حالت تحقیق مشاوره ای: مرور کیفی روش های پژوهش حاضر و پیامدهای تحقیقات آینده
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|8394||2008||15 صفحه PDF||سفارش دهید||6762 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Journal of Vocational Behavior, Volume 73, Issue 3, December 2008, Pages 343–357
Research regarding mentoring relationships has flourished during the past 20 years. This article reviews the methodology and content of 200 published mentoring articles. Some of the major concerns raised in this review include over reliance on cross-sectional designs and self-report data, a failure to differentiate between different forms of mentoring (e.g., formal versus informal), and a lack of experimental research. Implications and suggestions for future research are offered.
In the last two decades there has been a tremendous surge in interest and research on the topic of mentoring. Dozens of studies have documented the numerous advantages of mentoring for the organization, the mentor and the protégé (e.g., Allen et al., 2004, Allen and O’Brien, 2006, Kram, 1985 and Scandura, 1992). Although a substantial body of research has accumulated, as an area of scientific inquiry, workplace mentoring is a relatively new focus of study. As noted by Reichers and Schneider (1990) in their stage model describing the pattern by which new concepts are advanced, critical review and summarization of existing literature plays a key role in the evolution of concepts. Only recently have such qualitative (Noe et al., 2002, Ragins, 1999a and Wanberg et al., 2003) and quantitative (Allen et al., 2004) reviews of the mentoring literature appeared. In addition to summarizing content knowledge concerning what we know about mentoring such as protégé career outcomes (Allen et al., 2004) or the role of gender in mentoring relationships (Ragins, 1999a), existing reviews have been critical of various methodological issues associated with mentoring research. For instance, a methodological review of diversity and mentoring relationships underscored several research and design considerations (Ragins, 1999b). Some of the major concerns raised included over-reliance on cross-sectional designs and self-report data, a failure to differentiate between different forms of mentoring (e.g., formal versus informal), lack of dyadic data, and the use of psychometrically questionable measures. Given the frequency with which methodological concerns are highlighted by mentoring scholars, a systematic methodological review of the mentoring literature seems warranted. A body of research is often judged in terms of the methodology utilized. Thus, it is important to review the methods that have been employed in order for an area of research to progress rather than stagnate (Casper et al., 2007 and Scandura and Williams, 2000). A focus on methodology within the mentoring literature is critical at this point because methodological choices influence both the breadth and depth of what we know about these important organizational relationships. For example, mentoring relationships progress through distinct stages over time marked by unique issues and challenges (Kram, 1985). As such, longitudinal research strategies are essential to fully appreciate the dynamic nature of mentoring. Mentoring relationships are also inherently dyadic and the failure to consider both partners’ perspective greatly limits our understanding of mentoring (Eby, Rhodes, & Allen, 2007). Unambiguously defining and operationalizing mentoring is also vital given the differences that have been found between formal and informal mentorships (cf. Wanberg et al., 2003). Finally, as with any area of research, the failure to use content and construct valid measures leads to serious threats validity and interpretive problems (Cook & Campbell, 1979). To help facilitate the progressive theoretical development of the construct of mentoring, a comprehensive review of the organizational mentoring literature was conducted. Although methodology is the primary focus of our review we also examine some substantive issues, such as how frequently various content domains have been examined and the types of mentoring relationships studied by researchers. The present review proceeded with two specific objectives in mind. First, we wanted to summarize the current state of mentoring research and pinpoint the most pressing methodological concerns given the nature of the mentoring construct. This will allow us to make methodological recommendations that will extend our understanding of organizational mentoring. Second, the methodological features of existing mentoring research are juxtaposed against the findings from several reviews of organizational and management research in general (Aldag and Stearns, 1988, Scandura and Williams, 2000, Schriescheim et al., 1993 and Stone-Romero et al., 1995) and the work-family literature in particular (Casper et al., 2007). This provides a frame-of-reference for evaluating the state of mentoring research. Accomplishing these two objectives will help advance mentoring scholarship in important new directions.
نتیجه گیری انگلیسی
Choices with regard to research settings, research designs, and analyses have critical implications for the accumulation of knowledge regarding an area of research (Scandura & Williams, 2000), as do the substantive questions that are the focus of study. Our review supports many of the criticisms leveled against mentoring research. It also identifies substantive areas of mentoring scholarship that are well-studied and topical areas that deserve greater attention. In the sections that follow we discuss the specific findings of our review and offer suggestions to move mentoring scholarship forward. 3.1. Research designs Research design results are summarized in Table 2. Of the 207 studies coded, 178 were empirical, with the vast majority being exclusively quantitative (89.9%). As such, an important question is whether mentoring scholarship could benefit from more qualitative research. Given the changing nature of organizational careers and the current state of mentoring research, we believe that the answer is a resounding “yes.” The linear, stable and hierarchical careers that were commonplace in the 1980s when Kram’s initial qualitative research on mentoring was conducted have been replaced with careers that take more varied forms and are marked by greater instability and inter-organizational mobility (Arthur & Rousseau, 1996). The boundaryless and protean careers of today emphasize new career skills and psychological ways of viewing one’s career (Arthur and Rousseau, 1996 and Briscoe and Hall, 2006). This suggests that the types of assistance that mentors provide to protégés, and the venues within which this assistance is provided, may differ today from 20 years ago (Eby, 1997). With an emphasis on in-depth description and careful appreciation of the context (Locke and Golden-Biddle, 2002 and Van Maanen, 1979), qualitative research on mentoring has substantial utility in helping us understand the role of mentoring and contemporary careers. In terms of research setting, 96.0% of the studies examined mentoring in a field setting and in terms of design, 5.1% used an experimental design. The high base rate of field research is not surprising because mentoring is an interpersonal relationship that is arguably best understood from the perspective of adults working in organizational settings. Further, conducting experimental research in organizational settings is difficult and time consuming. However, by not using controlled settings and experimental designs it is impossible to establish cause-and-effect relationships among key constructs (Cook & Campbell, 1979), which is an essential step in theory testing (Bacharach, 1989). It is also important to underscore that mentoring represents a basic social process that could seemingly benefit from careful study in a highly controlled environment. In fact, research on other types of close relationships such as friendship and romance routinely use laboratory settings to uncover basic relational processes such as interpersonal attraction (e.g., Montoya & Horton, 2004) and relational commitment (e.g., Gonzaga, Keltner, Londahl, & Smith, 2001). Laboratory research on mentoring would allow us to develop and test theories related to topics such as the attraction process between mentors and protégés, the exchange of tangible and intangible resources within the mentoring dyad, and the development of relational processes such as trust and disclosure in mentorships. For example, several mentoring studies have successfully used laboratory experiments to examine mentor-protégé attraction (Allen, 2004 and Olian et al., 1988). One example for future research would be to manipulate factors such as the race and gender of mentors and provide controlled stimuli regarding mentoring behaviors exchanged during the course of the mentorship. This would permit an objective evaluation of whether or not the same mentoring behaviors provided by a female/minority are viewed similarly to those provided by a male/nonminority. The near absence of laboratory research on mentoring may be a reflection of the values of journal reviewers and editors, as well as researchers. Studies conducted under highly controlled laboratory conditions often generate concerns regarding whether or not “real mentoring” is being studied. In response to this criticism we are reminded of Mook’s (1983) discussion of the merits of “external invalidity” (p. 379), in which he eloquently discusses the flawed logic underlying the frequent dismissal of laboratory research on the grounds that it is artificial and lacks generalizability. Experimental research is not intended to be generalizable across people, time, and settings. Rather, it is used to test the validity of specific theoretical propositions that may (or may not) have utility in understanding some comparable phenomena in the field. Because testing core theoretical propositions is the building block of theory (Bacharach, 1989), we strongly recommend greater consideration of laboratory-based experimental research in journals that routinely publish organizational research on mentoring. In terms of time horizon, of the 176 studies in which data were collected (i.e., excluding the two meta-analyses) 90.9% used a cross-sectional design. The limited examination of mentoring relationships using longitudinal designs places a theoretical constraint on our understanding of mentoring. Mentoring research has grown and become a major topic of interest primarily based on the belief that it leads to beneficial outcomes such as career growth and favorable job attitudes. Yet for the most part, researchers have not made a convincing case that this reflects a causal relationship. In fact, some have argued that protégés are singled out for mentoring because they demonstrate positive work attitudes or high performance potential (Ragins & Cotton, 1993). In one notable longitudinal study conducted with student–faculty mentors in a doctoral program, Green and Bauer (1995) found that greater use of mentoring in the student’s first year did not contribute to student performance and commitment at the end of year two after controlling for student attitudes, experience, and ability as measured at program entry. These findings underscore the possibility that other factors such as protégé characteristics may underlie the link between mentoring and positive outcomes. However, it is uncertain how these results generalize to workplace mentorships or the extent that a one-year time lag is long enough for mentoring to have a positive effect. Another example is that cross-sectional studies have examined willingness or intent to mentor others and related it to factors such as previous mentoring experience and altruism (e.g., Allen, 2003, Allen et al., 1997, Arthur and Rousseau, 1996, Aryee et al., 1996 and Ragins and Cotton, 1993), but no longitudinal studies have been conducted to reveal the temporal precedence of any variables and actual mentoring behavior by mentors. As a final example, we have a limited understanding of relationship phases in mentoring. Kram’s (1985) seminal work on mentoring discusses the time-bound nature of mentoring and the unique relational dynamics associated with the phases of initiation, cultivation, separation, and redefinition. To date there has been no independent examination of mentoring phases using longitudinal designs. Moreover, no longitudinal research exists on the phases associated with formal mentorships even though the overall length of time associated with Kram’s mentoring phases is considerably longer than the duration of most formal mentorships. In summary, longitudinal research is sorely needed to examine these and other taken-for-granted assumptions about organizational mentoring. This should include long-term studies that track mentor and protégé characteristics, as well as both formal and informal relationships, at multiple time points from initiation through termination/redefinition. 3.2. Sources of data and focus of inquiry Results concerning sources of data and focus of inquiry are reported in Table 3. Almost all of the studies used survey-based methods for data collection (94.4%). A small percentage of studies (6.3%) used multiple data collection methods. The heavy reliance on surveys raises validity concerns. Triangulation in the form of multiple methods of data collection increases construct validity because it provides a more holistic assessment of the construct under study and reduces mono-method bias (Campbell and Fiske, 1959 and Jick, 1979). Triangulation is also important because all research methods have inherent weaknesses and obtaining convergent results across different methods increases the confidence that one can place in research findings (Jick, 1979 and McGrath, 1982). Paper-and-pencil surveys in particular have limitations such as increased susceptibility to response sets, greater potential for misunderstanding questions, and the inability to follow-up on respondent answers. Thus, mentoring research may benefit from more varied means of data collection. As mentioned previously, this may include experimentally designed research. Studies that include a combination of field-based survey research and a parallel experimental lab study would be especially useful for addressing both internal and external validity (see for example, Allen, 2004). We also found limited triangulation in terms of data sources. Only 18.2% of the studies included in the present review collected data from multiple sources. Moreover, the protégé was the focus of inquiry in 80.2% of the studies. The mentor was the focus in 30.9% and 27.5% of the studies focused on dyadic processes (e.g., cross-gender dynamics, mentor-protégé similarity). Only 7.2% of the studies reviewed focused on the organization. The limited use of multi-source data and heavy focus on the protégé raise several concerns. A mentoring relationship is an inherently dyadic and complex process, with the mentor and the protégé each enacting different roles and responsibilities in the relationship (Allen, 2007). Mentors and protégés also report different benefits (Eby, Durley, Evans, & Ragins, 2006; Ragins & McFarlin, 1990) as well as costs (Eby, 2007) in a mentoring relationship. This suggests that data from both perspectives is necessary to fully understand a mentoring relationship. For example, it would be informative to conduct research in which mentor personality data was self-reported and then correlated with protégé reported outcomes. This would yield some insight into mentor characteristics associated with beneficial protégé outcomes. 3.3. Research content domains Table 4 contains the results regarding the general content areas that have been the focus of mentoring research. The most common theme is research that investigates predictors of mentoring (31.9%), closely followed by outcomes associated with mentoring (30.0%) and both predictors and outcomes (21.7%). Therefore, we seem to have a balanced understanding of both the antecedents and consequences of mentoring, at least from the protégé’s perspective (see Table 3). Less attention has been given to research reviews (7.7%), theory development (8.2%), and measurement development (3.9%). The field of mentoring could benefit from more concerted attempts to advance theory. This seems especially important in light of the recognition that career patterns and organizational experiences have changed dramatically in the last few decades (Briscoe & Hall, 2006). Moreover, it is important for any area of study to review past findings in an effort to appropriately guide future directions. 3.4. Type of mentoring Results regarding the type of mentorships studied are provided in Table 5. We found that 39.8% of the articles reviewed did not specify the type of mentorship studied. Another 28.4% of studies included samples comprised of multiple types of mentorships. Failure to specify the type of mentoring examined poses potential interpretive problems because differences have been repeatedly found in the outcomes associated with formal and informal mentoring relationships (e.g., Chao, Walz, & Gardner, 1992; Ragins & Cotton, 1999). Likewise, studies that include both formal and informal mentoring may be “mixing apples and oranges” to some extent if fundamental aspects of the relationship differ across these two types of relationships. In future research it may be useful to clearly specify the type of mentoring under study. 3.5. Measurement of mentoring Table 6 includes the results regarding the measurement of mentoring. Of the empirical studies reviewed 44.9% measured psychosocial mentoring. Of those that measured psychosocial mentoring, the most commonly used measure was that developed by Noe (1988) (29.1%), followed by Scandura (Scandura, 1992 and Scandura and Ragins, 1993) (25.3%). A considerable number of researchers used study-developed measures or cited measures only recorded once (30.4%). A total of 50.6% of empirical studies reviewed included a measure of career mentoring. Again most of these studies used the Noe (24.7%) or the Scandura (24.7%) measure.The practice of using study-developed measures of mentoring raises concerns. Construct validity is questionable when measures are used with limited psychometric support. Our review also reveals a paucity of research focusing on measurement within the mentoring literature. Indeed, as shown in Table 4 only eight of the 207 studies reviewed focused on measurement development. Perhaps more importantly, the mainstay measures of mentoring support are based on Kram’s (1985) taxonomy of mentoring functions. As mentioned previously, it seems important to re-examine the mentoring functions identified by Kram over 20 years ago to verify that they are as applicable today as they were in the early 1980s. It should also be noted that the most commonly used measures of mentoring (i.e., Noe, 1988 and Scandura, 1992) have not been tested for equivalence with mentor samples, despite commonly being used for mentor reports of mentoring provided (e.g., Allen & Eby, 2004). This is important in that mentor and protégé reports of career and psychosocial support tend to be only modestly correlated (e.g., Allen, Eby, & Lentz, 2006). We are also aware of no research on cross-cultural measurement equivalence, even though Kram’s conceptualization of mentoring has been applied to other cultural contexts (e.g., Bozionelos, 2006). For mentoring research to advance, it seems critical that more effort be directed toward examining measurement issues in mentoring research. For example, in a recent study, Pellegrini and Scandura (2005) found that the Mentoring Functions Questionnaire (MFQ-9) demonstrated excellent psychometric properties among unsatisfied protégés, but not among satisfied protégés. 3.6. Country of origin The results regarding sample country of origin are presented in Table 7. As shown, the vast majority of research has been based on samples from the United States. Globalization is affecting all aspects of work within organizations. Mentoring relationships are likely no exception. However, little is known regarding the generalizability of research findings regarding workplace mentoring across nonwestern cultures. Our review shows that only a handful of studies have been conduced with non-North American samples. Are there different perceptions of mentoring relationships in other cultures? Do mentoring relationships take on the same developmental importance in nonwestern cultures as has been found in U.S. research? How well do the widely accepted functions of mentoring identified by Kram (1985) apply to other cultures? Even within the U.S., little is generally known regarding cross-ethnic mentoring relationships. Cultural values such as collectivism/individualism have significant effects on social relationships (Chen, Brockner, & Chen, 2002). For example, research suggests that Asians may be less likely to seek help from relationships outside of the family compared to Americans (Zhang, Snowden, & Sue, 1998). It seems likely that such differences may dramatically influence norms and expectations regarding mentorships in the workplace.3.7. Mentoring research in comparison to other research areas Another way to view the results of the present review is to compare our findings to other methodological reviews of the literature. Several reviews of organizational and management research exist. Some are selective in terms of the content domain covered (e.g., Casper et al., 2007 focused exclusively on the work-family literature) and others in terms of the journals reviewed (e.g., Scandura & Williams, 2000 reviewed management research in general in 3 specific journals) and others in terms of type of research (Eby, McCleese, & Butts, in press; focus on qualitative research). It should also be noted that existing methodological reviews examined some, but not necessarily all, of the methodological features discussed in the present review. Nonetheless, in the following paragraphs we examine the methodological features of mentoring research against other areas of scholarship. Mentoring research was similar to other areas of research on several counts. Juxtaposed against research in applied psychology, management, and social psychology (Eby et al., in press) as well as work-family (Casper et al., 2007), mentoring research is characterized by slightly greater use of qualitative (1% and 1% versus 4.5%, respectively) and combined (qualitative and quantitative) (2% and 1% versus 5.6%, respectively) research approaches. Qualitative research has many merits, including greater ecological validity and rich, descriptive accounts of real world phenomena (Lee et al., 1999 and Van Maanen, 1979). In other respects, mentoring research is based on a more limited methodological repertoire. We found that mentoring research (9.1%) was less likely to utilize longitudinal designs than was organizational and management research (29% in Aldag & Stearns, 1988; 14.4–22.6% in Scandura & Williams, 2000). However, this may be changing as nine studies that used a longitudinal design were published in 2005–2006. Mentoring research was also less likely to use experimental/quasi-experimental designs (5.1%) compared to all articles published in the Journal of Applied Psychology between 1975 and 1993 (32–49% in Stone-Romero et al., 1995) as well as organizational and management research (32% in Aldag & Stearns, 1988; 7–15% in Scandura & Williams, 2000). The mentoring studies that we reviewed were also less likely to use multiple methods (6.3%) or multiple sources (18.2%) compared to either work-family (23% and 24%, respectively, Casper et al., 2007) or organizational and management (19% for multiple methods, Aldag & Stearns, 1988 and between 11.1–26.3% for multiple people, Scandura & Williams, 2000) research. Again, this appears to be changing though as 11 studies incorporating multiple data sources were published in 2005–2006. In conclusion, our findings suggest that we can characterize mentoring research as primarily adopting quantitative, correlational, cross-sectional research designs in field settings where data are collected from a single source (typically the protégé) using a single method of data collection. This seemingly paints a bleak picture of the current methodological state of mentoring research. However, we emphasize that this may simply reflect an area of research that is still at a relatively early stage of development. Initial studies within a research area may justifiably focus on establishing relationships among variables through the use of expedient and low cost research methodologies before proceeding to more costly, time-consuming, and complex studies. However, in order to ensure the advancement of this field of study, the time has clearly come to make such investments. As demonstrated by the recent increase in longitudinal and multisource studies, more complex research appears to be beginning to emerge. We hope the results of the current review help to continue to inspire continuing efforts along these lines.