personality, classroom behavior, and student ratings of ... · personality, classroom behavior, and...

Journal of Educational Psychology1985, Vol. 77, No. 4,394-407

Copyright 1985 by the American Psychological Aaaociaton, Inc.0022-0663/85/$00.75

Personality, Classroom Behavior, and Student Ratings ofCollege Teaching Effectiveness: A Path Analysis

Stephen Erdle, Harry G. Murray, and J. Philippe RushtonUniversity of Western Ontario

London, Canada

This study tested the hypothesis that classroom teaching behavior mediatesthe relation typically found between personality and college teaching effec-tiveness. Colleagues rated 37 full-time college instructors on 29 personalitytraits, and trained observers assessed the frequency with which the same in-structors exhibited 95 specific classroom teaching behaviors. Instructionaleffectiveness was measured by global end-of-term student ratings averagedover a 5-year period. Path analyses revealed that approximately 50% of therelation between personality and teaching effectiveness was mediated byclassroom behavior. Results are discussed in terms of the validity of studentratings of teaching and in relation to Dunkin and Biddle's (1974) model ofclassroom teaching.

Student ratings are currently the mostwidely used measure of teaching effective-ness in North American colleges and uni-versities. Seldin (1980) reported that 95%of liberal arts colleges considered studentratings in the evaluation of teaching per-formance. Moreover, reviews by McKeachie(1979), Murray (1980), and Marsh (1984)concluded that student evaluations canprovide reliable and valid information oncertain aspects of college teaching effec-tiveness. Evidence for the validity of stu-dent ratings includes the fact that ratingsshow relatively low correlations with extra-neous variables such as class size and severityof grading and relatively high correlationswith objective measures of studentachievement (Cohen, 1981), and with theo-retically relevant instructor characteristicssuch as personality traits and classroom be-haviors. The latter teacher attributes wereclassified by Dunkin and Biddle (1974) as"presage" and "process" variables, respec-tively.

Personality and Teaching Effectiveness

Teacher personality traits have often beenhypothesized to be associated with student

Requests for reprints should be sent to StephenErdle, Department of Psychology, University of West-ern Ontario, London, Ontario, Canada N6A 5C2.

evaluations of college teaching. Althoughseveral early investigations using instructor'sself-reports of personality failed to supportthis belief (e.g., Bendig, 1955; Sorey, 1968),more recent studies in which personality wasmeasured by methods other than self-reportyielded positive results. Colleague ratingsof teacher personality have been shown torelate to student evaluations of teaching,with the effective teacher perceived asshowing leadership, objectivity, low anxiety,and extraversion (Murray, 1975); extraver-sion and task orientation (Murray, 1978);and leadership, extraversion, supportiveness,objectivity, and nonauthoritarianism(Rushton, Murray, & Paunonen, 1983).Student ratings of teacher personality alsohave been shown to relate to student evalu-ations of teaching, with the effective teacherperceived as showing ascendency, responsi-bility, emotional stability, sociability, origi-nal thinking, personal relations, and vigor(Costin & Grush, 1973); dynamism, prag-matism, amicability, and high intelligence(Sherman & Blackburn, 1975); and affilia-tion, achievement, endurance, nurturance,understanding, changeability, and extra-version (Tomasco, 1980).

In sum, a reasonably consistent pattern ofpersonality characteristics associated withstudent ratings of college teaching hasemerged from research. The highly ratedteacher is perceived by faculty peers and bystudents as showing leadership, objectivity,

394

PERSONALITY, CLASSROOM BEHAVIOR, AND TEACHING 395

and high intellect on the one hand and sup-portiveness, extraversion, and emotionalstability on the other. Due to the correla-tional nature of the research, however, causalinterpretations have been made with diffi-culty. Rushton, Murray, and Paunonen(1983) suggested that a reciprocal causalinteraction may exist in which success andreinforcement at teaching fosters a "teacherpersonality," which in turn leads to greatersuccess in the classroom. They pointed out,however, that there are good reasons forbelieving that the direction of causality isprimarily from personality traits to teachingeffectiveness rather than vice versa. Im-plicit in the notion of personality trait is theidea of cross-situational and longitudinalstability of behavior (Allport, 1937; Guilford,1959; Jackson & Paunonen, 1980), whichmakes it likely that an instructor's traitsinfluence his or her teaching effectiveness.

Classroom Behavior and TeachingEffectiveness

Several studies reported correlations be-tween student ratings of instructors' specificclassroom behaviors and student ratings ofoverall teaching effectiveness. For example,Mintzes (1979) found that highly ratedpsychology instructors showed behaviorssuch as speaking expressively or emphati-cally, giving a preliminary overview of thelecture, using multiple examples of concepts,and addressing students individually byname. Other studies showed that studentrating were predictable from behaviors re-flecting enthusiasm and interaction (Solo-mon, 1966); motivation of students, struc-turing of course material, skilled presenta-tion, and mastery of subject content (Desh-pande, Webb, & Marks, 1970); and consid-eration and competence (Keaveny &McGann, 1978). Because students assessedboth classroom behavior and teaching ef-fectiveness inthese investigations, it is pos-sible that correlations obtained were due inpart to judgement biases such as a "halo ef-fect." Similar results occurred, however,when behavior and effectiveness were inde-pendently assessed. Murray (1983), for ex-ample, using trained observers' ratings ofclassroom teaching behavior, identifiedspecific behaviors reflecting enthusiasm,

clarity, and rapport that were positively re-lated to student evaluations of overallteaching effectiveness. Similarly, Crantonand Hillgartner (1981) reported a variety ofrelations between student instructionalratings and behavioral frequency countsderived from videotapes of classroomteaching.

The classroom behaviors found to corre-late positively with student ratings ofteaching effectiveness seem to fall into twogeneral categories: those that convey en-thusiasm and/or rapport and thereby elicitstudent interest and participation (char-ismatic behaviors); and those that reflectpreparation, structuring of materials, andclear exposition of concepts (organizationalbehaviors). Charismatic behaviors includespeaking expressively, relating material tostudents' interests, and use of movement andgesture. Organizational behaviors includegiving a preliminary overview, stating ob-jectives, and using headings. It is interestingto note that Frey (1978) arrived at a similartwo-dimensional model of college teachingeffectiveness through factor analysis of be-haviorally oriented student rating forms.

Although causal interpretations again arerestricted by the correlational nature of theresearch, two considerations support thehypothesis that classroom teaching behaviordetermines student ratings of teaching ef-fectiveness. First, the normal temporalprecedence of teaching behavior to teachingevaluation is consistent with such a hy-pothesis. Second, and more important,there is experimental evidence supporting acausal relation between teacher behaviorsand student ratings. For example, Murrayand Lawrence (1980) reported that teacherstrained on specific classroom behaviors suchas vocal variation and expressive movementobtained significantly higher student ratingsthan did control teachers.

Personality and Classroom Behavior ofEffective Teachers

Given that both personality traits andclassroom behaviors appear to influencestudent evaluations of teaching, the presenthypothesis was that these instructor char-acteristics are themselves related, such thatclassroom behavior mediates the influence

396 S. ERDLE, H. MURRAY, AND J. P. RUSHTON

of personality on teaching effectiveness.Several considerations support such an in-terpretation. First, many conceptualiza-tions of personality view traits as hypothet-ical constructs inferred from prototypic be-haviors. From this perspective, the effect ofa teacher's personality on student ratings (orany other criterion) must be mediated by hisor her behavior. Given that student-teachercontact is largely restricted to classroomsettings, this behavior is most likely to beclassroom behavior. Second, theorists ineducational psychology (e.g., Dunkin andBiddle, 1974) have proposed models ofclassroom teaching in which "presage"variables, such as teacher personality traits,determine "process" variables, such as spe-cific classroom behaviors, which in turn de-termine "product" variables, such as studentevaluations of teacher effectiveness. A finalconsideration stems from the empirical datapresented above. An intuitive congruenceappears to exist between the types of per-sonality traits and the types of classroombehaviors found to be associated with stu-dent ratings of teaching effectiveness. Forexample, charismatic classroom behaviorsmight be expected to correlate with person-ality traits such as extraversion and sup-portiveness, whereas organizational class-room behaviors might be expected to corre-late with traits such as orderliness andleadership. In light of these considerations,a causal model in which classroom behaviormediated the effect of personality onteaching effectiveness was constructed.Path analytic techniques were used to testthe adequacy of this model.

Table 1Peer Rating of Personality Traits: Reliabilityand Correlation With Student InstructionalRatings

Method

Subjects

The sample of teachers consisted of 35 male and 2female full-time faculty members of varying rank, eachof whom had taught for at least 3 years in the depart-ment of psychology at University of Western Ontario.Because of the small number of women, all analyseswere collapsed across sex.

Measures of Personality

All faculty members in the department, includingthose participating in the study, were mailed a set of 29trait adjective names (see Table 1), along with trait

Personalitytrait

MeekAmbitiousSociableAggressiveIndependentChangeableSeeks definitenessDefensiveDominant

EnduringAttention - seekingHarm-avoidingImpulsiveSupportingOrderlyFun-lovingAesthetically

sensitiveApproval-seeking

Seeks helpand advice

Intellectuallycurious

AnxiousIntelligentLiberalShows leadershipObjectiveCompulsivenessAuthoritarianExtravertedNeurotic

Reliabilityof

mean rating8

.83

.92

.89

.88

.71

.74

.85

.77

.93

.91

.94

.83

.89

.86

.61

.90

.72

.78

.76

.74

.80

.88

.88

.92

.81

.70

.72

.88

.65

Correlation withstudent ratings

-.08.35*.52*.03.18.46*.12

-.30*.35*

.42*

.50*-.41*

.20

.43*

.40*

.39*

.34*

.32*

.11

.46*-.33*

.36*

.55*

.57*

.40*

.26-.32*

.51*-.22

aBased on an average of 10.51 raters.*p < .05.

definitions and instructions on how to rate severalnamed colleagues on 9-point scales. The instructionsemphasized that ratings were to be based solely onpersonal observation and were to be made relative toother university professors rather than to other peoplein general. Twenty of the traits were adapted from thePersonality Research Form (PRF; Jackson, 1974), anomnibus personality inventory based on H. A. Murray's(1938) need definitions. Two traits, extraversion andneuroticism, represented scales on the Eysenck Per-sonality Questionnaire (EPQ; Eysenck & Eysenck,1975). Another seven dimensions were selected becausethey were found to be useful in H. G. Murray's (1975)study of personality and college teaching. Between 8and 15 faculty peers (Af = 10.51) assessed each partic-ipating faculty member on the 29 personality traits.


Measures of Classroom TeachingBehavior

Each of the 37 professors was observed during threeseparate 1-hr class periods by each of 3 to 11 trainedclassroom observers (M = 6.41). Thus, faculty mem-bers were observed for an average of 19.23 hr each overa period of approximately 3 months. The observers,students enrolled in an educational psychology course,were randomly assigned to teachers. Prior to visitingclasses, observers were given approximately 4 hr ofgroup training in recording classroom behaviors fromvideotaped lecture segments. The instructors hadgiven written permission for classroom observation tooccur, but did not know when student observers wouldbe present. The observers summarized their 3 hr ofobservation of each teaoher on a standardized behav-ioral rating form developed by H. G. Murray (1983) thatwas called the Teacher Behaviors Inventory and com-prised 95 specific classroom behaviors (see Table 2 forpartial list) rated on 5-point frequency of occurrencescales.

Measure of Teaching Effectiveness

Formal end-of-course student evaluations of in-structors are required in all courses at the University ofWestern Ontario. In the department of psychology,students evaluate teachers on ten 5-point rating scales,assessing various aspects of classroom performance.The last item, "How would you rate your instructor interms of general, overall effectiveness as a teacher?",was used as the criterion of teaching effectiveness.Overall effectiveness ratings were averaged over allundergraduate courses taught over the past 5 years (M= 10.73 courses per teacher) to obtain a single measureof teaching effectiveness for each professor. A split-halfreliability coefficient of .91 was obtained for this mea-sure by correlating mean student ratings from odd- andeven-numbered courses across all professors and cor-recting the resulting correlation by the Spearman-Brown formula.

Results

Reliability of Personality and ClassroomBehavior Ratings

Split-half reliabilities were computed foreach of the personality traits by correlatingmean ratings of odd- and even-numberedjudges across all professors and correctingresulting correlations by the Spearman-Brown formula. As shown in Table 1, meanrater reliabilities for the 29 personality traitsranged from .61 to .94 (M = .82), indicatingconsiderable agreement among faculty peersin their ratings of colleagues.

Split-half reliabilites, corrected by theSpearman-Brown formula, also were com-

puted for each of the 95 classroom behaviors.Because of the relatively small number ofobservers rating some professors (as few as3), the reliability of 46 of the 95 classroomteaching behaviors was too low (below .50)to justify inclusion in subsequent statisticalanalyses. As seen in Table 2, the remaining49 classroom behaviors had moderate relia-bilities ranging from .50 to .86, with a meanof .64.

Relation of Personality and ClassroomBehavior to Teaching Effectiveness

Mean peer ratings of 21 of the 29 person-ality traits correlated significantly withstudent ratings of teaching. As shown inTable 1, the significant correlations rangedin magnitude from .30 to .57, with an averageof .42. As in previous research, the highlyevaluated teacher was characterized by traitssuch as leadership, objectivity, extraversion,supportiveness, emotional stability, highintellect, ambition, endurance, and domi-nance.

Mean observer ratings of 26 of the 49classroom behaviors also correlated signifi-cantly with student ratings. As shown inTable 2, significant correlations ranged inmagnitude from .27 to .66 {M = .44). Againresults corresponded to those of earlier re-search. The effective instructor exhibitedtwo general types of classroom teaching be-havior: one reflecting charisma and exem-plified by behaviors such as speaking ex-pressively, use of humor, relating subjectmatter to student interests, and encouragingstudent participation; and the other re-flecting organization and exemplified bybehaviors such as giving a preliminaryoverview, using headings to organize mate-rial, and giving multiple examples of con-cepts.

Construction of Composite Personalityand Classroom Behavior Dimensions

The next step in data analysis was theconstruction of composite personality andclassroom behavior dimensions for use insubsequent analyses. Aggregation of indi-vidual personality traits and classroom be-haviors into composite dimensions provides


Table 2Observer Ratings of Classroom Behavior: Reliability and Correlation With StudentInstructional Ratings

Behavior

Gestures with hands or armsPlays with chalk or pointerExhibits facial gesturesAvoids eye contactSmokes or drinks while teachingLeans on desk or lecternSmiles or laughs while teachingSpeaks in an expressive wayStutters, mumbles, or slurs wordsSpeaks in a monotoneDoes not speak clearlyGives multiple examples of concepts

Fails to define new termsUses audiovisual aidsUses graphs or diagramsUses headings to organize lectureRelates lecture to readingsPuts outline on blackboardLecture follows logical sequenceCovers little material in lectureGives preliminary overview of lectureOrganizes lecture with pointsProvides outline of the courseTells jokes or humorous anecdotes

Shows interest in subject matterRelates material to student interestsDescribes relevant personal experiencesReads lecture from prepared notesUses a variety of different activitiesStates own point of viewFocuses on controversial issuesGives advice re tests or examsStates objectives of lectureSuggests supplementary readingsAdvises as to preparation for papersSuggests mnemonic aids

Addresses students by nameSensitive to student feelingsTolerant of other points of viewAvailable for consultationFlexible regarding requirementsOffers to help studentsEncourages questions or commentsAsks questions of studentsEncourages student participationAsks if students understandExpects students to answer questionsEncourages independent thinkingIncorporates student ideas into lecture

Reliabilityof

mean rating"

.61

.59

.68

.57

.71

.63

.74

.63

.65

.60

.68

.51

.64

.86

.75

.77

.60

.76

.52

.55

.70

.58

.56

.71

.56

.60

.51

.58

.75

.53

.65

.51

.50

.81

.63

.65

.77

.57

.50

.65

.80

.59

.60

.68

.66

.61

.64

.61

.60

Correlation withstudent ratings

.29*

.00

.27*-.45*

.17-.15

.31*

.57*-.33*-.57*-.38*

.41*

-.42*-.16

.14

.36*-.01

.17

.66*-.53*

.33*

.30*

.05

.39*

.54*

.66*

.14-.17

.04

.06

.19

.34*

.38*

.20

.18

.26

.01

.56*

.53*

.43*

.11

.14

.17-.12

.35*

.26-.05

.42*

.43*a Based on average of 6.41 ratera.*p < .05.


more reliable measures of these two classesof variables. Second, a much smaller ratioof independent variables to sample size wasrequired to avoid unstable results in pathanalyses (see Kerlinger & Pedhazur, 1973).Finally, multicollinearity (i.e., intercorrela-tions above .80) within both personality andclassroom behavior domains would havemade the interpretation of regression andpath analyses extremely difficult. Conse-quently, preliminary factor analyses wereundertaken to identify underlying person-ality and classroom behavior dimensions ofthe effective university instructor.

Mean peer ratings of the 21 personalitytraits that correlated significantly with stu-dent evaluations of teaching were subjectedto a principal-components factor analyses.Cattell's (1966) scree test of eigenvaluesyielded a two-factor solution. The first twofactors, accounting for 66.1% of the totalvariance in mean ratings, were rotated to avarimax criterion to aid in the interpretationof underlying personality dimensions. Asshown in Table 3,14 of the 21 traits loadedhigher than .55 on one of the two rotatedfactors and lower than .40 on the other.Inspection of the factor loadings led to theinterpretation of Factor 1 as an Achieve-ment-Orientation dimension, and Factor 2as an Interpersonal-Orientation dimension.A professor who scored high on Factor 1would be characterized by dominance,ambition, leadership, intelligence, and en-durance. A professor who scored high onFactor 2 would be characterized as nonau-thoritarian, supportive, nondefensive, fun-loving, objective, and aesthetically sensitive.Professors were assigned factor scores on thetwo underlying personality dimensions byaggregating mean ratings for the personalitytraits loading higher than .55 on that di-mension (and lower than .40 on the other).These criteria for inclusion were used tominimize the correlation between scores onthe two personality composites. This stepwas necessary because highly correlatedscores would restrict interpretation of therelative contribution of the two personalitycomposites to the prediction of student in-structional ratings and classroom behaviordimensions. Alpha coefficients for theAchievement Orientation and Interpersonal

Table 3Results of Principal-Components FactorAnalysis of the 21 Predictive PersonalityTraits, Rotated to a Two-Factor VarimaxSolution

Loadings

Personalitytrait

DominantAmbitiousShows leadershipIntelligentEnduringIntellectually curiousAttention-seekingHarm-avoiding

AuthoritarianSupportingDefensiveFun-lovingObjectiveAesthetically sensitive

SociableEx travertedChangeableLiberalApproval-seekingOrderlyAnxious

Tactor Factor1 2

.93

.92

.87

.86

.84

.80

.77-.59

.02

-.15.03.30.11

-.01.21.32

.05

.06

.33

.11

.10

-.25

-.90.85

-.79.77.72.66

.44 .79

.57 .64

.59 .68

.46 .84

.44 .42

.31 -.46-.25 -.48

Note. Boxes indicate traits defining a given factor.

Orientation composites were .94 and .89,respectively, indicating considerable unidi-mensionality in each case. Professors*scores on the Achievement Orientation andInterpersonal Orientation composites werefound to correlate .27 with each other and .51and .45, respectively, with student ratings ofteaching.

Mean observer ratings of the 26 predictiveclassroom behaviors also were subjected toa principal-components factor analysis.Again, a scree test yielded a two-factor so-lution. The first two factors, accounting for50.5% of the total variance in mean ratings,were rotated to a varimax criterion. Asshown in Table 4, 21 of the 26 classroombehaviors had loadings greater than .55 onone of the two rotated factors, and no vari-able loaded on more than one factor. In-spection of the factor loadings led to the in-terpretation of Factor 1 as a Charisma di-mension and Factor 2 as an Organization


dimension. Professors who scored high onFactor 1 exhibited classroom behaviors suchas speaking expressively, smiling or laughingwhile teaching, showing interest in thesubject matter, and encouraging studentparticipation. Professors who scored highon Factor 2 engaged in such classroom be-haviors as giving a preliminary overview ofthe lecture, organizing the lecture withheadings or points, and giving multiple ex-amples of concepts. Professors were as-signed factor scores on each of these di-mensions by aggregating mean ratings ofclassroom behaviors loading .55 or higher ona given factor. Alpha coefficients for theCharisma and Organization composites were.93 and .88, respectively, indicating consid-erable unidimensionality in each of theclassroom behavior composites. Factorscores on the Chariama and Organization

dimensions correlated .45 with each otherand .63 and .53, respectively, with studentratings of teaching.

Relations Among Personality Composites,Classroom Behavior Composites, andStudent Ratings of TeachingEffectiveness

Zero-order correlations were computedbetween personality composites and class-room behavior composites. The Achieve-ment Orientation personality composite waspositively related to the Charisma classroombehavior composite (r — .42) but unrelatedto the Organization composite (r = .13).The Interpersonal Orientation personalitycomposite was positively related to both theCharisma (r = .42) and the Organization (r= .45) classroom behavior composite. These

Table 4Results of Principal-Components Factor Analysis of the 26 Predictive Classroom Behaviors,Rotated to a Two-Factor Varimax Solution

Loadings

Classroombehavior

Factor1

Factor2

Smiles or laughs while teachingTells jokes or humorous anecdotesRelates material to student interestsShows interest in subject matterExhibits facial gesturesSpeaks in an expressive waySpeaks in a monotoneGives advice re tests or examsSensitive to student feelingsEncourages independent thinkingAvailable for consultationEncourages student participationAvoids eye contactIncorporates student ideas into lecture

Gives preliminary overview of lectureOrganizes lecture with pointsLecture follows logical sequenceUses headings to organize lectureStates objectives of lectureCovers little material in lectureGives multiple examples of concepts

Gestures with hands or armsStutters, mumbles, or slurs wordsDoes not speak clearlyFails to define new termsTolerant of other points of view

.81

.77

.74

.69

.67-.67.67.66.64.59.59-.57.55

-.05.05.33.23-.03.42-.48.00.32.21.26.09-.28.24

-.02.01.33.16.13-.08.43

.24-.49-.44-.53.30

.86

.80

.78

.74

.72

.69

.59

.26

.11•.20-.47.51

Note. Boxes indicate traits defining a given factor.


InterpersonalOrientation

.32* .31* . 10 .30*

Student ratings ofteaching effectiveness

Figure 1. Path analysis of composite personality dimensions and classroom behavior dimensions inrelation to student ratings of teaching effectiveness.

results show that there are significant rela-tions between teacher personality and pat-terns of classroom behavior.

To test the proposed causal model thatclassroom behavior mediates the relationbetween personality and teaching effec-tiveness, a path analysis was undertakenusing the personality and the classroom be-havior composites as predictors and meanstudent ratings of teaching as the criterion.The results of this analysis are presented inFigure 1. The path coefficients are stan-dardized beta weights derived from a seriesof multiple regression analyses (see Ker-linger & Pedhazur, 1973). Overall, the fourpredictors in the analysis accounted for 57%of the variance in student ratings of teachingeffectiveness, R (4, 32) = .75, p < .001.

As explained below, 34% of the relationbetween the Achievement Orientation per-sonality composite and perceived teachingeffectiveness was found to be mediated bythe two classroom behavior composites.Twenty-seven percent of the relation wasmediated by Charisma, and 7% was me-diated by Organization.

The percentage of the relation betweenAchievement Orientation and student rat-ings that was mediated by Charisma (27%)was the sum of two components. Twenty-

one percent of the relation was directlymediated by Charisma. This percentagewas obtained by multiplying the path coef-ficient between Achievement Orientationand Charisma (.33) by the path coefficientbetween Charisma and sttfdent ratings ofteaching (.32) and dividing this product bythe correlation between Achievement Or-ientation and student ratings (.51). Theadditional 6% of the effect mediated byCharisma was due to the correlation betweenAchievement Orientation and InterpersonalOrientation. This percentage is obtained bymultiplying the correlation between the twopersonality composites (.27) by the pathcoefficient between Interpersonal Orienta-tion and Charisma (.33) and the path coef-ficient between Charisma and student rat-ings of teaching (.32) and dividing thisproduct by the correlation betweenAchievement Orientation and the criterion(.51).

Only 7% of the relation between Achieve-ment Orientation and student ratings ofteaching effectiveness was mediated by Or-ganization. Through procedures identicalto those described above, it can be seen thatless than 1% of this relation was directlymediated by the classroom behavior com-posite, whereas the remainder was mediated


indirectly by the correlation betweenAchievement Orientation and InterpersonalOrientation.

Sixty-six percent of the relation betweenAchievement Orientation and the criterionof teaching effectiveness was found to beunmediated by the two classroom behaviorcomposites. Sixty-one percent of the rela-tion was a direct effect of AchievementOrientation on student ratings, whereas theadditional 5% was a direct effect due to thecorrelation between the two personalitycomposites. The direct effect percentagewas obtained by dividing the direct pathcoefficient (.31) by the correlation betweenAchievement Orientation and the criterion(.51). The percentage of the relation thatwas a direct effect due to the correlationbetween the personality composites wasobtained by multiplying the correlation be-tween the two personality composites (.27)by the path coefficient between Interper-sonal Orientation and the criterion (.10) anddividing by the correlation betweenAchievement Orientation and the criterion(.51).

Fifty-nine percent of the relation betweenthe Interpersonal Orientation personalitycomposite and rated teaching effectivenesswas mediated by the two classroom behaviorcomposites. Twenty-nine percent of therelation was mediated by Charisma, and 30%was mediated by Organization. These per-centages were obtained in the same fashionas were those for Achievement Orientation.Twenty-three percent of the relation be-tween Interpersonal Orientation and studentratings was directly mediated by Charisma(.33 X .32/.45), whereas 6% was mediated byCharisma because of the correlation betweenInterpersonal Orientation and AchievementOrientation (.27 X .33 X .32/.45). Thirtypercent of the relation between Interper-sonal Orientation and the criterion ofteaching effectiveness was directly mediatedby Organization (.45 X .30/.45); virtuallynone of the relation was indirectly mediatedby Organization through Achievement Or-ientation (.27 X .01 X .30/.45).

Forty-one percent of the relation betweenInterpersonal Orientation and student rat-ings was unmediated by the classroom be-havior composites. Twenty-two percent of

the relation was a direct effect of Interper-sonal Orientation (.10/.45), and the re-maining 19% was a direct effect due to thecorrelation between Interpersonal Orienta-tion and Achievement Orientation (.27 X.31/.45).

In sum, 59% of the relation between In-terpersonal Orientation and student ratingsof teaching was mediated by Charisma (29%)and Organization (30%), and 34% of therelation between Achievement Orientationand student ratings was mediated by Cha-risma (27%) and Organization (7%). Overall,therefore, 46.5% of the relation between thetwo personality composites and studentratings was mediated by the Charisma andOrganization classroom behavior compos-ites, 28% and 18.5%, respectively. It seemsthat an instructor who scored high onAchievement Orientation received highstudent ratings of teaching partly because heor she engaged in a high frequency of class-Table 5Correlations Between 21 PredictivePersonality Traits and Two ClassroomBehavior Dimensions

Personality trait

DominantAmbitiousShows leadershipIntelligentEnduringIntellectually curiousAttention-seekingHarm-avoiding

AuthoritarianSupportingDefensiveFun-lovingObjectiveAesthetically sensitive

SociableExtravertedChangeableLiberalApproval-seekingOrderlyAnxious

Classroom behaviordimension

Charisma

.32*

.26

.48*

.29*

.27*

.27*

.58*-.25

-.28*.32*

-.19.47*.30*.44*

.48*

.56*

.48*

.48*

.40*

.06-.23

Organization

-.02.03.31*.11.14.19.12.08

-.32*.54*

-.36*.23.56*.14

.36*

.24

.19

.37*

.27*

.33*-.11

* p < .05.


Personalitytrait

Student ratings ofteaching e-f-fectiveness

Figure 2. General form of causal model relating individual personality traits to student ratings ofteaching effectiveness via the two classroom behavior composites.

room behaviors described as charismatic,whereas one who scored high on Interper-sonal Orientation received high studentratings partly because he or she engaged ina high frequency of both charismatic andorganizational classroom behaviors.

Analysis of Individual Personality Traits

Individual personality traits were ana-lyzed to provide a more detailed interpre-tation of the relations identified above aswell as to examine traits excluded from thepath analysis, such as extraversion, orderli-ness, and anxiety. To further demonstratethat personality traits and classroom be-haviors associated with teaching effective-ness are themselves related, zero-order cor-relations between the 21 predictive person-ality traits and the Charisma and Organi-zation classroom behavior composites werecomputed. As can be seen in Table 5,25 of42 correlations were found to be significantat the .05 level, with coefficients ranging ashigh as .58. Charismatic classroom behaviorwas most strongly related to the personalitytraits of attention-seeking (.58), extraversion(.56), sociability (.48), leadership (.48), li-beralness (.48), changeability (.48), andfun-lovingness (.47). Organizational be-

havior, on the other hand, was most stronglyrelated to the traits of objectivity (.56),supportiveness (.54), liberalness (.37), de-fensiveness (-.36), sociability (.36), order-liness (.33), and authoritarianism (-.32).

Further path analyses were undertaken totest the causal model that classroom teach-ing behavior mediates the relation betweenspecific personality traits and teaching ef-fectiveness. In each of 21 separate pathanalyses, the predictor variables consistedof the two classroom behavior compositesplus one of the 21 personality traits knownto predict teaching effectiveness. Meanstudent ratings of teaching served as thecriterion in all analyses. The analyses wereconducted individually to provide an ac-ceptable ratio of independent variables tothe sample size. The general form of thecausal model for these analyses is depictedin Figure 2.

Table 6 indicates the proportion of therelation between each of the personalitytraits and student ratings of teaching thatwas mediated by the Charisma and Organi-zation classroom behavior composites.Overall, 50% of the relation between per-sonality traits and effectiveness ratings wasmediated by classroom behavior, withCharisma (34%) accounting for approxi-


Table 6Estimated Direct and Mediated Effects of Personality on Student Ratings of TeachingEffectiveness in %

Personalitytrait

AmbitiousSociableChangeableDefensiveDominantEnduringAttention-seeking

Harm-avoidingSupportingOrderlyFun-lovingAesthetically sensitiveApproval -seekingIntellectually curious

AnxiousIntelligentLiberalShows leadershipObjectiveAuthoritarianBxtraverted

Average

Direct

63464637666456

85357831291965

5856495628314550

Charisma-mediated

32364030342635

15350551575724

3134332936413934

Effect

Organization-mediated

05181433001009

00301718142411

1110181536281616

Totalmediated

37545463343644

15652269718135

4244514472695550

mately twice as much mediation as did Or-ganization (16%). The direct effect of eachpersonality trait on student ratings was es-timated by dividing the path coefficientbetween the personality trait and studentratings by the correlation between thesesame two variables. The effects mediatedby Charisma and Organization were calcu-lated by multiplying the path coefficientbetween Charisma [Organization] and thecriterion by the correlation between Cha-risma [Organization] and the personalitytrait and dividing this product by the cor-relation between the personality trait andstudent ratings. The obtained results fur-ther support the view that approximately50% of the relation between personality andteaching effectiveness is mediated by specificclassroom behaviors.

Discussion

The teacher personality traits and class-room teaching behaviors identified in the

present study as significant correlates ofstudent ratings of college instruction are ingeneral agreement with results of previousresearch. The highly rated teacher wasfound to exhibit two types of personalitytraits: one type reflecting AchievementOrientation (e.g., dominance, intelligence,leadership) and the other reflecting Inter-personal Orientation (e.g., supportiveness,nonauthoritarianism, nondefensiveness).Similar personality traits were found to beassociated with perceived teaching effec-tiveness in studies reported by Costin andGrush (1973); Murray (1975,1978); Rushton,Murray, and Paunonen (1983); Sherman andBlackburn (1975); and Tomasco (1980).The highly rated teacher also was demon-strated to engage in two general types ofclassroom teaching behavior: one type re-flecting Charisma (e.g., use of humor, en-couraging participation) and the other re-flecting Organization (e.g., using headings,stating objectives). Again, these resultsgenerally replicate the findings of previous


investigations (e.g., Cranton & Hillgartner,1981; Frey, 1978; Mintzes, 1979; and Murray,1983).

More important, the present findingssupport the hypothesis that personalitytraits and classroom behaviors are them-selves related. The Interpersonal Orienta-tion personality composite was found to bepositively related to both the Charisma andOrganization classroom behavior compos-ites, and the Achievement Orientation per-sonality composite was found to correlatepositively with the Charisma composite.Additional evidence was provided by the factthat 25 of 42 simple correlations betweenindividual personality traits and classroombehavior composites were significant, withcoefficients ranging as high as .58. In-structors who engaged in a high frequency ofcharismatic classroom behaviors were ratedby their colleagues as attention-seeking,extraverted, sociable, showing leadership,liberal, changeable, fun-loving, aestheticallysensitive, approval-seeking, supporting,dominant, objective, intelligent, nonautho-ritarian, enduring, and intellectually curious.On the other hand, instructors who exhibiteda high frequency of organizational classroombehaviors were perceived by faculty peers asobjective, supporting, liberal, sociable,nondefensive, orderly, nonauthoritarian,showing leadership, and approval-seeking.

The proposed causal model in which per-sonality influences teaching effectiveness viaclassroom behavior was also supported.Path analysis revealed that 46.5% of therelation between Achievement Orientationand Interpersonal Orientation on the onehand and rated teaching effectiveness on theother was mediated by the two classroombehavior dimensions. Fifty-nine percent ofthe relation between Interpersonal Orien-tation and teaching effectiveness was me-diated by classroom behavior, with Organi-zation and Charisma mediating approxi-mately equal amounts of this relation. Itappears, therefore, that teachers character-ized by high levels of Interpersonal Orien-tation engaged in a high frequency of class-room behaviors reflecting Charisma andOrganization and these behaviors in turnled to positive end-of-term evaluationsfrom students. Thirty-four percent of therelation between Achievement Orientation

and teaching effectiveness was shown to bemediated by classroom behavior. This ef-fect was mediated primarily through Cha-risma, with Organization contributing to therelation only because of overlapping variancebetween the two personality composites.This suggests that teachers characterized byhigh levels of Achievement Orientation werepositively evaluated by students partly be-cause they exhibited classroom behaviorsreflecting Charisma. Organizational be-haviors, on the other hand, seemed to playvirtually no role in mediating this relation.Athough instructors high in AchievementOrientation might have been expected toengage in a high frequency of Organizationalclassroom behaviors, it may be that theachievement-oriented instructors who par-ticipated in this study devoted more time toresearch than to preparation for teaching.

The path analyses undertaken to examinerelations between individual personalitytraits and student ratings of teaching alsoprovided evidence for mediation by class-room behavior. Overall, 50% of the sharedvariance in ratings of personality andteaching effectiveness was accounted for bycharismatic and organizational classroombehaviors. Moreover, the pattern of me-diation supported intuitive expectations.For example, the relation between the per-sonality trait of orderliness and teachingeffectiveness was mediated primarily by theOrganization composite, whereas the rela-tion between attention-seeking and teachingeffectiveness was mediated primarily by theCharisma composite.

Several limitations of this research shouldbe kept in mind. First, the small sample sizerestricted the complexity of causal modelstested. Thus, the exact interpretation ofwhich specific teaching behaviors mediatedwhich specific personality traits was notpossible. Second, the present findings maynot generalize to methods other than thelecture method or to teaching content areasother than psychology. Finally, the corre-lational design of this research does not allowthe direct interpretation of causality. It ispossible that the causal model proposedhere, which assumes that teacher personalityinfluences student ratings of teaching ef-fectiveness via classroom teaching behavior,is inaccurate. The plausibility of alternative

406 S. ERDLE, H. MURRAY, AND J. P, RUSHTON

models, however, is reduced by theoreticalconsiderations reviewed earlier, by the nor-mal temporal precedence of personality toclassroom behavior to student evaluationsof teacher effectiveness, and by previousempirical evidence indicating a causal rela-tion between classroom behaviors and stu-dent ratings (e.g., Murray & Lawrence,1980).

In spite of possible drawbacks, this re-search has important implications for boththe validity of student instructional ratingsand the understanding of teaching at theuniversity level. The finding that studentratings of overall effectiveness are relatedpredictably to independently assessed per-sonality traits and classroom behaviors ofcollege instructors provides some evidencethat student ratings converge with theoret-ically relevant variables. Several research-ers (e.g., Small, Hollenbeck, & Haley, 1982)have claimed that student ratings are invalidbecause they are substantially influenced bythe personality of the instructor. Contraryto this claim, the present study suggests thatinstructor personality is reflected in specificclassroom teaching behaviors, which in turnare validly rated by students. In more gen-eral terms, the present research providesevidence supporting the pattern of relationsproposed in Dunkin and Biddle's (1974)model of classroom teaching. It seems"presage" variables (teacher personalitytraits) lead to "process" variables (classroombehaviors), which in turn determine "prod-uct" variables (student evaluations ofteaching). Given the present results, future re-search might test more complex models of col-lege teaching effectiveness. Studies of thiskind will further both the understanding ofand the assessment of college instruction.

References

Allport, G. W. (1937). Personality: A psychologicalinterpretation. New York: Holt, Rinehart &Winston.

Bendig, A. W. (1955). Ability and personality char-acteristics of introductory psychology professorsrated as competent and empathetic by the students.Journal of Educational Research, 48,705-709.

Cattell, R. B. (1966). The scree test for the numberof factors. Multivariate Behavioral Research, 1,245-276.

Cohen, P. A. (1981). Student ratings of instruction

and student achievement: A meta-analysis of mul-tisection validity studies. Review of EducationalResearch, 51,281-309.

Costin, F., & Grush, J. E. (1973). Personality corre-lates of teacher-student behavior in the collegeclassroom. Journal of Educational Psychology, 65,35-44.

Cranton, P. A., & Hillgartner, W. (1981). The rela-tionship between student ratings and instructor be-havior: Implications for improved teaching. Ca-nadian Journal of Higher Education, 11,73-81.

Deshpande, A. S., Webb, S. C, & Marks, E. (1970).Student perceptions of engineering instructor be-haviors and their relationship to the evaluation ofinstructors and courses. American EducationalResearch Journal, 7,289-305.

Dunkin, M. J., & Biddle, B. J. (1974). The study ofteaching. New York: Holt, Rinehart & Winston.

Eysenck, H. J., & Eysenck, S. B. G. (1975). Manualof the Eysenck Personality Questionnaire. SanDiego, CA: Educational and Industrial TestingService.

Frey, P. W. (1978). A two-dimensional analysis ofstudent ratings of instruction. Research in HigherEducation, 5,69-91.

Guilford, J. P. (1959). Personality. New York:McGraw-Hill.

Jackaon, D, N. (1974). Personality Research FormManual. Goshen, NY: Research PsychologistsPress.

Jackson, D. N., & Paunonen, S. V. (1980). Personalitystructure and assessment. In M. R. Rosenzweig &L. M. Porter (Eds.), Annual Review of Psychology,(Vol. 31, pp. 503-551). Palo Alto, CA: Annual Re-views.

Keaveny, J. J., & McGann, A. F. (1978). Behavioraldimensions associated with students' global ratingsof college professors. Research in Higher Education,9, 333-345.

Kerlinger, F. N.( & Pedhazur, E. J. (1973). Multipleregression in behavioral research. New York: Holt,Rinehart & Winston.

Marsh, H. W. (1984), Students' evaluations of uni-versity teaching: Dimensionality, reliability, validity,potential biases, and utility. Journal of EducationalPsychology, 76,707-754.

McKeachie, W. J. (1979). Student ratings of faculty:A reprise. Academe, 62,384-397.

Mintzes, J. J. (1979). Overt teaching behaviors andstudent ratings of instructors. Journal of Experi-mental Education, 48,145-153.

Murray, H. A. (1938). Explorations in personality.Cambridge, MA: Harvard University Press.

Murray, H. G. (1975). Predicting student ratings ofcollege teaching from peer ratings of personalitytraitB. Teaching of Psychology, 2,66-69.

Murray, H. G. (1978). Teacher ratings, studentachievement, and teacher personality traits. Paperpresented at the annual meeting of the CanadianPsychological Association.

Murray, H. G. (1980). Evaluating university teach-ing: A review of research. Toronto, Canada: On-tario Confederation of University Faculty Associa-tions.

Murray, H. G. (1983). Low-inference classroomteaching behaviors and student ratings of college


teaching effectiveness. Journal of EducationalPsychology, 75,13&-149.

Murray, H. G., & Lawrence, C. (1980). Speech anddrama training for lecturers as a means of improvinguniversity teaching. Research in Higher Education,13,73-90.

Rushton, J. P., Murray, H. G., & Paunonen, S. V.(1983). Personality, research creativity, and teachingeffectiveness in university professors. Sciento-metrics, 5, 93-116.

Seldin. P. (1980), Successful faculty evaluationprograms. Crugers, NY: Coventry Press.

Sherman, B. R., & Blackburn, R. T. (1975). Personalcharacteristics and teaching effectiveness of collegefaculty. Journal of Educational Psychology, 67,124-131.

Small, A. C, Hollenbeck, A. R., & Haley, R. L. (1982).

The effect of emotional state on student ratings ofinstructors. Teaching of Psychology, 9, 205-211.

Solomon, D. (1966). Teacher behavior dimensions,course characteristics, and student evaluations ofteachers. American Educational Research Journal,3, 35-47.

Sorey, K. E. (1968). A study of the distinguishingpersonality characteristics of college faculty who aresuperior in regard to the teaching function. Disser-tation Abstracts, 28-12(A), 4916.

Tomasco, A. T. (1980). Student perceptions of in-structional and personality characteristics of faculty:A canonical analysis. Teaching of Psychology, 7,79-82,

Received November 20,1984Revision received April 3,1985 •

Psychological Documents to Be Discontinued

At its February 2-3,1985, meeting, the Council of Representatives voted to ceasepublication of Psychological Documents (formerly the Journal Supplement Ab-stract Service) as of December 31, 1985, with the publication of the December1985 issue of the catalog. Continued low submissions, decreasing usage, andrising costs for fulfillment of paper and microfiche copies of documents werereasons given for discontinuing publication of the alternative format publication,which was begun in 1971 as an "experimental" publication.

Authors who wished to submit documents for publication consideration in 1985were required to do so by July 1. Authors revising documents were required tocomplete all revisions and submit them for final review no later than July 1.Fulfillment of orders for paper and microfiche copies of documents presently inthe system and of those documents entered during 1985 will continue throughDecember 31,1986.