are tailored to those who learn and teach in doctoral training programs. For prospective doctoral students, the results suggest that students who value research training should explore the re- search training environment of programs they are considering because the overall research training environment influences indi- vidual research mentoring relationships and development of re- search self-efficacy beliefs. At the same time, students also need to consider the person variables (e.g., past level of research interest) that they bring to their research training as a way to dispel possible expectations that the“right”mentor can create high levels of research interest, self-efficacy, or productivity. For faculty, the results offer some encouragement that efforts to mentor students’ research development are associated with greater research produc- tivity among students. At a more systemic level, the data invite consideration of the research training environment and mentoring activities at the program and departmental level. Training pro- grams that value research may benefit from discussion about the extent to which faculty feel supported in their efforts to be research mentors. Faculty discussions of mentoring may also encourage sharing of strategies or collaborative efforts, which may decrease the likelihood that some faculty will bear a disproportionate bur- den for research mentoring of students. This type of discussion may also encourage faculty to explore ways that they can manage elements of the mentoring relationships that seem particularly burdensome. Because the research training environment contrib- utes to students’research self-efficacy, faculty in training pro- grams may also wish to consider specific ways in which they, individually and collectively, nurture research self-efficacy among their students. Faculty also may wish to help their own students who plan on careers in academe develop their own skills as research mentors. For example, helping advanced doctoral stu- dents organize and direct their own research team, composed of undergraduate students and graduate peers, provides an opportu- nity to develop research self-efficacy, receive research mentoring, and begin developing skills as a future research mentor.

In summary, this study suggests that research mentoring expe- riences make a notable contribution to students’research produc- tivity. We encourage further innovation, collaboration, and evalu- ation in this area to promote continued development of scientist– practitioners in counseling psychology. 329 RESEARCH MENTORS References Atkinson, D. R., Neville, H., & Casas, A. (1991). The mentorship of ethnic minorities in professional psychology.Professional Psychology: Re- search and Practice, 22,336–338.

Bahniuk, M. H., Dobos, J., & Kogler Hill, S. E. (1990). The impact of mentoring, collegial support, and information adequacy on career suc- cess: A replication.Journal of Social Behavior and Personality, 5, 431–451.

Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations.Journal of Personality and Social Psychol- ogy, 51,1173–1182.

Bishop, R. M., & Bieschke, K. J. (1998). Applying social cognitive theory to interest in research among counseling psychology doctoral students:

A path analysis.Journal of Counseling Psychology, 45,182–188.

Bowen, D. D. (1985). Were men meant to mentor women?Training and Development Journal, 39,31–34.

Brems, C., Johnson, M. E., & Gallucci, P. (1996). Publication productivity of clinical and counseling psychologists.Journal of Clinical Psychol- ogy, 52,723–725.

Brown, S. D., Lent, R. W., Ryan, N. E., & McPartland, E. B. (1996).

Self-efficacy as an intervening mechanism between research training environments and scholarly production: A theoretical and methodolog- ical extension.Counseling Psychologist, 24,535–544.

Cronan-Hillix, T., Gensheimer, L. K., Cronan-Hillix, W. A., & Davidson, W. S. (1986). Students’view of mentors in psychology graduate train- ing.Teaching of Psychology, 13(3), 123–127.

Fassinger, R. E. (1997, August).Dangerous liasions: Reflections on fem- inist mentoring.Paper presented at the 105th Annual Convention of the American Psychological Association, Chicago, IL.

Galassi, J. P., Brooks, L., Stoltz, R. F., & Trexler, K. A. (1986). Research training environments and student productivity: An exploratory study.

Counseling Psychologist, 14,31–36.

Gelso, C. J. (1993). On the making of a scientist–practitioner: A theory of research training in professional psychology.Professional Psychology:

Research and Practice, 24,468–476.

Gelso, C. J. (1997). The making of a scientist in applied psychology: An attribute by treatment conception.Counseling Psychologist, 25,307– 320.

Gelso, C. J., & Lent, R. W. (2000). Scientific training and scholarly productivity: The person, the training environment, and their interaction.

In S. D Brown & R. W. Lent (Eds.),Handbook of counseling psychology (3rd ed., pp. 109–139). New York: Wiley.

Gelso, C. J., Mallinckrodt, B., & Judge, A. B. (1996). Research training environment, attitudes towards research, and research self-efficacy: The Revised Research Training Environment Scale.Counseling Psycholo- gist, 24,304–322.

Green, S. G., & Bauer, T. N. (1995). Supervisory mentoring by advisers:

Relationships with doctoral student potential, productivity, and commit- ment.Personnel Psychology, 48,537–561.

Hanish, C., Horan, J. J., Keen, B., St. Peter, C. C., Ceperich, S. D., & Beasley, J. F. (1995). The scientific stature of counseling psychology training programs: A still picture of a shifting scene.Counseling Psy- chologist, 23,82–101.

Hill, C. E. (1997). The effects of my research training environment: Where are my students now?Counseling Psychologist, 25,74–81.

Kahn, J. H., & Gelso, C. G. (1997). Factor structure of the Research Training Environment Scale—Revised: Implications for research train- ing in applied psychology.Counseling Psychologist, 25,22–37.Kahn, J. H., & Scott, N. A. (1997). Predictors of research productivity and science-related career goals among counseling psychology doctoral stu- dents: A structural equation analysis.Counseling Psychologist, 25,38– 67.

Kish, L. (1965).Survey sampling.New York: Wiley.

Krebs, P. J., Smither, J. W., & Hurley, R. B. (1991). Relationship of vocational personality and research training environment to research productivity of counseling psychologists.Professional Psychology: Re- search and Practice, 22,362–367.

Lark, J. S., & Croteau, J. M. (1998). Lesbian, gay, and bisexual doctoral students’mentoring relationships with faculty in counseling psychology:

A qualitative study.Counseling Psychologist, 20,754–777.

Luna, G., & Cullen, D. (1998). Do graduate students need mentors?

College Student Journal, 32,322–331.

Mallinckrodt, B. (1997). Discovering research training environments that fit the students we select: The challenge of providing for diverse needs.

Counseling Psychologist, 25,68–73.

Mallinckrodt, B., Gelso, C. J., & Royalty, G. M. (1990). Impact of the research training environment and counseling psychology students’Hol- land personality type on interest in research.Professional Psychology:

Research and Practice, 21,26–32.

Mintz, L. B., Bartels, K. M., & Rideout, C. A. (1995). Training in counseling ethnic minorities and race-based availability of graduate school resources.Professional Psychology: Research and Practice, 26, 316–321.

Noe, R. A. (1988a). Women and mentoring: A review and research agenda.

Academy of Management Review, 13,65–78.

Noe, R. A. (1988b). An investigation of the determinants of successful assigned mentoring relationships.Personnel Psychology, 41,457–479.

O’Brien, K. M. (1995). Enhancing research training for counseling stu- dents: Interuniversity collaborative research teams.Counselor Educa- tion and Supervision, 34,187–198.

Parker, L. E., & Detterman, D. K. (1988). The balance between clinical and research interests among Boulder model graduate students.Professional Psychology: Research and Practice, 19,342–344.

Phillips, J. C., & Russell, R. K. (1994). Research self-efficacy, the research training environment, and research productivity among graduate stu- dents in counseling psychology.Counseling Psychologist, 22,628–641.

Ragins, B. R., & McFarlin, D. G. (1990). Perceptions of mentor roles in cross-gender mentoring relationships.Journal of Vocational Behav- ior, 37,321–339.

Royalty, G. M., Gelso, C. J., Mallinckrodt, B., & Garrett, K. (1986). The environment and the student in counseling psychology: Does the re- search training environment influence graduate students’attitudes to- wards research?Counseling Psychologist, 14,9–30.

Royalty, G. M., & Magoon, T. M. (1985). Correlates of scholarly produc- tivity among counseling psychologists.Journal of Counseling Psychol- ogy, 32,458–461.

Royalty, G. M., & Reising, G. N. (1986). The research training of coun- seling psychologists: What the professionals say.Counseling Psychol- ogist, 14,49–60.

Turban, D. B., & Dougherty, T. W. (1994). Role of protege personality in receipts of mentoring and career success.Academy of Management Journal, 37,688–702. Received November 1, 2000 Revision received October 1, 2001 Accepted October 3, 2001 330 HOLLINGSWORTH AND FASSINGER