Student Personality Differences Are Related to Their Responses on Instructor Evaluation Forms

The relation of student personality to student evaluations of teaching (SETs) was determined in a sample of 144 undergraduates. Student Big Five personality variables and core self-evaluation (CSE) were assessed. Students rated their most preferred instructor (MPI) and least preferred instructor (LPI) on 11 common evaluation items. Pearson and partial correlations simultaneously controlling for six demographic variables, Extraversion, Conscientiousness and Openness showed that SETs were positively related to Agreeableness and CSE and negatively related to Neuroticism, supporting the three hypotheses of study. Each of these significant relations was maintained when MPI, LPI or a composite of MPI and LPI served as the SET criterion. For example, the MPI-LPI composite correlated.28 with Agreeableness,.35 with CSE and –.28 with Neuroticism. Similar correlations resulted for MPI and LPI. Hierarchical multiple regression demonstrated that the CSE was an independent predictor of MPI ratings, Agreeableness was an independent predictor of LPI ratings, and both the CSE and Agreeableness were independent predictors of MPI-LPI composite ratings. Neuroticism did not emerge as an independent predictor because of the substantial correlation between CSE and Neuroticism (r =.53) and because CSE had greater predictive capacity. This is the first study to incorporate the CSE construct into the SET literature.

Keywords: student evaluation of teaching; core self-evaluations; Agreeableness; neuroticism; Big Five

Despite persistent controversy, virtually all universities and colleges in the USA use the aggregated judgements of students to estimate the relative teaching effectiveness of instructors (e.g. Clayson [12]; Clayson and Sheffet [13]). Such student evaluations of teaching (SETs) routinely serve as prime indicators of teaching quality in deliberations regarding tenure, promotion and merit awards (McPherson, Jewell, and Myungsup [29]). Similar practices also exist or are developing in countries in other parts of the world, such as Australia, Cambodia, China, Italy, Japan, Malaysia, Romania, Sweden, Turkey and the UK (e.g. Braga, Paccagnella, and Pellizzari [6]; Bystrom [7]; Chen and Watkins [11]; Liaw and Goh [27]; Nash [32]; Raboca and Solomon [34]; Shah and Nair [37]; Stapleton [38]; Stringer and Irwing [39]; Ustunluoglu and Can [40]). Nevertheless, the validity of the SET approach has been broadly criticised on several grounds for decades.

Various sources of bias in SETs have been reported. For example, higher instructor evaluations have been found to be associated with elevated student grade expectations (e.g. McPherson, Jewell, and Myungsup [29]), male instructors (e.g. McPherson et al.), younger instructors (e.g. McPherson et al.), female students (e.g. Bonitz [3]), higher level courses (e.g. Goldberg and Callahan [18]), elective courses (e.g. Scherr and Scherr [35]) and smaller classes (e.g. Bedard and Kuhn [2]). In addition, instructor evaluations have been found to be related to course workload or difficulty (e.g. Scherr and Scherrr), prior interest in the subject matter (e.g. Scherr and Scherrr), different disciplines (e.g. Basow and Montgomery [1]), cross-cultural response styles (Dolnicar and Grun [15]) and student perceptions of the instructor personalities (Clayson and Sheffet [13]).

Nevertheless, one largely neglected area of inquiry in regard to potential biassing factors involved in SETs is the personality of the student rater. For example, Bonitz ([3]), in her doctoral dissertation, could locate only two studies that dealt with the relation of student personality to SET ratings. However, rather than focusing on conventional personality traits, one focused on the student social styles (Schlee [36]) and the other used a trait measure of mood (Munz and Munz [31]). An earlier study also showed student needs for achievement to be related to instructor evaluation (Kovacs and Kapel [26]).

Although, there is a paucity of research on the student personality in this context, most of the promising traits appear to be factors in the Big Five personality approach, the most widely accepted model of personality (Costa and McCrae [14]; Goldberg [16]; John and Srivastava [20]). The five factors in the model are Openness, Conscientiousness, Extraversion, Agreeableness and Neuroticism. It is widely assumed and supported by empirical evidence that these five factors or fundamental traits represent the most comprehensive yet economical way to describe the similarities and differences between people.

These five basic personality traits can be described in the following ways (Costa and McCrae [14]; Goldberg [17]): persons high on Openness tend to exhibit higher levels of imagination, willingness to experiment, intellectual curiosity, aesthetic interest, tolerance for diversity and emotional depth; they tend to be intelligent, perceptive, analytical, reflective, creative, cultured, refined and sophisticated. Persons high on Conscientiousness tend to manifest higher levels of dutifulness, order, self-discipline, deliberation, competence and achievement striving; they tend to be organised, responsible, reliable, practical, thorough, hardworking, thrifty, cautious and serious. Persons high on Extraversion tend to demonstrate higher levels of warmth, positive emotions, gregariousness, activity, assertiveness and excitement seeking; they tend to be talkative, enthusiastic, bold, spontaneous, adventurous and sociable. Persons high on Agreeableness tend to display higher levels of compliance, trust, straightforwardness, altruism, modesty and tender-mindedness; they tend to be warm, kind, cooperative, unselfish, polite, generous, flexible and fair. Persons high on Neuroticism tend to show higher levels of vulnerability, self-consciousness, anxiety, impulsiveness, angry hostility and depression; they tend to be tense, nervous, envious, unstable, discontented, insecure, emotional, guilt-ridden and moody.

For the present research, only four studies could be located that have provided data on the relation of levels of the Big Five in the students to their evaluations of instructors. Of these, one was reported in a journal article (Patrick [33]), one in a single-page summary of a paper presentation (Calton et al. [8]), one in a doctoral dissertation (Bonitz [3]) and one in a master's thesis (Chan [9]). Patrick found that higher Agreeableness was associated with more positive evaluations of the teaching ability of instructors in a sample of 176 education students at a Kansas university. Calton et al. assessed the Big Five in a sample of 200 students at the University of Nottingham. They found that Neuroticism was negatively related, while Conscientiousness and Extraversion were positively related, to instructor evaluation scores on two criteria. One criterion dealt with eight general teaching performance items, and the other dealt with items pertaining to the specific instructor behaviours. In the Bonitz dissertation, 610 college students rated one of four hypothetical professors, who were briefly described in four short vignettes. Male and female students were higher were on Agreeableness, Conscientiousness and Openness gave more positive ratings. For males, Neuroticism was negatively correlated with instructor ratings. In the Chan thesis, the Agreeableness and Openness of 588 students correlated positively with their ratings of the teaching performance of their professors.

Overall, the results of these four studies were rather inconsistent. There is evidence that the higher SETs were associated with higher Agreeableness in three, lower Neuroticism in two, higher Conscientiousness in two, higher Openness in two and the higher Extraversion in one.

In tangentially related research, Neuroticism and Agreeableness also figured prominently in regard to satisfaction with one's major and with college life. Bradford ([5]), in a sample of psychology, engineering and education majors, found that students lower in Neuroticism and higher in Agreeableness were more satisfied with their major in each of these three disciplines. However, among those in psychology and engineering, students high in Conscientiousness and Openness also were more satisfied with their major. In another study, Lounsbury et al.([28]) found that lower Neuroticism and higher Agreeableness, Conscientiousness and Extraversion were associated with greater satisfaction with college.

Another promising trait potentially related to student ratings of instructors is the core self-evaluation (CSE) construct put forth by Judge, Locke, and Durham ([24]). Generally, thinking positively of oneself and having confidence in one's own abilities characterise a person with a high CSE. According to Judge et al. CSE is reflected in dispositional levels of four traits well established in the psychological literature: Neuroticism, self-esteem, generalised self-efficacy and locus of control. A person high on the CSE dimension is low on neuroticism, high on self-esteem and self-efficacy and has an internal rather than an external locus of control. According to Judge et al. ([22]), these four traits can be described briefly in the following manner. Neuroticism is 'the tendency to have a negativistic cognitive/explanatory style and to focus on negative aspects of the self', self-esteem is 'the overall value that one places on oneself as a person', generalised self-efficacy is 'an evaluation of how well one can perform across a variety of situations', and locus of control refers to 'beliefs about the causes of events in one's life – locus is internal when individuals see events as being contingent on their own behaviour' (303–304).

The CSE approach was developed by Judge et al. ([24]) within the context of dispositional determinants of job satisfaction. Much research over the years has supported the initial contention that higher CSEs are associated with greater job satisfaction (e.g. Chang et al. [10]; Judge and Bono [21]). In regard to the present SET context, no study has included CSE as a predictor of student ratings of instructors. However, higher CSE scores were associated with greater satisfaction with university life (Miller and Nicols [30]) in somewhat tangentially related work.

Given their potential biassing effects and the lack of existing literature, there is a clear need for further research on the relation of student personality to SETs. The four studies implicating the Big Five have inherent limitations, and the results are somewhat inconsistent. The only conventional peer reviewed article is that by Patrick ([33]), but the major focus of that contribution was on how student ratings of instructor personality and expected grades are related to student course and instructor evaluations, rather than on the relation of student personality to SETs. The Bonitz ([3]) dissertation is based on an experiment conducted through a commercial online service, in which students were randomly assigned to one of four 110-word vignettes that briefly described professors with exactly the same characteristics, except that they were either men or women and taught either psychology or statistics. However, there was no opportunity for students to interact with and perceive a real professor, and actual SETs importantly involve such preceding exposure. The brief account of the conference presentation by Calton et al. ([8]) is short on detail and only descriptively reports bivariate correlations between SETs and three of the Big Five. Only Chan's ([9]) graduate thesis abstract is available online, and only one sentence provides descriptive information about the correlations between SETs and two Big Five variables. The thrust is on the personality of the instructor. In regard to student CSE, the possible relation to SETs has not received any attention.

The present study was conducted to determine whether the Big Five and CSE personality variables relate to SET at the university level. It seems reasonable to make certain predictions in regard to Agreeableness, Neuroticism and CSE. Three of the four previous studies (i.e. Bonitz [3]; Chan [9]; Patrick [33]) found that higher Agreeableness was linked to more positive SETs. As assessed with the Big Five Inventory (BFI: e.g. John, Naumann and Soto [19]), a person high on Agreeableness is kind, considerate, forgiving, helpful, trusting, cooperative and unselfish and is not a person who tends to find fault with others, start quarrels, be aloof and cold, and be rude. One would expect a student with these qualities to be more likely to produce a more favourable evaluation of an instructor than a student with the opposite qualities. Therefore, one hypothesis is the following:

Hypothesis 1: Higher student Agreeableness is associated with higher SETs.

Two of the previous studies (i.e. Bonitz [3]; Calton et al. [8]) found some evidence that higher Neuroticism is associated with lower SETs. A person high on Neuroticism on the BFI tends to be moody, emotionally unstable, tense, not relaxed, depressed, blue, a worrier who gets nervous and upset easily, does not handle stress well and does not remain calm in tense situations. It seems reasonable to expect a student with these limitations in an academic setting to be more frustrated and likely to produce a less favourable evaluation of an instructor than a student with the opposite characteristics. Therefore, a second hypothesis is the following:

Hypothesis 2: Higher student Neuroticism is associated with lower SETs.

Although two previous studies (i.e. Bonitz [3]; Calton et al. [8]) found that Conscientiousness was positively related to SETs, two (i.e. Bonitz [3]; Chan [9]) found that Openness was positively related to SETs, and one (Calton et al.) found that Extraversion was positively related to SETs, hypotheses were not put forward in regard to these three Big Five variables. According to the depictions on the BFI, there seems little inherent justification to expect that Conscientiousness, Openness or Extraversion should have a general linear relation to SETs.

In contrast, it seems quite probable that higher CSE is associated with higher SETs. As stated before, Judge et al. ([24]) developed the CSE approach with the belief that it should predict work satisfaction. Research indeed has shown that the CSE construct predicts work satisfaction (e.g. Judge and Bono [21]), but it also predicts life satisfaction (e.g. Chang et al. [10]), and even satisfaction with university life (Miller and Nicols [30]). According to the core self-evaluations scale (CSES) constructed by Judge et al. ([22]), persons high on CSE tend to believe that they succeed when they try, they are successful in completing tasks, that they achieve the success they deserve, that they determine what happens in their lives, that they are able to cope with the most of their problems and that they are satisfied with themselves. Persons low on CSE are much more likely to have doubts about their competence, to feel not in control of their work, to feel not in control of their career success, to feel depressed, to feel worthless, and to see things as relatively bleak and hopeless. In other words, high CSE persons have confidence in their abilities and think positively of themselves. Given the qualities of the high CSE person and that research has shown a robust relation between CSE and different indicators of satisfaction, it seems logical to expect that a high CSE student generally would feel confident and positive in class and that this would be reflected in the tendency to provide more favourable SETs. Therefore, a third hypothesis is the following:

Hypothesis 3: Higher student CSE is associated with higher SETs.

In addition to being the first study to examine student CSE in the context of SETs, improvements over earlier studies also were built into the present research strategy. For example, the methodology was designed to determine whether the predictive capacities of any of the six student dispositional variables are redundant. As well, the study was designed to determine whether student personality variables relate similarly to SETs when a most preferred or a least preferred instructor is being evaluated. In addition, the study was designed to examine relations between the six personality variables and the SETs without and with several demographic and dispositional variables statistically controlled.

The initial sample consisted of 148 students enrolled in four sections of introductory psychology at Cape Breton University in Sydney, Nova Scotia, Canada. One student failed to provide any data. Of the remaining 147 participants, there were 52 men and 95 women with a mean age of 19.9 years (SD = 3.22). The 144 participants who provided complete data were retained for analysis.

The single-page most preferred instructor evaluation form began with the following instructions: 'Among the course instructors that you have had or currently have at Cape Breton University, whom would you consider your MPI? Keep focused on this MPI as you complete the following instructor evaluation form'. Students then responded to 10 items, such as the following on a five-point 'Strongly Agree' to 'Strongly Disagree' Likert scale: 'The instructor is well organised and prepared'; 'The instructor makes clear and effective classroom presentations'; 'The instructor clearly states the student's responsibilities in relation to the course'. Finally, students responded to an eleventh item: 'How would you rate this instructor in general teaching ability?' Responses were made by placing an 'X' somewhere on a line anchored by '1 Poor' and '10 Excellent'. The participant's response was coded as the closest number to the 'X'.

The single-page least preferred instructor least preferred instructor (LPI) form began with the following instructions: 'Among the course instructors that you have had or currently have at Cape Breton University, whom would you consider your least preferred instructor? Keep focused on this least preferred instructor as you complete the following instructor evaluation form'. Students then responded in the same format to the same 11 items used on the MPI evaluation form.

Openness, Conscientiousness, Extraversion, Agreeableness and Neuroticism were assessed with the BFI (John, Naumann, and Soto [19]; John and Srivastava [20]). Respondents choose a number on a Likert scale from '1' (Disagree strongly) to '5' (Agree strongly) to indicate the extent to which they see each of 44 brief descriptions as pertaining to themselves. Examples of items include 'Is curious about many different things' (Openness), 'Makes plans and follows through with them' (Conscientiousness) and 'Is talkative' (Extraversion). Some items are reverse scored, such as 'Tends to find fault with others' (Agreeableness) and 'Is relaxed, handles stress well' (Neuroticism). John et al. provided an array of reliability and validity information. For example, reported internal consistency estimates are.83 for Openness,.82 for Conscientiousness,.86 for Extraversion,.79 for Agreeableness and.87 for Neuroticism.

CSEs were gauged with the core self-evaluations scale (CSES) developed by Judge et al. ([22]). Respondents choose a number on a five-point Likert scale anchored by '1' (Disagree strongly) and '5' (Agree strongly) to indicate the extent to which they agree or disagree that each of 12 statements describes themselves. Half of the items are reverse scored. Examples of regular items are 'I determine what will happen in my life' and 'Overall, I am satisfied with myself'. Examples of reversed items are 'I am filled with doubts about my competence' and 'There are times when things look pretty bleak and hopeless to me'. Judge et al. present extensive evidence of the reliability and validity of the CSES. For example, the CSES has a unitary factor structure, has a test-retest reliability of.81, and has shown internal consistency coefficients ranging from.81 to.87.

Six questions asked the age and gender of the respondent, gender of the MPI and LPI, number of courses the respondent had been enrolled in at Cape Breton University, and number of instructors the respondent had been exposed to at Cape Breton University.

To eliminate potential order effects, eight different questionnaire booklets were constructed. Each began with a consent cover page and ended with a page containing six demographic questions. Four pages separately containing the BFI, CSES, MPI and LPI measures were inserted between the first and last page in eight different orders. The BFI and the CSES always were adjacent, and the MPI and LPI always were adjacent. Four orders placed the two personality measures first and four orders placed the instructor evaluation forms first. The booklets were stacked for distribution ordered by version.

Booklets were distributed at the beginning of a mid-semester lecture in four sections of introductory psychology after a short oral message requesting participation. Participation was voluntary and anonymous. No incentives were provided. All students consented to participate.

Responses to the 22 items of the MPI and LPI forms were standardised, and the mean z score was used as an overall or average instructor rating. This Combined MPI and LPI had a Cronbach's alpha reliability of.84. Similarly, responses to the 11 items of the MPI and the LPI were standardised separately, and their respective mean z scores served as the MPI and the LPI ratings. The MPI and the LPI had Cronbach's alpha of.90 and.86, respectively. Alpha reliabilities also were computed for the six personality variables and all were adequate: Openness (.77), Conscientiousness (.71), Extraversion (.87), Agreeableness (.78), Neuroticism (.81) and CSES (.83). Descriptive statistics for these measures and the six demographic variables are presented in Table 1.

Table 1. Means, standard deviations, minimums and maximums for the variables in the study (N = 144).

Table 2 displays Pearson correlations between the variables with two-tailed tests of significance. What stands out most is that there were positive correlations between the three instructor evaluation criteria and student Agreeableness and CSE, and negative correlations for Neuroticism, but no correlations for Openness, Conscientiousness and Extraversion.

Table 2. Correlations between the variables in the study (N = 144).

1 Notes:

• 2 p < .05,
• 3 p < .01,
• 4 p < .001.

To formally test the hypotheses that student Agreeableness, Neuroticism and CSE are associated with SETs, partial correlations were computed with the six demographic variables controlled. Partial correlations also were computed in the same manner for Openness, Conscientiousness and Extraversion. All of these calculations used one-tailed significance tests. Although relations were not predicted for Openness, Conscientiousness and Extraversion, one-tailed tests were used to ensure fair comparisons to Agreeableness, Neuroticism and CSE. Results are shown in Table 3. Each of the three hypotheses was supported. Higher student Agreeableness was associated with more positive evaluations of instructors on the Combined MPI and LPI, MPI and LPI measures, thus supporting Hypothesis 1. Higher student Neuroticism was associated with more negative evaluations of instructors on the three instructor evaluation criteria, thus supporting Hypothesis 2. Higher student CSES scores were associated with more positive evaluations of instructors on each of the three instructor evaluation measures, thus supporting Hypothesis 3. In comparison, there were no significant corresponding partial correlations for Openness, Conscientiousness or Extraversion.

Table 3. Partial correlations between student personality and instructor evaluation scores with demographic variables controlled (N = 144).

• 5 Notes: Tests of significance were one-tailed.
• 6 p < .05.
• 7 p < .01.
• 8 p < .001.

Hierarchical multiple regression analyses were conducted to determine whether relations between instructor evaluations and student Agreeableness, Neuroticism and CSE were overlapping or independent. For each instructor evaluation criterion, a regression equation was computed with the six demographic variables entered as a block on the first step and Agreeableness, Neuroticism and CSE entered stepwise on the second step. With Combined MPI and LPI as the dependent variable, the demographic controls accounted for a non-significant 5.6% of the variance in the criterion, F(6, 137) = 1.37, p = .233, CSE accounted for another 8.2%, F(1, 136) = 12.91, p < .001, and Agreeableness accounted for a final increment of 4.6%, F(1, 135) = 7.65, p < .01. With MPI as the dependent variable, the demographic controls accounted for 9.1% of the variance, F(6, 137) = 2.28, p < .05, and CSE accounted for an additional 6.8%, F(1, 136) = 10.92, p < .001. With LPI as the dependent variable, the demographic controls accounted for a non-significant 4.8% of the variance, F(6, 137) = 1.15, p = .335, and Agreeableness accounted for another 5.3%, F(1, 136) = 8.08, p < .01. Therefore, student CSE and Agreeableness were independent predictors of Combined MPI and LPI, student CSE was the sole independent predictor of MPI, and student Agreeableness was the only independent predictor of LPI. Student Neuroticism did not surface as an independent predictor of any of the three instructor evaluation variables.

In the preceding analyses with Combined MPI and LPI and with MPI as the dependent variables, Neuroticism did not emerge as a predictor because CSE was a stronger predictor, and because CSE and Neuroticism, which were correlated to a fair degree (r = −.53), tended to account for common variance in these two instructor evaluation criteria. Support for this dynamic was illustrated in a supplementary analysis in which the two hierarchical regression equations were recalculated without CSE in the potential predictor pool. With Combined MPI and LPI as the criterion, after the demographic variables entered the equation, Agreeableness entered first and accounted for an additional 6.9% of the variance in the criterion, F(1, 136) = 10.78, p < .001, and Neuroticism entered second and accounted for a further 3.7%, F(1, 135) = 5.99, p < .001. With MPI as the criterion, after the demographic variables, Neuroticism entered and accounted for a final increment of 3.3% of the variance, F(1, 136) = 5.15, p < .05.

A final set of partial correlations was computed with the six demographic variables as well as Openness, Conscientiousness and Extraversion controlled. The resulting partial correlations are shown in Table 4. All nine coefficients were significant and in the same direction as those reported earlier without the additional Big Five controls.

Table 4. Partial correlations between student personality and instructor evaluation scores with Openness, Conscientiousness, Extraversion and demographic variables controlled (N = 144).

• 9 Notes: Tests of significance were one-tailed.
• 10 p < .05.
• 11 p < .01.
• 12 p < .001.

Supplementary analyses using Hotellings t showed that correlations between MPI and each of the dispositional variables were not significantly different than correlations between LPI and each of the dispositional variables for the data in Tables 2–4. For the partial correlations, the lowest comparison p value was.30. However, the difference between the Pearson correlations involving CSE in Table 2 closely approached significance (p = .056).

The results of the present study demonstrated clearly and consistently that students' personalities were related to the ratings that they provided on instructor evaluation forms. All three hypotheses were supported. Of the Big Five personality variables, higher Agreeableness and lower Neuroticism each were associated with higher SETs. Higher CSE also was associated with higher SETs. These relations were evident with simple bivariate Pearson correlation and with partial correlation simultaneously controlling for student age, student gender, gender of MPI, gender of LPI, number of university courses in which the student had been enrolled, and the number of university instructors to which the student had been exposed. Furthermore, these relations persisted when Openness, Conscientiousness and Extraversion served as additional controls. These relations also were maintained with three different instructor evaluation variables: the MPI ratings, the LPI ratings, and the Combined MPI and LPI ratings.

Hierarchical multiple regression analyses with the six demographic variables controlled also determined that CSE and Agreeableness independently accounted for variance when the Combined MPI and LPI ratings served as the dependent variable. However, only CSE independently accounted for variance with MPI ratings as the criterion. As well, only Agreeableness independently accounted for variance when LPI ratings were the criterion. Presumably, the predictive capacities of Neuroticism were subsumed by those of CSE because of the superior predictive capacity of CSE and the substantial correlation (r = −.53) between CSE and Neuroticism.

Comparisons of the correlations and partial correlations throughout the analyses indicated too that Agreeableness, CSE and Neuroticism each correlated to about the same degree with the MPI and the LPI ratings. From these results, one should not expect SETs to be influenced by an interaction effect involving the degree to which professors are preferred and any of these three gauges of student personality. In other words, Agreeableness, CSE and Neuroticism appear to relate in a general way to professors all along the continuum of student instructor preference.

As hypothesised, student Agreeableness, Neuroticism and CSE surfaced as predictors of SETs while Extraversion, Conscientiousness and Openness did not. Given the characteristics associated with Agreeableness, Neuroticism and CSE, as outlined earlier in this article, this is not so surprising. We probably should expect more positive evaluations from students who are more agreeable, who are less neurotic, who are more emotionally stable, who have high self-esteem, who believe that they control their destiny and who have confidence in their abilities. In contrast, the qualities attached to the Extraversion, Conscientiousness and Openness constructs seem not to suggest a general negative or positive stance towards completing SETs.

The novel research strategy adopted in the present study produced results that show both consistencies and inconsistencies with past work. In regard to Agreeableness, the present results are in agreement with those of Patrick ([33]), Bonitz ([3]) and Chan ([9]), but Calton et al. ([8]) found no relation between Agreeableness and SETs. In regard to Neuroticism, the results are consistent with those of Calton et al. and with those of Bonitz, at least for males, but not with those of Patrick and Chan, who found no relation. In regard to Extraversion, the results are consistent with those of Patrick, Bonitz and Chan, who found no relation, but inconsistent with those of Calton et al. who reported a positive relation between Extraversion and SETs. In regard to Conscientiousness, the results coincide with those of Patrick and Chan, but not with those of Calton et al. and Bonitz, who found that Conscientiousness and SETs were positively correlated. In regard to Openness, the results are consistent with those of Calton et al. and Patrick, but not with those of Bonitz and Chan, who found that Openness was correlated positively with SETs. Of course, as stated earlier, there were no previous studies of the potential link between CSE and SETs.

The present study is the first of which we are aware to investigate and to report a relation between the CSE of students and their ratings of instructors. The positive nature of the correlation is consistent with other research that has found that higher CSE is associated with greater job satisfaction (e.g. Judge and Bono [21]), life satisfaction (e.g. Judge et al. [25]), and satisfaction in other contexts (e.g. Boyar and Mosley [4]; Miller and Nicols [30]). The present work also showed that student CSE and Neuroticism both are related to SETs, but that CSE is the stronger predictor and has the capacity to eliminate Neuroticism as a predictor in a hierarchical multiple regression framework. This provides important additional evidence in the ongoing discussion of the issue of overlap between the CSE and the Big Five (e.g. Judge, Heller, and Klinger [23]; Judge et al. [22]).

Although the current research approach at first glance might seem somewhat artificial, it is quite reflective of what actually occurs in the process of SETs. The present study has several strengths. The method is based on the responses of students to real professors, and all of the activities and interactions that occur over the teaching of an entire course, as opposed to, for example, the limited instructor vignettes used in the experiment of Bonitz ([3]). The approach also allowed students to provide evaluations of two instructors rather than one and, explicitly by MPI and LPI rating, data are gathered about the estimated range of responses the students are likely to provide in their actual instructor ratings during their university years. The strategy also enabled the comparison of MPI and LPI, and how dispositional and other variables might differentially influence MPI and LPI ratings. However, the method does have one noticeable drawback. The researcher cannot determine whether or not some students are rating the same instructor, and consequently cannot estimate the extent to which students' ratings are based on actual or on perceived strengths and weaknesses of the instructor, perhaps driven by the dispositional characteristics of the rater.

The present results have important implications for SETs. If the distributions of Agreeableness, Neuroticism and CSE are identical in all classes, then comparisons of averaged instructor evaluations for different instructors may be considered a valid approach. However, there is reason to assume that such distributions of dispositions often are not the same in different classes. If this is the case, then such a comparative approach may be adversely affected by biases resulting from differing student levels of Agreeableness, Neuroticism and CSE. Such differences in dispositional distributions may arise because students with different personalities are drawn to different disciplines and course options within those disciplines, or because of what students know beforehand about the instructors in whose courses they enrol. Other types of selection processes also may play a part in biassing SETs. For example, perhaps, a greater proportion of students lower on Neuroticism and higher on CSE will be more likely to enrol in more selective courses where only better students will tend to do well. It follows too that introductory courses probably will have more than their share of students who are higher on Neuroticism and lower on CSE. Therefore, the commonly found association between higher level courses and higher SETs (e.g. Goldberg and Callahan [18]) may in part result from these underlying dynamics. Furthermore, above and beyond these potential contaminating features that may bias comparisons between instructors, other indirect effects on SETs are possible. For example, within any one class, having a greater than usual proportion of students who are low on Agreeableness, high on Neuroticism and low on CSE may result in a more negative classroom climate, and this may have a negative impact on instructor behaviour and effectiveness and, in turn, this may influence students of all dispositional stripes to provide less flattering SETs.

This is the first study, according to our search of the literature, to examine the relation of both the Big Five and CSE to SETs and the first to research the association between CSE and SETs. Given the promise of the present results, there are ample opportunities for various types of future empirical inquiry. For example, in addition to replications of the present study with larger samples, one key research path should lead to the determination of the robustness and magnitude of the relations between Agreeableness, Neuroticism and CSE in a large university sample of instructors and students with actual end-of-semester institutional SETs. Another path could lead to the determination of the degree to which the associations between the student dispositional variables and SETs stem from perceived vs. actual differences in instructors arising from student personality differences, or from differences in rating generosity arising from student personality differences.

Although post-secondary institutions increasingly have come to accept the necessity of SETs as a valid part of the instructor evaluation arsenal for judgments of suitability for tenure, promotion and merit rewards, recent research results such as those reported here continue to fuel longstanding controversies regarding the validity and appropriateness of SETs for such purposes.

The perennial overwhelming desire among university and college administrators for a simple numerical rating system for the effectiveness of instructors is incontrovertible. It also appears that resistance among faculty members gradually has dwindled somewhat over the years. Nevertheless, the voluminous research has continued to expand and, as in other areas of scientific inquiry, the practice of SETs should remain open to periodic scrutiny and revision if the accumulating empirical results suggest that to be an appropriate tack.

Stewart McCann, PhD, is a professor in the Department of Psychology at Cape Breton University in Sydney, Nova Scotia, Canada. His major research interests are in social psychology, personality psychology and political psychology, but he also has long-standing interests in educational psychology and educational issues.

Christopher Gardner graduated from Cape Breton University with a BSc in psychology in May 2013.