You are on page 1of 17

Research ill Higher Educatlon, Vol. 3f{, No.

6, J997

FACULTY/STUDENT MENTOR PHOGRAM:

Effects on Academic Performance and Retention

Toni A. Campbell and David E. Campbell

A university laculty/student mentor program was evaluated for its effects on academic performance and retention. A matched pairs design was used in which 339 unde~gradl1ates assigned to mentors were paired with nonrnentored studenls based on gender, ethnicily, GPA, and enterlngl enrollment status. The results showed a higher GPA lor rnentored sludents (2,45 VS. 2.29), more units completed per semester (9.33 vs, 8.49), and lower dropout rate (14.5% vs. 26.3%). Amount of mentorprotege contact was positively correlated with GPA. Academic achievement and raten lion were unrelated to gender andelhnicity 01 the mentor, the protege, or Ihe gender and ethnic match between the two.

During the past decade, American universitie have made a concerted effort to create carnpu environments that reflect the diversity of the general population, especially wirh regard 10 gender andethniciry, As the efforts to attract and retain students of underrepresented minorities have inten ified, colleges and universities have implemented a variety of support programs. One type of program that has become particularly popular is an adaptation of the apprentice model of graduate education; this is the faculty mentor program in which faculty members are encouraged to serve as mentors to' undergraduate students. As used here, memoring refer. 1.0 a situation in which a more-experienced member of an organ ization maintains a relation hip with a las -cxperienced, of len new member to the organization and provide! information, apport, and guidance so as to enhance the les -experienced member'S chance of succe s in the organization and beyond. We will refer Ie the more-experienced member of such a. relationship as .11 mentor and lhe Less-experienced indi vidual as a protege.

Current research on mentoring deals with two types of setting: business organization, and educational environments, The literature from business source often treats mentoring as an organizational strategy for training and

Toni A. Campbell, Son lose State University; Davie! E. Cnmpbcll, Humboldt State University.

Address correpondence 'to: lbni A. Campbell, ChU't1 Developmern Program. Sat! lose State Universiry. S"II Jose, CA 95192-0074.

727

728

CAMPBELL AND CAMPBELL

develeping personnel within a firm, whereas other literature take a vocational perspective and suggests guidance for individual career development (e.g., Dreher and Ash, 1990; Fagenson, 1989; Kram, 1985). The research from educational sources ranges broadly from peer rnentoring in econdary education to studies of graduate educarion 1n doctoral programs (e.g., Queen, 1994; Wilde and Schau, 1991). The mentoring proces. itself takes on a variety of forms. In some ca. es, formal programs are adminl rered in which students are assigned to mentors. In other. tudents and mentors develop relationships "naturally" with no formal support from the administratlon, Furthermore, the research literature is aceumalaring in the absence of a generally accepted definition of mentoring (Jacobi 1991). The result is a somewhat confusing array of studie loosely aligned with the concept of rnentoring.

While ome of the re earch as esses the achievement of specific mentoring outcomes (as does the tudy to be described here). much of the publi hed work appear more concerned. with examination (If the mentoring process itself and how it i perceived by the participant. 'Or example. Rice and Brown (1990) surveyed undergraduate tudents acting as peer mentor . Their report focuses on how the participants felt about the memoring and how these feelings correlated with personal characteristics such as interpersonal skill and self-perception. In other educational research students provided descriptive and evaluative infermarion on their faculty mentors and the mentoring relationship (e.g., Cesa and Fro er, 1989~ Frier on. Hargrove, and Lewi . 1994; Wilde and Schau, 1991).

Research on mentoring has not "been driven or dominated by theory. Rather, current efforts are directed at determining just what the forms are that mentoring might take .and what some of the correlates and consequences are of these different forms of mentoring. Development of theoretical constructs to make sense of all this "du (bowl ernpirici rn" should be forthcoming. Meanwhile. some tentative efforts in the direction of conceptual, thinking have been suggested. J acobi (1991) describes three reports of factor anal y es of mentor behavior. These studies point to the existence of at least two robust factors: direct career assistance (providing information, problem solving, sponsor hip) and 50- cial-emotional support (providing acceptance, counseling) (e.g., Noe 1988). While rno l research deal with the effects of the mentoring relationship on protege . ai least one report suggests a more complex model that includes effe-C1S on the mentors themselves. Busch (1985) presents data indicatlng that mentors : ee benefits for themselves that are derived from participation in the mentoring relationship.

Is ues of gender and ethnicity have received considerable attention in the mentoring literature. It bas been uggested that mentoring effects may be enhanced by gender matching (Fitt and Newton, (981). The perception of shared values, experiences, and social networks, for example. may encourage the development of a bond between mentors and proteges of the same gender

EVALUAT(oN, OF MENToAING

729

(Hughes, 1988). In a number of universiry 'programs, an effort bas been made to match student.'; and mentors, on the basis of gender, ethnieity, or both (e.g, Jchnson, 1989; Meznek, MaGrath, and Garcia, 1989). The results have been mixed with regard to matching on gender. Frlersen, Hargrove, and Lewis (1994) reported thal students responded more positively to female than to male mentors. Wilde and Schau (1991) reported that some of the female students in their study commented on the importance to them of having a female mentor, Bur when tbe survey data were analyzed for differences attributable to gender of meOII:or at protege, no differences were found on ratings of such factors as supportiveness and fostering of professional development.

In a survey of men and women in business, D.reher and Ash (J990) found that individuals experiencing extensive mentoring relationships received more promotions, had higher incomes, and were mote satisfied with their cornpensa- 60n than i,f;I,dividuals experiencing less extensive ll1.entorlng relationships .. No gender differences were found, however. Women did not report fewer mentorillgcxperiences. than mea. While rnentoring relationshipswere associated wjth measures of career success, differenJial rnentoring experiences could not account for the observed difference in incomes between men and women. There may even be mentoring siruations in which gender matching results ill decreased effectiveness. III Nee's research (988).. mentors matched with proteges of (he opposite gender reported thal their students used the relationship more effecti vely than did proteges of the same gender:

Mixed results are also found in the research 011 ethnic matching. In 3. study of' African-American college student-s, Frierson; Hargrove, and Lewis (994) found that students who had African-American mentors reported more posltive attitudes (him did proteges who had worked with white mentors. By contrast, Atkinsca Neville, and Casas (1991) surveyed psychologists who bad experienced mentori.ng during graduate school or in their early years in practice. They found no evidence thatethnic matches resulted in higher prote.ge ratings of perceived mentoring benefits man did cross-ethnic pairings.

The evaluation. research on mentoring programs in education h.as tended to use weak designs. Most of me work relies 011 sclf-report measures in a retrospective. correlational design With the data: gathered at a single point ill time (Jacobi. 1991). Often the data are subjective and reported without adequate evidence of reliability and validity. Weak statistical analyses are not uncommon (e.g., Frier on, Hargrove, and Lewis, Smith and Davidson, 1992). Also, given that a primary reason for the: development of campus rnentering programs is the retention and enhanced academic success of at-risk ethnic .IlUnoriLy tudents, it is noteworthy that such ourcomemeasures have nor been included in [he pubIished reports. Accordingto Jacobi's review (1991). one of the best evaluations of a formal memoring program was conducted by Cosgrove (L986). That study showed the effect of mentoring on satisfaction with the university environment

730

CAMPBELL AND CAMPBELL

and developmental gains. Mentored students were comp red with a control group, but there was no assessment of the program effects on tudent academic performance, In fact, a review of the literature ha failed to locate a single .report of a control-group study dealing with the effects of a university mentoringprogram on undergraduate retentionand performance, Such a study is the subject of this report.

The main goal of thi: research was to examine whethe and how participation with a faculty mentor played a role in academic ucces _ Of particular interest was whether meeting with a mentor during a student's first year at the univerity affected grade-point average. units complete-d, and retention rate. in. ubsequem years. It was expected that student s who had access to mentors would how higher SC0re on measures of academic success thanwould comparison students who had not been mentered, The specific hypothese are:

flypothesisJ. Students in the mentoriog program will achieve a higher level of academic performance as mea ured by grade-point average (GPA) and will

complete more units of redit,

Hypothesjs 2. Mentored students will have a higher retention rat at the university and will graduate at a higher rate.

H pothesis 3. Academic performance and retention will be unrelated to gender of mentor or protege, or to the match in gender etwe n lhe two. This expectation follows from the fact that the existing research on gender effects either has received no support (Dreher and Ash, 1990; Wilde and Schau, 1991) or is upported primarily by methodologically weak. measures that are subject to response bias (Frierson. Hargrove, and Lewis, 1994' Nee, 1988).

Hypothesis 4. Academic performance and retention will be unrelated to ethnicity of mentor or protege, or to the match in ethnicity ber .... een the two. Thi hypothesis is made ill view of the finding that ethnic effect were either not supported Atkinsoa, Neville, and Casas, 1991} Or b sed on ubjeciive and unvalidared report! by proteges (Frierson, Hargrove. and Lewis. I 994).

The number and duration of c ntacts between proteges and mentors should

have an impact on the beneficial outcomes of participation in a meruoring program. Krarn (1985) has suggested thatthe amount of lime mentors and protege are in contact is a factor in obtaining the potential benefits of mentoring. To the extent that dyads meet regularly and pend meaningful amountsof time together, student proteges hould have the opportunity do uch things as askquestion • discuss issues. observe the mentor, and set goals. We expected that the more extensive the contact between mentors and proteges, the greater the potential would be for positive effect on. tudents academic success. Thi led to:

Hypothesis 5. The number and duration of men lor-protege coniacts will be positively correlated with GPA and negatively correlated with retention rare.

EVALUATION OF MENTORING

731

Hypothe es 1 and 2 arc surnmative in nature (Posavac and Carey, 1992) ill that they address the overall effectiv ness of the mentoring program relative (0 its primary goals-retention and academic performance. Hypotheses 3, 4. and 5 are elements of formative evaluation; theyaddress the proces or form of the program and involve variables that may help to explain why the program achieves or fails to achieve its stated goals. For example hypoihesl 2 addre ses the possibility of an interaction effect. The alternati e (as oppo ed to null) form of the hypothesi states that exposure to the menroring program wlll ha e a greater positive effect on protege. matched to same-gender mentors than it will on proteges matched to different-gender mentors. Data upporting such an irueraction would re uIt in greater understanding of program effect and could lead to modifications designed to improve the program.

In summary. the goal in this research was to test a. series of hypo the es about the effectiveness of an assigned rnentoring program against independently and objectively compiled data. This research does not test .1'1 theoretical model of meruoring, What it does is provide a straightforward test of whether an IIS-

igned mentoring program can contribute 1.0 student academic succe s. That is, will students who participate with an a signed mentor achie e higher gradepoint average', complete more units per erne ter, and be less likely to drop out of college than will student in a comparison group of nonmentored tudents?

METHOD

Program Description

The mentoring program evaluated here served as a. retention efforiae a large metmpolitan university on the West Coast. Its goal was to facilitate per ollal contacts between faculty and students. The purpose of these contacts was to provide a sistance [0 . tudents [0 help them succeed in reaching their academic goal and in graduating from the university.

Although allY student could apply LO participate, the target population Wa! students from ethnic group that were underrepresented at the university. Student were recruited in two way .. During the summer prior to ent ring the university as freshmen or as transfers. students in the target population were malled progrmll information and an invitation to participate. Students mo t of whom had already received the mailing, were also recruited dlredly during their participarion i,.l campus outreach programs such as Summer Bridge. Approximately 20% of all tudents who were contacted enrolled in the program

Faculty participants were recruited through university-wide distributions of the program description and an application form, About 10% of the entire faculty volunteered to mentor one to four students for an academic year. Students were matched with faculty on the basis of bared academic interests. Thus,

tudents who pecified a major on their application were assigned to a mentor

732

CAMPBELL AND CAMPBEll

_ ho e pec1alty wa. in the same e.g., history/history) or a related (history/ political science) field. Undeclared majors were assigned to mentors who agreed to work with uch students. Less than 1 % of the mentors, or students stated preferences (e.g., gender, ethnicity), and those few were accornmodeted whenever possible.

Mentors and students were encouraged to meet regularly throughout the year but were not required to adhere (0 a particular struCLUCC in their mentoring relationship. Mentors were asked to maintain a log of their contacts with protege that included date duration, and the general content of their meetings. These logs provided documentation that contacts between the student and the mentor had actually occurred.

In addition to their private meetings. the program offered a number of activities to create opportunities for mentors and students to spend rime together. During the academic year six work hops provided training on subjects as varied as mentoring styles and campus resources and career network development. There were social events uch as luncheons and free tickets to university theatrical productions. Small grant ($$500) were available [0 encourage mentors and tudents to initiate re earch projects or to attend profes ional meeting together.

Sample

Written data in the form of the mentors' logs recording the content and duration of contacts with prot6g~s wereavailable for 339 students. These were the students whose retention and academic achievement data: were evaluated in the present : ludy. The tudents participated in the rnentoring program during their first year at the university. Protege data were e amined [or three waves of participant beginning with the fall of 1992.

For comparison purposes. the mentored tudents were mal hed with 339 tudenn who had not enrolled in the program. The matching wa s done by mean of a computer routine that earched through all univer ity record for students who had not participated in the program but who were comparable in other characteristics. Each prOtege was matched with a control student \ ho matriculated in the same semester and year, was of the arne gender. arne ethnic group, arne entering class, and had the same entering OPA (for freshmen the entering GPA was their high school' OPA; for transfers it was their previous cumulative college GPA). After an exact match on year, gender, ethnicity, and class level wa . achieved, the program optimized on GPA The mean difference in GPA between the prorege and control group wru .00 (S.D. = .05 '. All comparative data were pulled from routinely compiled administrative records,

The resulting ample of 678 matched students was 37% male and 63% female. "ntering ch s level was 47% freshmen and 53% tran fer {primarily ju-

EVALUATION OF MENTORING

niors), Mean entering GPA was 2.82 (S.D. = 0.49). The ethnic distribution was 69% Latino, 22% African American, 3% Native American, and the remaining 6% were from a variety of other ethnic group.

The group of mentors was comprised of 126 faculty, administrators, and staff. Mentors werevolunteers who agreed to meet with their assigned studem(s) throughout one academic year. Many mentors "reenlisted" at least once; 74 participated for one year, 45 for two years, and 7 for three years. Participants from every college and from most departments served as mentors. Tltey represented a range of faculty levels (part-time instructor to dean). More females (72) than male- (54) volunteered. The majority of the mentor were Cauea ian (91). A number of ethnic minority mentors also participated in the program (Latino, 15; Asian 14; African American, 5; Native American. 1).

Variables

Academic achievement was mea. ured by GPA after the first semester, econd semester, and cumulatively at the lime the data were gathered, which could total as many as three years for some students, The mean number of units completed per emester was based on the first year of ' enroll men It at the university. Retention was indicated by whether the student was still enrolled in the spring of 1995 or had withdrawn from the university for say rea on other than graduation. Contact sheets maintained by rhementcrs were used to determine the number of contacts and total number of minutes of mentor-protege contact during the first year that the student participated in the program.

RESULTS

A series of r-tests W3$ conducted separately on each of the academic success variables: first ernesrer, econd semester, and cumulative (jPA, dropout rate, and graduation rate. In each lest, protege Scores were compared will). control group cores. Table 1 present the means and standard deviation for each variable by group. As the table shows, mentored students completed an average af .84 units more per ernester than did students in the control groups, c(338) = 2.63, P < .01-

There were consistent differences ill. GPA favoring the mentored students, The greatest programmatic impact on GPA occurred during the Iirst emester, At the end of their first emester at the university. protege grade were.3 of a grade point higher than those of tudents in the control group; the respective means were 2.50 and 2.20, I{J38) = 438, p < .001. This pattern of differences continued into the second ernester when the proteges (M = 2.32) exceeded their control eM = 2.14 t(338) = 3.4. p < .001 and was al 0 found curnulari cly (prorege M ;; 2.45, control M = 2.29, t(338) = 2.85, p < .01),

733

734

CAMPBEU AND CAMPBELL

TABLE 1. Mean Differences Between Mentored tudents and Paired Coutmls on Academic Achievement and Retention Measures

PrOlege Control
Mean (S,D .. ) Mean (S.D.)
Units/Sern, 9.33 (4.08 8.49 (4.53)"
1st Sem, OPA 2.50 (0.93) 2.20 (1.11)*'"
2nd Sem. GPA 232 1.12) 2.14 (1.22)1<*
Cum. GPA 2.45 10.81) 2.29 (0.9·)*
Drop Rate 14.5% 26.3%"·
Graduated 6.0% 6.0% p<.OI~ "p < .001.

Note.' Comparisons tested with I for dependent sample', except or drop rate. which WllS tested with chi-square,

Only one of the two measures of retention revealed ef'Cects fo-r rnentoring.

The dropout rate (students who failed to reenroll in any semester) among prottge was. about half of that for students in the control group, 14.5% versus 26.3%, l?(l) = 14.56, P < .00 I. There was no difference between the two groups on their rate of graduation. It may bepremature, however, 10 as ess graduation rate becau e only 6% of the student have completed enough units to graduate.

Consistent with our hypothesis, Table 2 bows that academic success and retention rate were unrelated to the gender of the protege or of the mentnr, Male and female protege did not diller igniflcaatly in units completed. the variou measures of GP ,or dropout rate, Likewise, pro~ge gender affected neither the average number of contacts nor the total duration ofthose contacts with mentors. Mentors of either gender apparently w re equally capable of encouraging proteges. With one exception, the measures of academic success and contact did not significantly differ for. mdents of male or female mentors. The one exception to that pattern was that female mentors had an average of 1.45 more contacts witb their proteges than did male mentors (female M = 7.95. male M = 650. l(334 = 2.68. p < .OJ).

As we have noted above, there is little empirical evidence but persi tent anecdotal support for the view that gender matching between proteges and mentors enhances the outcomes of mentoring program. In the present ample there were 209 students who were of the same gender a their mentors and 128 who were not As the two Jeft columns of Table 3 demonstrate, gender matching had very little effect on academic performance, There was, however. one significant difference favoring the gender-matched student on the total number of contacts with their mentors. The means were 7.88 contact for gender-matched swdenlS

EVALUATION OF MENTORrNG

735

TARLE 2. Academi Achievement and entacts as a Function of Protege Gender
ond 1entl)f Gender
Mentee (student) Menlor (faculty)
Male Female Male Fcmal~e
N :125 213 149 188
Unus/ em. 9.08 (4.0 ) 9.48 (4.08) 9.57 (3.98) 9.12 4.16)
lst Seru, GPA 2.44 (0.94) 2.54 (0.93) 2.58 (0.91) 2.44 (0.9S)
2nd Sem, GPA 2.19 (1.14) 2.39 (1.10) 2.43 (1.10 2.23 (1.13)
Cum. GPA 2.36 (0.81) 2.51 (0.81) 2.51 (0.80 2.40 (0.81)
Drop Rate 15.1% 14.1% 16.1% 13.3%
# Ccmacts 6.65 (4.23) 7.66 (5.33) 6.50 (4.321 7.95 (5.35)**
Contact TIme 111.34 131.63 111.34 13'5.02
(131.63) 1 179..13) 169.52) (1.58.66) *1' < .01.

Note: Comparisons tested with I for lndepeudem sample . except for drop rille. which was tested whh chi-square.

TABLE 3 .. Academic Achi.cl'cment and ontaces as a Function ef Match Uehlfcen

Prlllte.ge and Mentor Gender
Female-
Genders Genders Male-Male Female
Matched Differed Match Match
N 209 128 73 135
UnitlilSem. 9,11 (4.0 ) 9.64 (4.09) 9.14 C'tr.)9) 9.14 (4.13
lSI Sem. GPA 2.49 (0.94) 2.53 (0.93) 2.51 (0.93) 2.47 (0.94)
2nd Sem. GPA 2.23 (1.11) 2.47 (LIZ) 2.19 (Ll2) 2.25 (Ll!)
Cum.GPA 2.41 (0.83) 2.54 (0.79) 2.36 (0.83) 2.42 (0.83)
Drop Rate 15.9% 12.4% 19.2% 14.l%
# COlllaCI'S 7.88 (5.04) .39 (4.69).u. 6.74 (3.99) 8.50 (5.50)*
Contact' Ime 123.59 126.30 96.07 138.21*
047.09) (187.86) (113.57) (J61.30)
- p < .05; np < .01.

Note: Comparisons tested wlrh I for independent samples. except for drop rare, which was tested with chi-square,

736

CAMPBELL AND CAMPBELL

and 6.39 contacts for their control, (335) = 2.72, P < .01. While number of contacts was greater for gender-matched pairs, the e contacts. were also briefer on average because the total cont ct time was lower for gender-matched proteges. although not significantly so.

A different way to examine tne impact of gender matdting is to assess whether there was a differential effect among matched dyads for gender of the dyad. The right columns of Table 3 pte ent the results of these analyses, As with the comparisons of marched and unmatched pairs, gender of the matched pairs al 0 was not associated with me academic performance outcome of mentaring. Similar to the pl viou analysis, however there were apparent effects far number of contacts andtotal duration of the contacts. Male-male pairs averaged 6.74 contacts while female-female pair averaged 8.50 contact, (206) = 2. 3. p < .05. The mean differences in duration of contacts were 96.07 minutes for male-male pairs and 138.21 minutes for female-female pair, 1(206) = 1.98, p > .05 .. Wrlile the female protege/mentor pairs met more often and for longer period of time than did the male pairs, these two dependent variables did not show any significant relationship to academic performance.

Et.hnicity

A central hypothesi, of thi evaluation was that ethnicity of the protege, the mentor, or the ethnic match of the pair would be unrelated either to academic performance or to retention. Table 4 presents the results relat.ed to this hypothesi . Analysis of variance on the, e means showed 110 significant differences

TABLE 4. At"ademic Achievement and Contacts ru D Function or Protege IEthniclty

African ative
Latino American American Other
N 235 75 9 20
Unils!Sem. 9.40 (3.99) 9.05 (4.51) 10.9<1 (2.87) US (3.89)
I Sl Scm. GPA 2.57 (0.&2) 2.25 (.1.01) 2.68 (0.90) 253 (0.69)
2nd Sem. GPA 2.35 (0.70) 2.09 (Ll3) 2.86 (J .1.1) 2.53, (1.14)
Cum. G.PA. 2.51 (0.82) 2.23 (0 .. 80) 2.66 (0.66) 2.56 (0.78)
Drop Rate 15.1% 13.3% l1.t% 5.0%
# Contacts 7.23 (4.71) 7 1 (5,71) 8A5 (5.47)
Contact Time 123.03 116.28 159Y
(167.48) 149.18) I 72.29) EVALUATION OF MENTORING 737
TA8LE 5. Academic Achievf!ment and Contacts as a Function of
Mentor Etbnicit.y
African Native
Caucasian Latino Asian American American
N 224 58 :36 I! .,
UnitslSem. 8.97 10.25 9.74 B.98 11.36
(4.51) , 3.93) (3.99) (4.51) (3.84)
1st Sem, GP 2.50 2.62 2.45 2.46 2.19
(0.91 ) (0.96) (1.7) (0.57) (1.29)
2nd Sem. GPA 2.28 2.4& 2.29 1.98 2.81
(1.12) 112) (l.O7) ( 1.49) (0.82)
Cum. GPA 2.40 2.71 2.41 2.14 2.68
(0.84) (0.67) (0.85) (0.74) (0.68)
Drop Rate 15.6% 15.5% 11.1% 9.1% 0%
# Ccntacts 7.57 6.53 7.17 8.82 4.71
(5.14) (3.0l) (4.76) (8.81) (4.46)
Contact TIme 133.92 93.53 117.25 139.64 116.71
(170.47) (127. 6) (160.89) (203.20) (161.33)
NOIe: Comparisons tested wilb ANOVA. except for drop rate. which was rested with chl-squure. between protege ethnic groups 011 academic achievement or retention. Similarly, there were no differences due to erhnicityon the average number of contacts with mentors or on the total duration of comacts. An analysis of variance was conducted using ethnicity of the mentor as the grouping factor (Table 5). Again, the stati tical analysi revealed no significant differences on protege academic performance, dropout rate, or number of contacts. Similar to tile analysis of gender-matched dyads, the i 'sue of ethnically matched proteg6/mentor pairs wa as es: ed, Table 6 . hows that among the 76 matched and 260 unmatched pair. there were no significant differences 0.0 any of the achievement or contact variable. This set of analyse clearly demonstrates that neither ethnicity of the protege nor of the mentor was a significant factor in the effe tiveness Of the mentoring program,

Amount of Contact

During their year in the program. proteges averaged 7.28 contacts (S.D. = 4.97) with their mentors. The mean total contact time was 124.25 minutes (S.D. = 163.41). Table 7 shows the correlations between degree of mentor-protege contact and the major academic outcome variables. While the correlati n are not large. all bUL one how a ignificant tendency for tudents who have greater contact with their mentors to perform better as measured by units completed

738

CAMPBELL AND CAMPBELL

TABLE 6. Academic cbievement aod ontacts as a Func:lion of Match Dell een PrQ1ige and Mentor Etlmielty

Ethnicity Elhnicity
Matched Differs
N 76 260
Uni'lS/Sem. 9.90 (4.18) 9.16 (4.06)
lst Sem. GPA 2.S1 (0.92) 2.50 (0.94)
2nd Scm. GPA 2.38 eLl7) 2.30 (l.J I)
Cum.GPA 2.57 (0.69) 2.42 (0.85)
Drop Rate 14.5% 14.6%
# COnraCI$ 7.28 (4.95) 7.34 (4.96)
Contact Time 110. [ 129.20
(137.66 (170.46) NOlI!: Comparisons te ted with t "for mdepcndeni groups, except r r drop rille, which Wall tested with chi-square.

and by GPA. The quared con-elations indicate that only 1.2% to 3.60/0 of tile variance in academic performance can be attributed to variation iII mentorprotege contact.

DISCUSSION

The first objective of this e aluative research was to determine whether the university mentoring program succeeded in enhancing the academic performance of undergraduate students. It was hypothesized that participation in: the program would result in more units completed per semester and higher grades (as measured by GPA). Thi bypotbe, is was supported. When compared with their matched controls, students in the program completed slightly less than one additional unit per semester and attained a higher grade-point average equivalent to between .2 and .3 of a grade polnt. Wl:!lIelhe increase in completed units is relatively mall for any particular seme ter the cumulative effect of units completed over a four- or five-year university career could represent an eight-

TAU.LiE 7 .. COr:~1:8tions Between Fr~qucm:y of Menbu··Prolege Contact, Total Contact Time and Academic Achievement

2n<L Sem.

GPA

Cumulative GPA

TOlal Units

1'1 Scm.

GPA

.05 .17**

No. of Contacts Contact Time

.12 .13'"

,11* .21"'1'+

.11* ,19"''''

*p < .05; *p < .01; .up < ,001. N = 338.

EVALUATION OF MENTOAING

10 len-unit difference between the tt 0 groups. It was also hypothesized that prot6ges would have a lower dropout rate than their matched controls. Support was again found; the dropout rate for protege wa approximately half that of their controls (14.5% vs. 26.3% respectively). If this differential. dropout rate is attributed to the rnentoring program, then the program was responsible :for retention of 40 targeted minority student. who would otherwi e have dropped out. If the time period of the study were extended to cover the typical period of an undergraduate education (four or fi e years a oppo ied to . everal semesters). then the actual number of retained miaoritie . could be can iderably larger.

The. e findings provide good support for the conclusion (hat the 'program being evaluated did indeed cau e the rep rted academic gains. The upport would have been even tronger if tudents had been randomly assigned to participate in rhe mentor program or the control group. Randomized control is. unfortunately, very difficult to achieve in organizational research outside the laboratory. Use of matched control was the most feasible approach in the research reported here. Each protege was paired with a control who was matched In gender, ethnicity, and prior GPA. By matching 011 GPA, students in the program and their controls were paired a'> do ely a. possible on mental ability, study kills academic motivation, and resource . GPA acted as a surrog te measure for underlying ability, motivation. and other factors relevant to academic p rf00110l1'lCe. To under land the logic here, con ider an alternative explanation for the enhanced GPA of proteges as compared to their con rols, One could suggest that students who volunteered for the mentor program were more motivated as students than tho. e who declined and v ere selected as controls. But, if academic mati arion i. viewed. as a relatively enduring characteristic, then it would have been operating in the past and would have been accounted For by marching students on the results of uch motivation-prior GPA. To argue that motivation suddenly shifted and was maintained thereafter for some students and that such a shift accounts Ior the apparent pr gram effect requires an awkward line of reasoning. Yet such reasoning mUSL remain as a possibility.

The use of randomized control groups is iii highly desirable practice in field research. Yet in the review of studies of university mentoring program for undergraduate . no reports were found in which randomized groups were used to ldentifyeffecrs on independently measured academic achievement, Only one study wa .. located that used a randomized-group de ign (Cosgrove. 1986), but il did not deal. pecifically with academic outcomes, The research presented here is unu ual in thal it combine experimental control and objective outcome measures of academic performance in an effort to identify program effectson targeted minorities and women.

It has been suggested that mentor-protege pair matched on gender. will be more effective than those of different genders (Hughes. 1988). The research has been mixed on this j ue, so an analyst of gender matching was included here. It was hypothesized that gender matching would have no significant effects on

739

740

CAMPBELL AND CAMPBELL

academic measures. The' results showed gender matching did not influence units completed GPA, or dropout rate. There was a difference, however; in number of contacts between mentor and their protege .. ln their first year in the program students who matched their mentors in gender totaled 7.9 contacts compared with 6.4 for those who did not match in gender. In a separate analy. is, it was found that the female-female pairs had contact more frequently than did the male-male pair (8.5 contacts vs. 6.7 contacts, respectively), Thi difference WaS associated with greater total contact time, again favoring the femalefemale pairs. Such a difference may be simply another validation of the traditional ex-role rereotype in American society. That is, female-female relationships are characterized a more openly social, more intimate, and more af~ fective than we corresponding male-male relationships. For our purposes, the nature of gender differences in social interaction is of Ie s import than the finding that gender matching was not associated with difference in academic outcome measures for the undergraduates included in 'this study.

The literature on ethnic match between mentor and protege parallels that on gender matching. That is, arguments and rationales an be found both supporting, and refuting the benefits of such matching. The empirical research is mixed onthi issue. We tested the hypothesis that ethnic matching would be beneficial but we did not anticipate support. The data on academic outcomes howed no differences thai could be attributed to ethniciry .. Specifically, the proteges belonging to differeut ethnic groups did not differ in academic succes •. Nor did the proteges differ when grouped by ethnicity of the mentor, Fin lly, there were no ignillcant difference in academic outcome measures when student-protege pairs who were of the same ethnic group were compared with pairs who differed in ethnic identity.

Previous research on gender and ethnic matching effects eems to show a pattern. When matching is shown to have an effect, the dependent variables are often measures of attitudes and preferences. For example, Frierson, Hargrove, and Lewis 1994) found gender and ethnic effect. on reported attitudes about the mentoring. Similarly. Ugbah and Williams (1989) found that studenu indicated preference for mentors of their OWn gender. When the dependent measure are objective and behavioral. effects associated with gender and ethnic march are less common. Thi may explain wby such effects are found les often in the organizational behavior literature where the outcome measures are often objective (e.g .. salary attained, rate of promotion). The factor that explain beliefs. preferences, and expres ed ati faction may be quit.e different from those that explain overt behavlor and its direct consequences. In the research reported here, no elf-report measure are involved, and no associations with gender or ethnic status were found.

The final hypothesis predicted that greater student contact with mentors will be associated with greater academic gains by the students, This hypothesis re-

EVALUATION OF MENTORING

741

ceived modest. upport. Low but significant correlations showed that the more contact (and more contact time) between mentor and protege, the greater the academic achievement (measured by units completed and by GPA). This may indicate that more exren ive mentoring will bring about slightly greater academic performance. Alternatively it may simply reflect the tendency for the more capable students 10 take greater advantage of an relevant re ources-eincluding access to mentors. U would be desirable to have data relating to which contacts were initialed by the student and which were initialed by the mentor. [f most were mentor initiated, then the c rrelations would appear to reflect program effectiveness. But even here. it could be argued that mentors prefer to initiate contact with the more apable and respon ive tuden . leading to the reported correlations between number of contacts and academic achievement.

The research reported here supportsseveral recommendation. For university mentor programs emphasizing objective criteria such as student retention and academic performance. it appears that matching student with same-gender or

ame-erhnicity mentors i. not an important consideration. However .. if tudent preference are a dri ving force in the operational details of the program. then such matching may deserve serious con lderation, This matching is ue de erves further exploration in future research. It may be thar the rna l important variables for optimal matching ba e liule to do with gender and ethnicity, Researchers may dowell LO explore matching on preferred learning style. world outlook, communication style, and a number of personality variable . Another direction for future work i to conduct descriptive studies of university mentor programs as currently conducted. A taxonomy of pas ible programs could be developed with the programs arrayed on such dimensions as degree of structure, university size, and nature of organizational climate.

We conclude with a comment all methodology. As long as field experiments u ing randomized group are rare in the literature. uncertainties will remain in the interpretation of social action programs. Yet. given the commitment of American universities (0 addre the challenge of social diversity, evaluation research rnu t proceed using all the experimental control and power attainable within each re earchsitnation, The u e of matched conuols and objective data constitutes a reasonable compromise for tho e seeking to understand the impact of their programmatic efforts, The pre ent study demonsrrares one path toward

uch understandmg ..

REFERENCES

Atkin on. D. R.. Neville, H .. and Casas, A. (19 1). The mentorship of ethnic minorities in professional psychology. Professional' Psychology: Research and Practice 22: 336- 338.

742

CAMPBELL AND CAMPBELL

Busch, J. W. (1935). Mentoring in graduate schools of education: Mentor' perceptions.

American Educational Research Journal 22: 257-265.

Ce a, I. L., and Fraser. S. C. (1989). A method of encouraging the development of good men tor- protege relntionships. Teaching of Psycho lORY 16: 125-128.

Cosgrove .. 1'. 1. (1986). TIle effects of particlpation in a mentoring-transcript progrllm on freshmen. Journal of College Studen: Persatmel 27: 119-124.

Dreher, G. F., and Ash, R. A. (1990). A comparative ~1lldy ofmentoring among men and women in managerial, professional. and technical position. Journal of Applied PsycfroloK}, 75: 539-546.

Fagenson, E. A. 1989). The mentor advantage: Perceived career/job experiences of proteges versus non-proteges. Journal 0/ Organizational BehlITljQr 10: 309- 20.

Fill, L. W., and Newton, D. A. (1981). When the mentor h; a man and the protege is a woman. Harvard Business Review 59: 56-60.

Frierson, H. T., Hargrove, B. K .. and Lewi • N. R. (1994). Black summer research suidents' perceptions related to research mentors' race and gender. JOllril(L{ oj College SII/dell( Development 35: 475-480.

Hughes. M. S. (1988). Developing leadership potential for minority women. In M. A. D.

Sagaria (ed.), Empowering WOlllen: Leadersllir' Developme/lt Sm;uegie,I 011 Campus .. New Directions for Stlldl'nr Services: No. 44 (pp, 63-75), San Francisco: Jessey-Bass.

Jacobi, M. (1991). Mentoring and undergraduate academic success: A literature review.

Review of Educational Research 61; 505-532.

Johnson, C. S. (1989). Mentoring programs. 10 M. L. Upcraft and 1. Gardner (eds.), The Freshman Year Experience: Helping Students Survive and Sue eed ill College. (pp, 118-l28). San Francisco: 10 ey-Ba ...

Kram, K. E. (1985). MetttorinK at Work: Deve!0pflumla/ Relationships ill Organizational Life. Glenview, II.; S ott. oresrnan & Co.

Meznek, J., McGfllln, P., and Garcia, F. (1989). The Puente Project, A Report TO the Board a/Governors, CCllifomia Community Colleges. Sacramenro: Office of [he Chancellor (£tHe Document Reproduction Service No. ED 307 920).

Nee, R. A. (1988). An inve ligu.i.an of the determinants of successful assigned mentoring relationships. Personnel P ... ,(;Iw/08-V 41: 457'-479.

Posavac, E. 1., and Carey, R. G. (1992). Program Evaiumion; 4lh cd, Englewood Cliffs, NJ: Prentice Hall.

Queen. K. W. (1994). Meeting affective needs of at-ri k adolescent . Psychological Repari 74: 753-754,

Rice, M. B., and Brown. R. D. (1990). De eloprnenral factors associated with elf-perceptions of mentoring competence and mernoring needs, Joumal of College Studen:

Devetopment 31: 293-299.

Smith, . P., and Davidson, W. S. (1992 . Mentoring and the development of AFricanAmerican grad.uate students. Journal of Coiieg« Smdenl Dellefopmellt 33: 531-539.

Ugbah, S., and Williams, S. A. t 1989). The mentor-protege relatlonship: Its impact on blacks in predominnnll·y while instiuniens, In J. C. Ham (ed.), BJack.\~ in Higher Educaiion: Overcoming lire Odds (pp, 29-42). Lanham, MD: Universjty Press of America.

Wilde, J. B .• and Schau, C. G. (1991). Mentoring in graduate schools of education:

Mentees' perceptions. Journal of Experimental Education 59: 165-:179 ..

Received August 2. 1996.

Copyright of Research in Higher Education is the property of Kluwer Academic Publishing and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.

You might also like