INDEX:

• Introduction
• Theoretical Framework
• The Findings of Previous Research Studies
• Research Questions For This Study
• Methods
• Measurement of Independent Variables
• Measurement of Dependent Variables
• Results
• Exploratory Analysis
• Discussion
• Implications For Further Research
• Practical Considerations

“Doctoral Advisors and Their Protégés: Does Mentoring Matter?”

By Dr. Laura Gail Lunsford, University of Arizona South

The Editors' Comments

Editors’ note - This paper is the abstract of the dissertation which won the 2009 Hope Richardson “Best Mentoring Dissertation” from the IMA. The award was bestowed at the 2009 IMA Annual Conference at Las Vegas, Nevada on March 5th.

While the article’s focus is higher education advisor mentoring of doctoral students, the Editors feel that many of the findings of this research study have important implications and practical value for mentoring programs in every setting. We commend this article and the editors’ comments along this side to you for your thoughtful consideration.

Introduction

The advisor-student relationship is at the heart of doctoral education in America. Significant resources and organizational efforts are devoted to promoting it. For example, the Council on Graduate Schools lists mentoring as one of six factors leading to Ph.D. completion (Council of Graduate Schools, 2007). Hundreds of millions of dollars are spent on programs, such as the McNair Scholars (US Department of Education, 2007), which foster faculty-student mentoring. The National Science Foundation considers mentoring important enough to confer annually a Presidential Award for Excellence in Science, Mathematics and Engineering Mentoring with a $10,000 grant (National Science Foundation, 2007).

This study examines doctoral education mentoring using a large-scale survey conducted at two major American research universities. It explores specific doctoral supervisory behaviors and the influence of those behaviors on doctoral student outcomes. This is important for two reasons.

1. First, it tests directly whether there is a measurable relationship between mentor behaviors and important student outcomes.

2. Second, it examines conditions that promote or weaken the effects of mentor behaviors on student outcomes.

The results suggest two important findings.

1. First, mentoring is more complex than previously reported in the literature. There are different types of mentoring and each has a distinct path and outcomes.

2. Second, mentoring involves networks. Most doctoral students have a network of supportive and helpful relationships. A clearer understanding of the mentoring relationship should include an appreciation of this network and its operation.

How's THAT for a "wake up call" !?

Attention all you higher education institutions! Check out these clear indications of why YOU too should be making mentoring a KEY ROLE in faculty-student and advisor-student relationships. If you truly want students to succeed, do what has been shown to work!

The author states that this article reviews the results of a study which "examines doctoral education mentoring". This implies that all advisors in the study were performing mentoring functions. As you read, it will become clear that this was not the case in nearly half of the cases.

For our own clarity, perhaps a more accurate statement might be that the study was of "doctoral student advisement for the extent to which it included mentoring".

Theoretical Framework

A variety of theoretical approaches are found in the mentoring literature.

• A resiliency approach (Garmezy, 1985) is the basis for mentoring research on children and youth (Rhodes, 2002)
• while social exchange (Ensher, Thomas, and Murphy, 2001) and,
• developmental approaches (Green and Bauer, 1995) characterize the research on adult mentoring relationships.
• Mentoring researchers have proposed that there may be a variety of mentoring networks, including the traditional one-on-one relationship, which may better describe individuals mentoring relationships (Higgins and Kram, 2001).
• To complicate matters further, most research on doctoral education considers mentoring as one of many factors that influence student retention, rather than an area of independent focus (Council of Graduate Schools, 2004).

A psychosocial, developmental approach is used here because it provides insight into why an individual may be receptive to being mentored and why someone might engage in mentoring. Psychosocial development theory provided the framework for the early work on mentoring (Kram, 1985; Levinson, 1979; Valliant, 1977) and is the basis for most mentoring inventories, including the one used in this study.

There is a developmental window in young adulthood, during the stage of identity development, when individuals most benefit from having a mentor (Erikson, 1968). Individuals must have a need, which can be met by a more experienced person, to want or to benefit from a mentor (Levinson, 1979; Valliant, 1977). This notion of a developmental window comes from identity development theory. Identity development proceeds along two dimensions, commitment (high or low), and exploration (high or low), which yield four identity statuses (Marcia, 1966; Marcia, Waterman, Matteson, Archer, and Orlofsky, 1993). Identity commitment may be an important condition for mentoring receptivity.

This article is one of the best descriptions of dissertation research the Editors have seen in many years. Further, this study is also one of the best constructed the Editors have seen in many years. That is, no doubt why those who reviewed the submitted abstracts gave THIS study the 2009 IMA Dissertation Award !

Never-the-less, for the sake of adequately guiding our readers, we wish to state the following.

Just as in the first "result" stated in the article's section above ("mentoring is more complex than previously reported in the literature"), so is the theoretical framework for mentoring more complex than what is stated in this article.

For example, a significant element of using a developmental approach to mentoring, in any setting and with any age group, is the use of a phased or staged developmental model for the development of persons. Readers who wish more information on this are referred to the articles on the IMA web site regarding the "Concerns-Based Adoption Model" (CBAM).

The Findings of Previous Research Studies

Mentoring is comprised of two functions:

1) Career, e.g. sponsorship, coaching, protection, and providing challenging assignments, and
2) Psychosocial, e.g. role modeling, acceptance, counseling, and friendship (Kram, 1985).

Previous studies have found that these functions exist across disciplines (Ferrer de Valero, 2005; Green and Bauer, 1995), across the years in graduate school (Clark, Harden, and Johnson, 2000; Kahn, 2000; Rose, 2003; Schlosser and Gelso, 2001; Tenenbaum, Crosby, and Gliner, 2001), and across gender (Tenenbaum et al., 2001; Zhao, Golde, and McCormick, 2005).

The remainder of this section reviews the findings related to departments and discipline; frequency; who is considered a mentor; race, gender, and citizenship; student outcomes; and several conditional effects.

Disciplinary and departmental characteristics affect the type and amount of mentoring students report.

• Students in research psychology programs are more likely to have a mentor than students in applied programs (Clark et al., 2000).
• Students in departments with shorter times to degree and higher completion rates are more likely to have a mentor than students in departments with longer times to degree and lower completion rates (Ferrer de Valero, 2005).
• Students in humanities and social sciences are more likely to receive psychosocial support (defined as academic advising and personal touch behaviors) and less career support from their mentors than students in the physical or biological sciences (Zhao et al., 2005).

The frequency of mentoring varies by method used to study it.

• Survey-based studies report 50-75% of doctoral students have a mentor (Clark et al., 2000; Cronan-Hillix, Gensheimer, Cronan-Hillix, and Davidson, 1986; Nettles and Millett, 2006).
• In contrast, mentoring is rare when assessed with in-person interviews (Kram, 1985; Levinson, 1979; Valliant, 1977). Half of mentored students report having more than one mentor (Clark et al., 2000).

Higgins and Kram's (2001) theoretical work on developmental networks suggests this finding may not be unique.

There is no consistency in who serves as a mentor for graduate students. Several studies define the advisor as the mentor (Green and Bauer, 1995; Schlosser and Gelso, 2001; Tenenbaum et al., 2001; Zhao, 2005). Clark et al (2000) broaden this definition to include any faculty member, while Kahn (2000) placed no limits on who could be considered a mentor. One researcher asked respondents to imagine an ideal mentor and answer the survey questions with that person in mind (Rose, 2003).

There are limited studies on race and gender and they have reported equivocal findings. Students of color, who have mentors of color, report significantly more psychosocial and career support as well as more comfort with the relationship (Ortiz-Walters and Gilson, 2005). However, having shared values, regardless of mentor race, was significantly related to these same outcomes. Women are slightly more likely to have female dissertation chairs than is to be expected by chance (Neumark and Gardecki, 1998) and female doctoral students graduate faster if there are more females on the faculty. However, female students with female dissertation chairs take longer to graduate than female students with male dissertation chairs. In contrast, Tenenbaum et al. (2001) find male and female students are more likely to have male mentors.

Citizenship has not been considered in mentoring studies, even though a sizable percentage of doctoral studies in programs in the U.S. are international students (Redd, 2006).

The few studies that link mentoring with protégé outcomes usually report positive outcomes. Outcomes have been defined as satisfaction with program or mentor; scholarly products such as publications or presentations; research interest; and time to degree.

• Graduate students who report psychosocial support from a mentor are significantly more likely, than those who receive less support, to be satisfied with their program or mentor (Tenenbaum et al., 2001; Zhao et al., 2005).
• Graduate students who have mentors are significantly more likely to have scholarly products, such as published articles or conference presentations, (Cronan-Hillix et al., 1986; Nettles and Millett, 2006; Tenenbaum et al., 2001), or interest in research (Kahn, 2000), than students without mentors.

Tenenbaum et al. (2001) report the sole negative outcome; receiving more psychosocial support, is significantly, negatively correlated with student productivity (number of presentations and publications).

Mentoring research has identified significant, but small, effects. This suggests the presence of interactions or conditional effects (Frazier, Tix, and Barron, 2004). Theoretically, identity commitment may be an important variable to consider.

The literature has identified two types of mentoring but it is not clear how the types influence one another. Education researchers point to the importance of disciplinary characteristics and citizenship when studying graduate students.

The Editors point out that it is vital for readers to remember that findings such as these were true in the specific settings and conditions in which the named studies were done.

Never-the-less, some mentoring initiatives are undertaken with the specific agenda of demonstrating that such findings can be changed, even reversed, when different results are desired and when effective mentoring is provided. In other words, do not assume that findings such as those the author reports are automatic results for every mentoring program. That is not the case.

For example (reference bullet point #3 under "Disciplinary characteristics"), by defining mentoring to include "career support", and by training mentors to provide such support, mentoring programs in humanities and social sciences can ensure that their proteges receive such support.

The author's "Frequency" findings related to method of data collection are somewhat mysterious. However, they have significant importance for interpretation of past research, and for planning by those who would do research in the future.

Research Questions

The overarching research question is does mentoring influence graduate student outcomes, and if so, do certain conditions strengthen or weaken that relationship?

Two hypotheses are tested here. In addition, the data are explored to assess the presence of mentoring networks and specific mentor behaviors that might have the most influence on outcomes.

• Hypothesis 1. Students who report more mentoring, i.e. psychosocial and career support, from their advisor will report higher satisfaction with advisor; more intellectual property events, publications, presentations; and more progress on degree milestones.
• Hypothesis 2. Significant relationships in hypothesis one will be influenced by the interaction of the two mentor functions with each other and with student discipline, citizenship, and identity commitment.

It is significant that the results of this study (see later) were 57% received mentoring from their advisors. This means that those who DID receive mentoring could serve as the research "treatment group" and those who DID NOT receive mentoring could serve as the "control group".

This makes this research is a very significant empirical study, one of the best kinds of research. That the sample is as large as it is makes the study even more important.

Methods

This section describes the sample, survey design, and administration.

SAMPLE -Individuals from two, public research-intensive institutions (U1 and U2), who began their doctoral program between August 2000 and January 2003, participated in an online survey during April 2006. Students had time to develop a relationship with their advisor and to produce scholarly work. The universities are located in the southeast and southwest of the United States and were selected because of their sizable doctoral student populations and for convenience. Twelve U1 departments are participating in a study on doctoral completion and are excluded from the sample at the institution's request. The number of students in the sample is 2,004 (1,073 from U1 and 931 from U2).

SURVEY DESIGN - Participants took 10-20 minutes to complete an online survey, which had four sections.

1. The first section asks about enrollment, completion of program milestones, and factors affecting degree progress, which were adapted from the Survey on Doctoral Education and Career Preparation (Golde and Dore, 2001).
2. Mentoring behaviors received from the advisor and the number of mentors are the focus of the second section.
3. The third section asks about scholarly productivity using questions developed for a similar population (Schneider, 2007).
4. The fourth section includes the 24-item Objective Measure of Ego Identity Status (OMEIS) (Adams, 1998), demographic, and career questions.

ADMINISTRATION - Individuals were invited by email to complete an anonymous, online survey.

Two reminder emails were sent seven days apart.

Participants were given the opportunity to enter an incentive drawing for $100 gift certificates. About 25% of the individuals (n= 505) could not be contacted because their email was returned as undelivered, 258 from U1 and 247 from U2. Thus, 1,499 individuals were contacted.

Measurement of Independent Variables

1. Mentoring - Mentoring is measured by two factors computed from fourteen behaviors. The factors represent the psychosocial and career functions of mentoring relationships. These behaviors are taken from the Advisor Working Alliance Inventory (AWAI) (Schlosser and Gelso, 2001). The items in the AWAI scale (as revised by this author) are presented in Table 1. Responses are on a 5-response Likert scale from 1 (strongly disagree) to 5 (strongly agree). Some items are reversed coded. Respondents also rated the extent to which they consider their advisor a mentor, how much and what type of mentoring they receive from others, and how many people they consider a mentor.

2. Identity Commitment - The short version of the Objective Measure of Ego Identity (OMEIS) was selected to measure identity status because of its established validity and reliability (Adams, 1998). There are 24 questions; six items for each of the four identity statuses, which cover three domains: occupation, politics, and religion. Subjects respond using a scale from 1 (strongly disagree) to 6 (strongly agree). The OMEIS has been used on samples of college students and young adults (Adams, 1998), although no studies were found with a doctoral student population.

3. Citizenship - Citizenship is assigned from the respondent's choice from these options: U.S. Citizen, Permanent Resident, or Non U.S. Citizen. Citizenship is dummy coded with U.S. Citizen as the referent.

4. Discipline - The respondents report their field of study, which were collapsed into six disciplines: education, engineering, humanities, mathematics, science, and social science. Discipline is dummy coded with engineering as the referent.

Clearly, because the AWAI factors are specific to graduate education advisement, some of these factors do not generalize well to other grade levels within education, EVEN when the focus is advisement, such as during middle and high school years.

Further, some of these factors may not generalize well to other settings, even when advisement is a key mentoring role, unless the graduate level language is revised to be more general, or to be more specific to another setting. This would be especially the case with the italicized items.

Identity commitment was assessed since the theoretical framework for the study showed that the issue of identity development may relate to mentoring since the period when identity is developed is also when individuals will most benefit from having a mentor.

Dependent Variables

Satisfaction with advisor - Satisfaction with advisor is computed from responses to one, 5-level Likert scale question; a 5 indicates more satisfaction.

Student productivity - Intellectual Property Events is a categorical yes/no variable. An intellectual property event includes any of the following: invention disclosures, patent applications, copyrights, patents granted/derived, and licensing agreements.

Degree progress - Progress toward degree is calculated from respondents' answers to eight questions about common degree milestones. There is a short time to the graduation milestone after a dissertation defense, thus these two milestones are collapsed. Respondents indicate if the milestone is a program requirement and if they have completed it. A modified Guttman scale is used to calculate this variable. If certain milestones are completed it indicates the candidate has completed all the previous steps, e.g. advancement to candidacy. These scores are assigned as follows:

• 7 - graduated or orally defended their dissertation
• 6 - orally defended their proposal
• 5 - written dissertation proposal
• 4 - advancement to candidacy.
• The remaining respondents receive a 1, 2, or 3 depending on the sum of these milestones: classes, master's degree, and comprehensive exams.

Covariates - There were two covariates in the analysis, number of semesters since beginning their doctoral program and number of semesters of undergraduate research

You may want to know . .

"Independent variables" are those factors that are present at the start of a process (like before mentoring happens).

"Dependent variables" are those factors which are dependent on, or created by the independent variables.

The point of identifying and evaluating these is to be able to determine which factors are significant in causing the desired (positive) or undesired (negative) results.

"Covariance" is the effect of two variables which interact and complicate the process of determining what causes what, and what might have happened anyway, regardless of whether mentoring happened or not.

Results

Sample Characteristics
Respondents - The sample is 53% female, 66% European American, 73% U.S. citizens, and 82% are enrolled at the time of the survey.

• About half of the respondents study engineering (14%) or science (38%).
• Two-thirds of the respondents are married or partnered.
• The majority of engineering students are international students.
• Significantly more women than men from U2 responded.
• There are no significant differences in the number of white students and students of color from the Universities.
• The response rate is 45%. Missing and excluded data reduces the sample to a response rate of 32%.

Mentors - Two-thirds (63%) of the advisors are full professors.

Hypothesis Testing
Hypothesis One - Computing the appropriate regression, to examine if the two mentoring functions are significantly related to the five doctoral student outcomes, tests the first hypothesis. The psychosocial mentoring function is positively, significantly related to satisfaction with advisor and the career mentoring function is positively, significantly related to all five outcomes. Thus, there is partial support for the hypothesis one.

Hypothesis Two -
Satisfaction with Advisor.

• The two mentoring functions significantly, positively interact to influence satisfaction with advisor. The interaction between career and psychosocial support is significant and positive.
• There is a significant, negative main effect of number of semesters of undergraduate research on satisfaction with advisor.
  The interaction of the mentoring variables provide partial support for hypothesis two.

Intellectual Property Events.

• The two mentoring variables are not significantly related to having an intellectual property event.
• Discipline and number of semesters of undergraduate research are significant as main effects.
  • Individuals who report more semesters of undergraduate research have odds of an intellectual property event 18% higher than those with one less semester of undergraduate research.
  • Students in majors other than engineering have significantly less odds of reporting an intellectual property event (28% less for science/math; 16% less for social science/humanities; and 9% less for education).

There is no support for hypothesis two for intellectual property events.

Publications. Discipline significantly interacts with the two mentoring variables to influence the number of publications.

• Receiving one-unit more of psychosocial support is significantly associated with lower odds of having a publication for students in the Humanities (30% less) and Social Sciences (17% less).
• Career support has the converse effect on publications. Receiving one-unit more of career support is significantly associated with greater odds of having an additional publication for students in social sciences (7 times greater), humanities (2 times greater), and education (3 times greater). Number of publications for engineering students is not affected by receiving more career support.
• The significant main effects are number of semesters since beginning the program, diffusion identity status and moratorium identity status.
• There is a significant main effect of mathematics, with students in mathematics being 11 times less likely to publish a paper than a student in engineering.

There is partial support for hypothesis two for because discipline significantly interacts with career support to influence publications.

Presentations.

• Career support is significantly related to number of presentations, but there are no significant, conditional effects; thus hypothesis two is not supported for presentations.
• Individuals are 40% more likely to have a presentation for each unit increase in career support.
• For a one-semester increase in having participated in undergraduate research, the odds of presenting at a conference increase 15%.
• Doctoral students in mathematics have lower odds (60%) of presenting than students in engineering.

Progress Toward Degree.
The two mentoring variables are not significantly related to progress on degree milestones but the mathematical model is significant because of the main effect of discipline. There is no support for hypothesis two for progress on degree milestones.

• Students majoring in science make significantly more progress toward degree than students in Engineering.
• Students with more undergraduate research experience report more progress on degree milestones than students with less undergraduate research experience.

Although the final response rate of 32% seems quite low, in fact it is quite typical of such studies. Imagine how low it would be if no incentive 'drawing" had been provided to motivate participation.

The hypotheses were originally stated earlier in the article. For clarity regarding conclusions, the hypotheses are presented here again.

There is only "partial support" for Hypotheses one (at this point in the discussion) since it includes both  "satisfaction with advisor" AND the results of more intellectual property events, publications, etc. and the latter are not addressed yet in the discussion.

When all data are considered the following interesting results are found:

(Hypothesis 1 part A). was proven. Students who reported more mentoring,( i.e. psychosocial and career support) from their advisor AL:SO reported higher satisfaction with their advisor.

However, (Hypothesis 1 part B) was NOT proven. Students who reported higher satisfaction with their advisor did NOT also report more intellectual property events, publications, presentations; or more progress on degree milestones.

What DID effect the number of intellectual property events and progress on degree completion was the area of discipline (field of study).

• Engineering had the highest effect
• Science & Math had 28% less effect
• Social Science/Humanities had 16% less
• Education had 9% less.

(Also Hypothesis 2). The two mentoring functions (psychosocial and career support nor the ) were NOT significantly effected by citizenship nor identity commitment. Neither were the number of intellectual property events or progress on degree effected by citizenship or identity commitment.

Exploratory Analysis

Mentoring is the second most frequently endorsed item of seven factors graduate students identify as contributing to their success. Over half of the doctoral students agreed or strongly agreed with a statement that their advisor is a mentor, another 25% are neutral about their advisor being a mentor, and 18% did not agree that their advisor is a mentor. There are no significant differences by gender, discipline, race, marriage/partnership status, or citizenship for students who considered their advisor a mentor and those who did not.

Most individuals report multiple mentors, including their advisor. Two-thirds of the respondents have a faculty mentor, other than their advisor. Two-thirds of the sample have a mentor outside of the faculty. Only 14% of the respondents report no other mentors.

Analysis of the individual AWAI items reveals collaboration to be significant. Introduction to professional activities, offering improvement ideas, and feeling respected are significantly related to three of the dependent variables. However, four behaviors (including feeling respected) are negatively related to outcomes. Therefore, not all mentor-like behaviors are associated with positive outcomes.

While it is important that, in this study, 2/3rds sought mentoring from faculty other than their advisors AND 2/3rds sought mentoring from outside the faculty, it must be understood that, in this study, only 57% felt advisors served as mentors.

The biggest unstated finding here is that, when mentoring is not received from those who officially should be providing mentoring, persons will seek multiple mentors elsewhere. The drive in adults to succeed is that strong. Note however, that this is unlikely to be the case with youth, especially those who have never benefited from adequate support and guidance.

The implications of this finding are offered in the section with that title below.

Caution is in order here when considering the use of these findings in YOUR mentoring work, especially those where the significance is negative. While "not all mentor-like behaviors are associated with positive outcomes" was true in THIS study, you may not find this to be true if you work with proteges who are not graduate doctoral students.

There are several reasons why this may only be the case in this setting and in this study. However, this article does not explain these other factors well enough for us to be guided by this finding. This is especially true when only 57% of the respondents considered their advisor to have functioned like a "Mentor".

Discussion

The results advance knowledge about mentoring in three areas by:

1. Providing evidence that mentoring is important but conditional on discipline and the type of mentoring
2. Providing evidence of mentoring networks; and
3. Examining the behaviors that constitute mentoring relationships.

In addition, the covariate undergraduate research, arguably a type of mentoring relationship, was unexpectedly related to doctoral student outcomes.

Conditional relationships - Psychosocial and career support were positively, significantly related to doctoral student outcomes in a main effects model. There was a large effect size for one dependent variable, satisfaction with advisor. However, the small, significant effect sizes found in previous research were found for presentations, publications, intellectual property events, and progress on degree milestones. Most studies stop at this point.

This study went one step further by introducing conditional effects, which presented a different view of how the type of mentoring influences student outcomes. The exception was satisfaction with advisor; which already had a large effect size. Psychosocial and career support significantly, positively interacted to influence satisfaction with advisor. The interaction of these variables has not been investigated but these results suggest they are dependent on one another, at least for ratings of advisor satisfaction. Having a personal relationship with an advisor might make a student more likely to take their career advice.

Variations Across Disciplines - This study replicated Tenenbaum's et al (2001) findings of negative outcomes. Why would students in humanities or social sciences who receive more psychosocial support have less odds of publishing than students in engineering ?

1. Biglan (1973) suggests that work proceeds differently in “pure” versus “soft” disciplines. In the pure disciplines there is more structure, more collaboration and research is done a team, laboratory environment. Individual work characterizes humanities. Thus, a struggling student may have more support built in the pure disciplines and may be left to flounder in the softer ones. If there is less support in soft disciplines then the role of the mentor is more important and a student may need more psychosocial support. Similar findings have been reported in the youth mentoring literature (Rhodes, 2002).

The disciplinary characteristics may also explain why receipt of more career support is related to higher odds of publishing for students in the humanities, social sciences, and education. Career support may be built into the pure disciplines through laboratory teams. Thus, the influence of advisor career support may not be as important. However, career support from an advisor may make a big difference for the students working alone in the humanities.

2. Most doctoral students are embedded in a network of mentors, providing evidence for Higgins and Kram's (2001) developmental network theory. Most students report multiple mentors and receive mentor-like support from a number of people, including fellow graduate students. Are certain individuals more likely to attract mentors? Perhaps there is a personality trait or mentoring personality causes a person to report (or attract) more mentors. The small effect sizes in previous studies might be partially explained by a failure to include mentoring received from more than one person.

3. This study explored which mentor behaviors positively and negatively influence protégés' outcomes. For example, collaboration and engagement in professional activities are positively, significantly related to several doctoral outcomes. Other behaviors, such as recognizing areas of improvement, welcoming input, and establishing a timetable have a negative, significant relationship with the doctoral outcomes. Feeling respected is positively related to satisfaction with advisor and intellectual property events but is negatively associated with presentations.

Instrumental and Sample Considerations - The findings need to be considered with caution.

1. The AWAI instrument might not encompass all of the behaviors that characterize mentoring relationships in graduate school. For example, introducing a student to a new literature was not one of the AWAI items but is an important professional, mentoring behavior.

2. The OMEIS was used to measure identity status but the data suggest it might not be the best measure.

3. Only currently enrolled or graduated students were included in this study. Thus, these findings might not apply to students who are on leave or who have left their program. A related concern is the lower response rate from international students, which means the findings might not apply to international students.

4. Finally, it might take years for the effects of mentoring to be realized, which might not show up in the proximal measures used here.

THESE RESULTS STATEMENTS ARE WHERE READERS ARE URGED TO FOCUS.

1A. Mentoring is important to graduate students (and others interested in learning and improving, we add). This is so much so that:

persons seek multiple persons to mentor them, (results #2, "networks")

AND . . .

seek mentoring elsewhere when those who are supposed to provide it do not.

1B. The typical ways that people work and the extent of collaboration (independently, as teams, etc.) vary with discipline (settings). "Informal mentoring" and support are more available in collaborative, team-oriented environments with the result that persons who are learning and seeking to improve find more support for those agendas.

NOTE - This should not suggest that these kinds of settings do not need more formal mentoring. Other research shows that where ever there is an expectation for learning and improvement, formal mentoring should be provided to support achieving that goal.

Clearly, those settings which are characterized by independent work and less collaboration and team work, tend to provide inadequate informal mentoring opportunities to support learning and improvement.

In these settings, formal mentoring is critical for staff and for organizational development and attainment of the organization's strategic objectives (retention, increased productivity, better results, etc.)

Implications For Further Research

This study has four implications for future research.

First, more work is needed to understand how citizenship and identity may influence mentoring and student outcomes. The measures of citizenship and identity commitment were problematic. Career commitment might be more appropriate to discern the relationship between mentoring and this theoretically important concept.

Second, future studies need to examine the influence of multiple mentors because so many students in this sample reported them; failure to consider multiple mentors may be why one reason for the small effect sizes found in the literature.

Third, the findings highlight a need to develop discipline specific outcomes. For example, number of publications is not a good measure for students in math since so few math students publish.

Fourth, few of these advanced or graduated students reported having no mentor but more work is needed to understand the relationship between mentoring and doctoral attrition. An attrition rate as high as 50% seems to be an inefficient use of scarce resources.

The Editors would encourage further research, especially for items 2 and 4 described by the author. In fact, such research studies would be excellent candidates for future awards like the one this dissertation received. (Se IMA home page)

Regarding the third item the author suggests, the Editors find that such a "need", as stated, is only felt when viewed in general. Persons who work in specific settings are well aware of the desired outcomes of that setting and their mentoring and focus (or lack of it) on intellectual property events and other outcomes is automatically adjusted to fit setting norms and expectations.

IT WOULD BE INTERESTING to study further and describe exactly what those norms and expectations are for diverse settings and how they effect the kind of mentoring.

It would also be interesting to attempt to relate those diverse norms and expectations to the extent of collaboration and teamwork in each of those settings.

Practical Considerations
The results provide two practical considerations for doctoral education.

1. First, the friend aspect of mentoring is usually emphasized (National Academy of Sciences, 1997) but doctoral students might benefit if career support is also emphasized. Career support had a positive effect on doctoral student productivity and mirrors findings in case studies on the development of exceptional talent, which found career guidance to be important (Sand, 2000; Zuckerman, 1977). Furthermore, the specific career behaviors of collaboration and building of professional networks appeared to be responsible for the positive relationship with doctoral student outcomes. Thus, these types of activities might be highlighted more in graduate programs.

2. Second, the findings suggest support of undergraduate research might be warranted because it was robustly related to several important doctoral student outcomes. The relationship between mentoring and outcomes disappears when talent is considered (Green and Bauer, 1995). Thus, this variable was included as a measure of talent. It was not found to be important as a talent variable but it was left in the analysis as a covariate because it was significantly related to the outcomes.

Number of semesters of undergraduate research is a crude measure of an activity, arguably a mentoring activity, which occurred years prior to the data collection. However, it was significantly, positively related to having an intellectual property event, publications, and progress on degree milestones. It was negatively, significantly related to satisfaction with advisor. This last, surprising finding might be explained because these students might have had higher expectations of graduate advisors, given their prior research experience. Undergraduate research might be an important pipeline into doctoral education and was clearly related to doctoral outcomes.

This study found that only 57% felt advisors served as mentors and that 2/3rds of these same respondents sought mentoring from faculty other than their advisors AND also from outside the faculty  The implications of this finding are that mentoring programs must ensure that:

1. Persons whose role is advisement, teaching, leading, and other such roles, deserve clear expectations that their work IS mentoring and a clear definition of what effective mentoring looks like.

2. Official, formal "mentors" must be adequately trained, supported and held accountable for functioning as effective mentors.