Decision Sciences Journal of Innovative Education

Volume 6 Number 1
January 2008
Printed in the U.S.A.

Using SERVQUAL to Measure the Quality

of the Classroom Experience
Michael Stodnick† and Pamela Rogers
Department of Management, College of Business Administration, University of North Texas,
1167 Union Circle Denton, TX 76201, e-mail: [email protected], [email protected]

ABSTRACT
Over the last three decades, higher education institutions have found themselves using
vernacular that was once chiefly found in business disciplines, such as value-added and
competitive advantage. With the rising costs of tuition, newer-generation students are
seeing themselves more and more as customers and universities are beginning to adopt
customer-centric strategies and missions. However, even with this paradigm shift, little
research has been done to extend traditional service management concepts to educa-
tional settings. This research attempts to bridge this gap by applying the SERVQUAL
scale, a well-validated and widely used service operations construct, to the classroom
environment. The findings show that the SERVQUAL scale exhibits both reliability and
convergent and divergent validity; in fact, in these regards, it outperforms traditional stu-
dent assessment scales. Moreover, the scale can explain significant amounts of variances
in student-related outcome variables such as satisfaction and learning. This innovative
approach to measuring classroom service quality does indeed show that students can be
viewed as customers and has far-reaching implications to all stakeholders in the delivery
of higher education.

Subject Areas: Service Quality, SERVQUAL, Student Evaluations, Student

Learning, and Student Satisfaction.

INTRODUCTION
Over the last three decades, higher education institutions have experienced dramatic
shifts in both their funding formulas and student populations. Creating a compet-
itive advantage, once a concept largely foreign to higher education, has become a
driving force (Oldfield & Baron, 2000). The myriad of stakeholders involved in or
influenced by higher education are now seeking evidence of institutions’ effective-
ness in achieving educational goals. Although consensus among these stakeholders
as to the definition of quality education may vary by segment, the stakeholders are
of the same mindset in calling for indicators that capture performance of all those
involved in executing and improving the delivery of higher education (Nedwek &
Neal, 1994).

† Corresponding author.

115
116 Using SERVQUAL in Education

The intensified competition within higher education mirrors that found within
the service sector in general. The response of many firms to the heightened call for
enhanced quality was to implement continuous improvement programs such as total
quality management and/or Six Sigma. A key tenet to these philosophies is that
organizations should continually assess customer perceptions of service quality.
Only when data are collected and analyzed can real improvements be made (Jensen
& Artz, 2005). Universities are giving serious consideration to the issue of service
quality assessment for a multitude of reasons, arguably the two most important of
which are: students report that word-of-mouth recommendations play a large role
in their decision to choose a university and both university quality assurance and
independent assessment evaluators place heavy emphasis on the student experience
as one of their assessment criteria (Cuthbert, 1996). The underlying theory is that
institutions that continually improve service quality and delivery are more likely to
generate high levels of customer satisfaction, resulting in both increased customer
loyalty (namely, a higher retention of the current student population), and decreased
costs of attracting new students (through positive word of mouth from the students
and higher independent ratings).
In this study, we focus squarely on one portion of what Petruzzellis,
D’Uggento, and Romanazzi (2006) call the “total student experience”—that is,
the quality of the classroom encounter. The use of student ratings to provide feed-
back about the quality of instruction developed out of protests in the late 1960s
from students who increasingly saw themselves as customers (Centra, 1993). Since
that time, a vast number of studies, including several meta-analyses, have shown
that the use of student ratings is both a reliable and valid measurement of teaching
quality. A review of assessment literature conducted by Brightman et al. (1993)
concluded that ratings provide normative data that can be used as a mechanism for
teaching improvement.
Recently, this customer-centric approach of service quality has gained mo-
mentum in educational literature as the increasing cost of education has created a
new generation of students with greater customer awareness than ever before. As
Oldfield and Baron (2000) pointed out, the “interaction between customer and ser-
vice organization lies at the heart of the service delivery.” Employees who deliver
the service, in this case the instructor, are of key importance to both the customers
they serve, the students, and the employer they represent, the university. In some
regards, the employee (instructor) may be the most visible route by which the
employer (the university) can distinguish itself.
The principal instrument used in service management and marketing litera-
ture to measure service quality is the SERVQUAL scale. However, even as higher
education continues to strive toward customer-oriented strategies, very little work
has been done to combine education literature with service management and mar-
keting research. This research bridges this gap by applying the SERVQUAL scale
within a classroom setting. Can SERVQUAL, a valid and reliable customer-centric
scale used to measure the quality of service delivery in environments as diverse
as retail and business consulting, be used to measure and thus ultimately improve
the quality of service delivery in higher education? In other words, can this well-
validated scale be innovatively applied to measure student perceptions of classroom
delivery? This question is of paramount importance to all stakeholders in higher
education. Better measures of the customers’ voices through their assessment of
Stodnick and Rogers 117

service quality may ultimately lead to improved educational experience (student),

increased professional development (instructor), higher university ranking (univer-
sity itself), better-qualified graduates (community), and so on.

LITERATURE REVIEW
Full reviews of SERVQUAL and student evaluation literature are well beyond
the scope of this article. Instead, after a brief summary of what the SERVQUAL
scale is and its inception, we will focus on applications of SERVQUAL in higher
education. The conceptual underpinnings of the SERVQUAL model were first
published in 1985 (Parasuraman, Zeithaml, & Berry, 1985). In that research, the
authors focused their discussion of service quality on what Gronroos (1984) labeled
“functional quality,” or the expressive performance of a service. They argued that
there are 10 distinct dimensions to service quality. However, 3 years later when
empirically deriving a service quality definition, the list of 10 was reduced to 5; the
5 dimensions and the descriptions the authors give are listed below (Parasuraman,
Zeithaml, & Berry, 1988).
r Tangible—physical facilities, equipment, and appearance of personnel;
r Reliability—ability to perform the promised service dependably and accu-
rately;
r Responsiveness—willingness to help customers and provide prompt ser-
vice;
r Assurance—knowledge and courtesy of employees and their ability to
inspire trust and confidence; and
r Empathy—caring, individualized attention the firm provides its customers.
Over the last 20 years, authors have used the SERVQUAL scale to measure
service quality in a wide selection of industries with varying success. The primary
emphasis of these studies has been to test whether SERVQUAL is an appropriate
measure of service quality in varying contexts and to determine the antecedents
and consequences of delivering superior service quality. For a full review of the
current and future states of SERVQUAL research, see Parasuraman and Zeithaml
(2002).
We have identified five studies that have applied SERVQUAL within a uni-
versity environment. Cuthbert (1996) pioneered this stream of research by exam-
ining the applicability of the SERVQUAL scale to measure student perceptions
of university-level service quality. The author began by testing the reliability of
the five SERVQUAL dimensions and found very weak results: Cronbach alpha
scores ranging between .01 and .52. Because of these lower-than-expected scores,
the SERVQUAL items were subjected to exploratory factor analysis. Seven factors
formed and, as the author pointed out, these new factors little resembled the original
five factors. The author concluded from these results that using the SERVQUAL
scale to measure university service quality seems inappropriate. No analysis was
performed to determine whether any of the items in SERVQUAL can be used to
predict student satisfaction or any similar dependent variable.
Several studies extended Cuthbert’s (1996) initial work in this area. Old-
field and Baron (2000) replicated the Cuthbert (1996) study 4 years later, using
118 Using SERVQUAL in Education

SERVQUAL to measure student perceptions of business and management fac-

ulty. Through an application of exploratory factor analysis, the researchers found
that three factors emerge: requisite, essential items that allow students to fulfill
their study obligations; acceptable, items that are preferable rather than essential to
student development; and functional, items outside the control of the instructor and
primarily derived from university rules. The authors did not test the link between
these factors and student outcome measures. Sahney, Banwet, and Karunes (2004)
explored the possibility of using SERVQUAL to measure student perceptions of
service quality in higher education in India. Their factor analysis suggested that
the SERVQUAL items were actually unidimensional. Although the authors showed
how the SERVQUAL items can be used in quality function deployment to improve
the university’s services, no analysis was done to indicate whether any SERVQUAL
items can be predictive in nature.
Two other studies used SERVQUAL in a more focused way. Hughey, Chawla,
and Khan (2003) employed the SERVQUAL instrument to measure service qual-
ity of computer labs, carrying out two studies separated by a 2-year interval. In
both studies, the SERVQUAL items load onto three factors: staff, services, and
professionalism. The authors conducted several tests to investigate whether gen-
der, academic standing, and time spent in labs influence a student’s perception
of service quality. They found that female students tended to rate the university
more highly on the services and professionalism constructs than did their male
counterparts. The only other significant result was that juniors rated staff higher
than seniors. O’Neill (2003) studied the application of SERVQUAL in a university
orientation setting. Students were asked to assess the quality of the orientation
process immediately after orientation, time t. One month later, time t + 1, the same
students were asked to reflect back on the orientation process and fill out another
survey. The author’s analysis showed that SERVQUAL was unidimensional at time
t, but three-dimensional at time t + 1. Neither of these two studies investigated
whether SERVQUAL was a predictor of student satisfaction or any other dependent
variable.
This article further develops the application of SERVQUAL in an academic
setting and makes several key contributions. To begin with, this study is the first
to use the SERVQUAL model within a particular class, which could be consid-
ered a narrower service encounter than previous studies have used. We make this
choice for several reasons. By selecting one class, we will not aggregate data to
the extent that much of the previous SERVQUAL education research has done.
Previous research that applied SERVQUAL to measure student perceptions of the
overall university may not be capturing all the important variance in these student
perceptions. For example, if a student has one excellent instructor and one poor
instructor and is asked about the quality of instruction at the university, they might
answer “average.” While technically this is correct, it would be hard to use their
response to make specific changes and improve service quality. It would not be
fair or a good use of resources (e.g., instructor time) for the university to tell the
excellent instructor to improve because overall instruction is “average.” Likewise,
if the poor instructor reviews the results of the study, he or she may feel that he
or she has no need for improvement. In contrast, our application will target the
instructor delivering the service, thereby giving the instructor detailed, actionable
information. The instructor can glean information and improvement areas not only
Stodnick and Rogers 119

when SERVQUAL is applied to a single class they teach (giving them specific in-
formation for that particular class) but also when they aggregate their scores across
all their classes (giving them more general information about their instruction
techniques).
A second significant contribution that this study makes is the use of a more
comprehensive methodology in answering the question of whether or not it is
appropriate to use SERVQUAL to measure quality of instruction. Simply testing
the dimensionality of SERVQUAL, as previous studies have done, is only a first
step. Another vital part of the equation is comparing SERVQUAL to other student
evaluation scales. When compared to other scales, SERVQUAL’s reliability and
its ability to predict other student measures and outcomes should be similar to
existing scales. The scale we use for comparison is Brightman et al. (1993), hereto
referred to as the Brightman scale. We have chosen the Brightman scale because
of its widespread use in pedagogy literature.
The underpinnings of the Brightman scale can be found in three places. First,
many of the items used in the scale are derived from Berkeley Student Description
of Teaching instrument developed by Davis, Wood, and Wilson (Wilson, 1986).
The Wilson (1986) article that describes the survey has been cited more than 45
times; examples include Boex (2000) and Pietier, Drago, and Schibrowsky (2003).
Furthermore, the Brightman scale relies heavily on two meta-analyses of pedagogy
literature—Centra (1987) and Feldman (1989). Essentially, both of these studies
put together a list of the most common items used in student assessments and
reported which items were best in predicting various outcome measures. These
listings heavily influence the items used in the Brightman scale. Brightman et al.
(1993) applied factor analysis to a 34-item survey resulting in a six-dimensional
scale: organization and clarity, communication ability, grading and assignments,
interaction with students, intellectual and scholarly, and student motivation. This
instrument is the one currently used at the university where this research is con-
ducted.
A third major contribution of this study is testing whether the SERVQUAL
scale has predictive ability in a classroom setting. For a service quality scale to be
meaningful and useful, it must not only reliably describe customer perceptions of
quality but also have a significant relationship with other customer measures in or-
der to be actionable. An abundance of service quality literature exists in this area; a
small selection of which is reviewed later. Outside of the classroom setting, service
quality has been shown to have significant positive impact on customer satisfac-
tion (Cronin & Taylor, 1992; Marley, Collier, & Goldstein, 2004; Voss, Parasurma,
& Grewal, 1998), customer loyalty (Aldaigan & Buttle, 2002; Lee, Lee, & Yoo,
2000; McDougal & Leevesque, 2000), and various profitability and market-related
performance measures (Kamurka, Mittal, de Rosa, & Mazzon, 2002; Silvestro &
Cross, 2000; Zeithaml, Berry, & Parasuraman, 1996). Within the classroom set-
ting, Jensen and Artz (2005) have shown that the positive relationship between
service quality and satisfaction with both instructor and course holds true. Peda-
gogy literature also indicates that students who feel that they receive high-quality
instruction report higher learning and development levels than students who do not
perceive quality instruction (Cabrera, Colbeck, & Terenzini, 2001). To be valid,
the instrument used to measure service quality in a classroom environment must
support this link. As Gibbs (1995) points out, students often prefer what is actually
120 Using SERVQUAL in Education

detrimental to their long-term development. As such, a survey must contain items

that are linked to student performance and learning and not just questions that
amount to nothing more than a popularity contest.
Finally, there is a concern in educational literature that student evaluations
are biased by grade expectations (see Eiszler, 2002, and Marks & O’Connell, 2003,
for brief reviews). If SERVQUAL is indeed applicable in a classroom setting, it
should be free from this bias. This study will be the first to investigate whether this
bias exists.

RESEARCH OBJECTIVES
The specific objectives of this study are
r to investigate whether the SERVQUAL scale will be reliable and valid in
a university classroom setting;
r to determine whether the SERVQUAL scale exhibits predictive validity by
testing its relationship with student satisfaction and learning measures;
r to compare the reliability and validity of the SERVQUAL scale to that of
another well-established student evaluation scale, Brightman et al. (1993);
and
r to explore whether the SERVQUAL scale is free of grade expectation bias.
HYPOTHESES REVIEW
Four sets of hypotheses were tested in this study. They are (alternative hypotheses
are omitted for brevity):
H1a : Student evaluations of service quality, as measured by the Brightman
et al. (1993) scale, are positively associated with student satisfaction
with the course.
H1b : Student evaluations of service quality, as measured by the SERVQUAL
scale, are positively associated with student satisfaction with the course.
H2a : Student evaluations of service quality, as measured by the Brightman
et al. (1993) scale, are positively associated with student satisfaction
with the instructor.
H2b : Student evaluations of service quality, as measured by the SERVQUAL
scale, are positively associated with student satisfaction with the in-
structor.
H3a : Student evaluations of service quality, as measured by the Brightman
et al. (1993) scale, are positively associated with student perceptions
of learning.
H3b : Student evaluations of service quality, as measured by the SERVQUAL
scale, are positively associated with student perceptions of learning.
H4a : There will be no significant correlation between a student’s expected
grade and their evaluation of service quality, as measured by the Bright-
man et al. (1993) scale.
H4b : There will be no significant correlation between a student’s expected
grade and their evaluation of service quality, as measured by the
SERVQUAL scale.
Stodnick and Rogers 121

METHODOLOGY
Sample and Data Collection
The sample for this research consisted of six undergraduate Operations Manage-
ment courses at a large southwestern university. Four sections of Introduction to
Operations Management, one section of Purchasing, and one section of Production
Planning and Control were surveyed. Although individual responses were anony-
mous, descriptive statistics of the students enrolled in the courses were calculated.
The total population size was 264, of which 58% were male and 42% female.
Eighty-eight percent of the population were from the business school, 7% from the
engineering school, 2% from arts and sciences, and 3% divided among the other
schools and/or undecided. Ninety-eight percent of the population were undergrad-
uate students composed of 74% seniors, 23% juniors, and 1% sophomores, while
2% of the population were postgraduate students.

Survey Instrument
This research used an anonymous online survey to collect the data. Each student
in the six classes was asked to voluntarily fill out the survey at the end of the
semester. The questions used on the survey were derived from previous studies. The
34 questions used to measure the six instructor rating constructs of the Brightman
scale were taken verbatim from the Brightman et al. (1993) study. All of these items
were measured on a 5-point Likert scale. The 19 questions used to measure the five
SERVQUAL dimensions were adapted from the study by Parasuraman, Zeithaml,
and Berry (1991). As suggested by the authors, the wording was changed to fit the
classroom environment (see Appendix A for the 19 SERVQUAL items). Following
the advice of Oldfield and Barron (2000), who argued that perception-only scores
should be used when there is a long time delay between assessing expectation and
performance, the survey used a perceptions-only scale, as approximately 3 months
intervene between the forming of expectations at the beginning of the semester
and the rating of performance at the end. The perceptions-only scale, sometimes
called SERVPERF, has been validated in a number of research settings (Cronin
& Taylor, 1992, 1994; Lee et al., 2000). The items measuring the five constructs
used 5-point Likert scales. One question was used to measure each of the following
dependent variables: overall student satisfaction with the course, overall student
satisfaction with the instructor, amount the student learned throughout the course
(Cabrera et al., 2001), and expected grade (Eiszler, 2002). The first three questions
used a scale of 1 (lowest) to 10 (highest), the expected grade was measured on a
scale ranging from 0 (F) to 4 (A). The total sample size derived from the online
survey was 198, which yields a response rate of 75%.

Reliability and Validity—Brightman Scale

Because both scales used in this study have been well established in previous
literature, a confirmatory factor analysis approach was used in scale development.
We began by testing the reliability of the six individual constructs. Reliability
assessment was done using two different methodologies: corrected item to total
correlation (CITC) (Kerlinger, 1986) and Cronbach alpha (Nunnally, 1978). The
CITC method posits that each item within a construct should be highly correlated
122 Using SERVQUAL in Education

Table 1: Factor development of the Brightman scale.

Factor Cronbach First Second Minimum Percent Var. Alpha–
(No. of Items) Alpha Eigenvalue Eigenvalue Factor Loading Explained AVISC
Presentation .93 5.5 .8 .682 68.2 .16
ability (8)
Organization & .92 5.2 .8 .705 64.3 .13
clarity (8)
Grading & .92 5.3 .8 .622 65.5 .15
assignments (8)
Intellectual & .89 3.0 .4 .858 74.8 .16
scholarly (4)
Student .86 2.4 .4 .867 78.5 .11
interaction (3)
Student .76 2.1 .6 .736 67.7 .06
motivation (3)

with the construct itself. Kerlinger (1986) recommends that every item within the
scale should have a CITC value that exceeds .4. The lowest CITC value for the 34
questions (on their respective six constructs) was .543 for the fifth item in the
grading and assignments construct; in fact, that was the only value below .6. The
Cronbach alpha values, measures of internal consistency, are presented in Table 1.
The lowest value, .758, is well above Nunnally’s suggested cutoff of .7. All others
are .86 or above.
After assessing the scales’ reliabilities, we turned to an exploration of both
convergent and divergent validity. Convergent validity is the extent to which in-
dicators are associated with each other and represent a single concept. Divergent
validity is the degree to which a construct and its indicators differ from other con-
structs and their indicators. We tested for convergent validity by examining: the
structure of the eigenvalues (a factor should only have one eigenvalue over 1.0);
percent of variance explained (the items in the factor should explain at least 40%
of the variance in the factor); and factor loadings of each construct (all factor load-
ings should exceed .4) (Ahire & Deveraj, 2001). Table 1 contains the results. Each
factor appears to converge toward unidimensionality. We assessed divergent valid-
ity for each construct by calculating the Cronbach alpha minus average interscale
correlation (AVISC) value. The difference between the two should be substantially
greater than zero. Although there is no statistical test of significance, difference
values of .3 and .4 have been used in the past (McDougal & Levesque, 2000;
Petrick, 2002; Spreng & MacKoy, 1996). The last column in Table 1 summarizes
the findings. All six of the scores presented are very low; indicating that the scales
may not be measuring six distinct concepts, and the scales may be suffering from
multicollinearity.
Because the main analysis will use multiple regression, a technique highly
sensitive to multicollinearity, we decided to explore this issue further. For each of
the six constructs variance inflation factor (VIF) scores were calculated. In this case,
the VIF scores tell how well the sixth factor can be predicted by the remaining five.
All of the VIF scores exceeded 5.0 except for the presentation ability factor, which
Stodnick and Rogers 123

Table 2: Development of two new factors.

Minimum
Factor (No. of Cronbach First Second Factor Percent Var. Alpha–
Items) Alpha Eigenvalue Eigenvalue Loading Explained AVISC
Learning .97 14.3 .9 .581 59.3 .39
environment (24)
Organization & .94 5.4 .9 .503 59.6 .36
clarity (9)

was 4.87. This result confirms those found in the discriminant validity section and
suggests that the six constructs do not appear to be sufficiently distinct from one
another.
To move forward and assess the true factor structure of the 34 items, we de-
cided to subject the 34 items to exploratory factor analysis. Because the items were
highly correlated, direct oblimin rotation was used. Two factors emerged from the
analysis. The first factor, which we label learning environment, is essentially a com-
bination of the first three Brightman et al. (1993) dimensions: presentation ability,
organization and content, and grading and assignments. The second factor, which
we label student involvement, is a combination of the three remaining Brightman
et al. (1993) constructs: intellectual and scholarly, student interaction, and student
motivation. This result is similar to the two-factor solution that Goldstein and Be-
nassi (2006) found. Only one item, the second on the original student motivation
construct, cross-loaded (i.e., had a loading of over .4 on both factors) and thus was
dropped from further analysis. We confirmed the reliability, unidimensionality, and
discriminant validity of these two new factors. The results are presented in Table 2.

Reliability and Validity—SERVQUAL Scale

The steps for developing the SERVQUAL scale mimicked those for the Brightman
scale. The results of the analysis are presented in Table 3. These results suggest that
the five dimensions of SERVQUAL are reliable, unidimensional, and divergent.
However, before proceeding to multiple regression, we also tested VIF scores for the

Table 3: Factor development of the SERVQUAL scale.

Minimum Percent
Factor Cronbach First Second Factor Var. Alpha—
(No. of Items) Alpha Eigenvalue Eigenvalue Loading Explained AVISC
Assurance (4) .89 2.9 .6 .776 74.0 .42
Empathy (4) .94 4.0 .4 .835 79.8 .48
Responsiveness (3) .92 2.6 .2 .932 86.5 .46
Tangibles (4) .82 2.8 .6 .769 69.8 .53
Reliability (3) .92 2.6 .3 .903 85.9 .43
124 Using SERVQUAL in Education

five dimensions. The highest VIF score was for reliability at 3.24. All other scores
were less than 2.0. These results confirm the appropriateness of using multiple
regression.

RESULTS AND DISCUSSION

Hypothesis 1—Predicting Student Satisfaction with Course
Multiple regression analysis was used to determine if either of the two scales could
predict a student’s overall satisfaction with the course, the dependent variable. The
first model tested the two revised constructs derived from the study by Brightman
et al. (1993). The standardized betas for learning environment and student involve-
ment were .478 (p < .01) and .284 (p < .05), respectively. The overall model was
significant at p <.001 and had an adjusted R-squared value of .489. This finding
confirms Hypothesis 1a—student evaluations, measured using the revisions to the
Brightman scale, are positively associated with student satisfaction with the course.
Furthermore, these results confirm those of earlier studies that have also shown that
organization and clarity and presentation ability (two of the three dimensions in our
learning environment construct) have the strongest effect on student satisfaction
(Feldman, 1989; Cabrera et al., 2001).
Next, the same hypothesis was tested using the five SERVQUAL dimensions
as independent variables. The results of the multiple regression are shown below
in Table 4. At the p <.05 level, three of the five SERVQUAL dimensions were
positively related to student satisfaction with the course: empathy, reliability, and
assurance. The other two dimensions, responsiveness, and tangibles, were not sig-
nificant. The adjusted R-squared value for this model, .472, is very similar to that of
the model which used the Brightman measures. This finding confirms Hypothesis
1b—student evaluations, measured using SERVQUAL, are positively associated
with student satisfaction with the course.

Hypothesis 2—Predicting Student Satisfaction with Instructor

Similar to the methods used to test Hypothesis 1, we employed multiple regres-
sion to test Hypothesis 2, using the revised Brightman scales as the independent

Table 4: Multiple regression results, dependent variable: student satisfaction with

course.
Factor Std. Beta t Statistic p Value
Assurance .197 1.99 .048
Empathy .606 3.23 .002
Responsiveness −.090 .567 .455
Tangibles .016 .210 .834
Reliability .289 2.36 .019
F value = 19.829, p < .001.
R-squared = .472.
Stodnick and Rogers 125

Table 5: Multiple regression results, dependent variable: student satisfaction with

instructor.
Factor Std. Beta t Statistic p Value
Assurance .268 2.19 .031
Empathy .470 3.48 .001
Responsiveness −.060 −1.10 .272
Tangibles −.060 −1.11 .271
Reliability .336 3.13 .002
F value = 54.795, p < .001.
R-squared = .717.

variables first. The standardized betas for learning environment and student in-
volvement were .642 (p < .001) and .252 (p < . 05), respectively. The model had
an F statistic of 94.3, which is significant at the p <.001 level. The adjusted R-
squared value for this model was .727. Note that this value is much higher than
the R-squared found in the tests to Hypothesis 1. This finding seems reasonable,
as there are many contextual variables outside of the control of the instructor that
may affect a student’s satisfaction with the course in general, such as the dif-
ficulty of subject, the time slot of course, the dynamism between students, the
classroom, and so on. Similar to the findings in Hypothesis 1, the results of this
model demonstrated that satisfaction with instructor is more heavily influenced by
learning environment than student interaction, although both are significant. These
results confirm Hypothesis 2a—student evaluations, measured using the Bright-
man scale, are positively associated with how satisfied those students are with their
instructors.
Comparable results were found when the five SERVQUAL dimensions were
used to measure students’ perceptions. The results are summarized in Table 5.
This model found the same three dimensions significant, listed here in decreasing
order of magnitude: empathy, reliability, and assurance. The other two variables
remained nonsignificant. The overall model was significant at p <.001, confirming
Hypothesis 2b—student evaluations, measured using SERVQUAL, are positively
associated with how satisfied those students are with their instructors. The adjusted
R-squared for this model, .717, is near that found in Hypothesis 2a, illustrating that
the two student evaluation scales explain roughly the same amount of variance in
a student’s satisfaction with the instructor.

Hypothesis 3—Predicting Student Perceptions of Learning

When the two revised Brightman et al. (1993) constructs were regressed against
student perceptions of learning, only learning environment was a significant pre-
dictor. Its standardized beta of .560 is significant at the p <.001 level. Student
involvement, standardized beta of .140, t statistic of 1.00, is not significant at the
p <.10 level. The overall model F value was 47.2, which is significant at the p <.001
level. The adjusted R-squared for this model was .504. The lower R-squared value
for student learning when compared to satisfaction with instructor is expected,
as prior research has revealed that a vast array of important variables beyond the
126 Using SERVQUAL in Education

Table 6: Multiple regression results, dependent variable: amount learned.

Factor Std. Beta t Statistic p Value
Assurance .313 2.03 .043
Empathy .406 2.26 .025
Responsiveness −.180 −1.04 .300
Tangibles .005 .07 .946
Reliability .193 1.36 .178
F value = 23.839, p < .001.
R-squared = .508.

control of the instructor affect student learning: student motivation, ability, per-
sonality, and so on. (Syler et al., 2006). These regression results serve to confirm
Hypothesis 3a—student evaluations, as measured through the revised Brightman
constructs, are positively associated with student perceptions of learning.
The results of the SERVQUAL model are depicted in Table 6. In this model,
only two dimensions were significant predictors of student learning at the p <.05
level: empathy and assurance. The three other variables were all nonsignificant.
The overall model is significant at p <.001 and had an adjusted R-squared value
of .508; again, this closely resembles the adjusted R-squared in the model from
Hypothesis 3a. These results confirm Hypothesis 3b—student evaluations, as mea-
sured through SERVQUAL, are positively associated with student perceptions of
learning.

Hypothesis 4—Correlation Between Expected Grade and Student

Evaluations
As discussed earlier, one major concern with the use of student evaluations is the
issue of grade expectation bias; in other words, are students who are expecting
to receive high grades biased in their evaluations (Eiszler, 2002)? A good student
evaluation instrument should not exhibit this bias. To test for this correlation, two
preliminary steps must be taken. First, for each of the two major student evaluation
scales one overall service quality score was devised. We achieved this in three dif-
ferent ways: the average score of all items (33 for Brightman, 19 for SERVQUAL),
average score of regression factor scores (three scores for Brightman, five scores for
SERVQUAL), and average score of exact factor scores (three scores for Brightman,
five scores for SERVQUAL). All three methods produced nearly identical results:
not only did significance results remain unaltered, parameter estimates were all
extremely close, and as a result, only the details for the average score method are
given. Second, as Centra (2003) pointed out, when testing for a correlation be-
tween student evaluations and expected grade, it is necessary to control for student
learning. This was done by using partial correlations. The partial correlation, which
controls for student learning, between expected grade and student evaluation, as
measured through the Brightman scale, is −.049, which is not significant at the
p <.10 level. Similar results are produced when using the SERVQUAL scale to
calculate a total student evaluation score; the partial correlation is −.083 (p > .10).
These findings suggest that once student learning is controlled for, expected grade
Stodnick and Rogers 127

has no significant relationship with student evaluations—in other words, students

expecting good grades do not necessarily rate instructors more highly than those
expecting lower grades.

CONCLUSION AND FUTURE RESEARCH

This research is the first to apply the SERVQUAL scale to measure student per-
ceptions of service quality in a classroom setting. Although the scale itself is well
established, the application of it to the classroom is innovative in nature. The find-
ings suggest that the SERVQUAL scale is reliable and exhibits both convergent and
divergent validity. In fact, in terms of scale development, SERVQUAL performed
better than a traditional student evaluation scale, the Brightman scale. In addition,
the SERVQUAL scale has been shown to display predictive validity, because a sig-
nificant positive relationship exists between individual dimensions of SERVQUAL
and two measures of student satisfaction as well as student learning. Indeed, with
far fewer questions (44% less), the SERVQUAL scale explains roughly the same
amount of variance in these student outcome measures as does the Brightman scale.
This parsimony may help prevent respondent fatigue and in the long run lead to a
more reliable assessment instrument.
In predicting the various student outcome measures, the behavioral dimen-
sions of SERVQUAL were the most effective. The dimension of empathy con-
sistently had the strongest impact on the dependent variables. The items for this
dimension, items largely ignored in the Brightman scale, suggest that students are
looking for customer-centric qualities in instructors; in other words, an instructor
who understands the individual needs of each student and is able to give personal-
ized attention. A second SERVQUAL dimension was significant in all three models
as well—assurance. This dimension shows that the students must feel confident
in both instructors’ knowledge of their fields and their impartiality in assessment.
The reliability dimension played a significant role in two of the three regression
models, indicating that both consistency and dependability are traits that affect
student satisfaction. Neither tangibles nor responsiveness had a significant effect
in any of the three regression models, suggesting that these two traits are not as
important as the three other SERVQUAL dimensions in shaping satisfaction and
learning. Instructors can use these findings, that is, the regression rankings, as a
resource allocation tool when devising teaching strategies and trying to improve
the quality of their classroom experiences.
Collectively these results demonstrate that a customer-centric service qual-
ity scale such as SERVQUAL can be applied in a classroom setting. Although
universities have been widely considered to be as close as possible to what is con-
sidered a pure service, and students are progressively seeing themselves more and
more as consumers of said service, there has been little effort to blend existing
service management literature into current academic research streams (Oldfield
& Baron, 2000). Because the effects of improving university service radiate and
multiply throughout the service value chain, this study’s findings are wide ranging
indeed. As such, innovative methods must be devised to capture the student’s voice
as active participant and customer in a service delivery encounter. As this study
has demonstrated, one such method is using scales such as SERVQUAL. This
128 Using SERVQUAL in Education

customer-centric approach can help instructors improve their service delivery, thus
increasing service quality for many of the stakeholders in the education model:
students obtain higher-quality classroom experiences, instructors receive informa-
tion for professional development, the university gains a better reputation, future
employers will get better trained graduates, and so on. Rather than focusing solely
on many of the structural elements of the classroom experience, as many student
assessment scales do, the SERVQUAL scale focuses on the behavioral aspects of
the classroom. So in addition to trying to improve structural components such as the
syllabus, outlines, handouts, exams, and so on, instructors can use the SERVQUAL
assessment scale to understand what behavioral traits they need to improve and the
latent construct names themselves are very intuitive, powerful, and easy to un-
derstand, particularly for business instructors who are familiar with service quality
terminology. These abstract level constructs, terms such as empathy and assurance,
certainly give the instructor a different perspective of the needs of their customers
(students) than do questions asking about very specific components like pace of
instructor’s speech, clarity of handouts, and so on. For example, results of student
surveys might show instructors that they need to improve their overall responsive-
ness to their customers (students). The instructor could do so by extending office
hours, checking e-mail, and/or phone messages more often and making use of chat
tools in software programs such as WebCT and Blackboard. In general, instructors
need to think of themselves as service providers in a common business sense. They
can learn more about improving their service delivery when they incorporate these
tacit customer-centric ideas and terminology into their assessment paradigms than
by focusing on traditional scales that heavily emphasize very specific structural
components of service delivery.
At this point, it is worth mentioning the limitations and extensions of this
research. The respondents to this survey were primarily business majors enrolled in
Operations Management classes. The findings should not be generalized until con-
firmed in a variety of settings. This study could be replicated in other departments
within business schools as well as across entirely different disciplines. Likewise,
this study focused on courses that were composed primarily of juniors and seniors,
future research can extend these results to lower-level courses as well as graduate
courses. Not only should the validity and reliability of the SERVQUAL instrument
be validated, but so should the strength of the relationships between the five in-
dividual dimensions and the dependent variables. For example, this study found
that empathy and assurance were the most significant predictors of satisfaction
and learning. Perhaps in different environments, the other dimensions may be dis-
covered to be more important. For example, in computer or science lab courses,
tangibles might be a significant predictor of satisfaction and/or learning.
A second limitation is the use of self-assessment to gauge student learning
and expected grade. Because this survey was anonymous, it was impossible to
link student responses with exam scores or grades earned. However, immediately
before the survey was distributed, students were given detailed sheets of all their
scores to date with a summary at the end indicating what their grades would be if
class ended that day. Self-reported scores have been used and shown to be reliable
in previous research (e.g., Cabrera et al., 2001). Another potential methodological
limitation is the possibility of common method variance exerting undue influence
on the data set. To assess common method, we used Harman’s Single-Factor test.
Stodnick and Rogers 129

Because of the limited sample size, we only tested the 19 SERVQUAL items as
well as the 4 student outcome measures. Common method variance is assumed to
exist if a single factor emerges from the unrotated factor solution and/or the first
factor explains the majority of the variance in the items (Malhotra et al., 2006;
Podsakoff & Organ, 1986). When the 23 items are subjected to exploratory factor
analysis, 6 factors emerge, the 5 SERVQUAL dimensions as well as 1 factor of
outcome measures. The first factor accounts for only 33% of the variance in the
data. Taken together these findings suggest that the data do not exhibit extreme
common method variance.
As discussed earlier, in addition to testing the generalizability of these find-
ings, another potential area for future research is to investigate the uniqueness of
the two scales. For example, is the SERVQUAL scale explaining the identical vari-
ance in the dependent variables (satisfaction, learning) that the Brightman scale is?
Using the terminology introduced by Goldstein and Benassi (2006), SERVQUAL
focuses primarily on the “process” portion of service delivery; can the SERVQUAL
items be combined with the “structure” items included in Brightman’s scale to cre-
ate an even more comprehensive educational service quality instrument? Future
researchers can explore, possibly using hierarchical regression, the overlap be-
tween the two scales—are elements within the scales unique or complementary?
In addition, some of the structural items in Brightman’s scale, such as organization
and clarity, may actually be antecedents to the behavioral items in SERVQUAL.
For example, clear organization might allow instructors to perform more reliably
or become more responsive to student needs.
A close parallel to the idea presented above is to determine if the two scales can
be parsimoniously combined in some manner to form one global scale maximizing
the amount of variance explained in student outcome measures. Perhaps what one
scale lacks, the other includes. In a similar vein, future researchers can determine
whether a more limited set of SERVQUAL items could be used to measure the
service quality in higher education. This study used 19 items; future research can
explore whether either a subset of those 19 items or newly created items can explain
as much variance in student outcomes variables as the scales used in this research.
Finally, this study used a general measure of student learning as a depen-
dent variable. Learning has been shown to be multidimensional in nature. Future
research can investigate the relationship between SERVQUAL and specific dimen-
sions of learning such as professional competencies, group interpersonal skills,
problem-solving skills, design skills, and so on (Cabrera et al., 2001). This type of
research would aid in showing whether any additional questions should be added
to the SERVQUAL scale in order to capture instructor behavior that can influence
these specific types of learning dimensions. Likewise, another dependent variable
that could be introduced could be aimed at teaching gains. Does the SERVQUAL
scale, and the resulting analysis the instructor would receive from using it, lead to
instructors improving their classroom performance?

REFERENCES
Ahire, S., & Deveraj, S. (2001). An empirical comparison of statistical construct
validation approaches. IEEE Transactions on Engineering Management, 48,
319–329.
130 Using SERVQUAL in Education

Aldaigan, A., & Buttle, F. (2002). SYSTRA-SQ: A new measure of bank service
quality. International Journal of Service Industry Management, 13, 362–381.
Boex, L. F. J. (2000). Attributes of effective economics instructors: An analysis of
student evaluations. Journal of Economic Education, 31, 211–227.
Brightman, H., Elliott, M., & Bhada, Y. (1993). Increasing the effectiveness of
student evaluation of instructor data through a factor score comparative report.
Decision Sciences, 24, 192–199.
Cabrera, A., Colbeck, C., & Terenzini, P. (2001). Developing performance indica-
tors for assessing classroom teaching practices and student learning. Research
in Higher Education, 42, 327–352.
Centra, J. (1987). Formative and summative evaluation: Parody or paradox? In
L. M. Aleamoni (Ed.), Techniques for evaluating and improving instruction.
San Francisco: Jossey-Bass, 47–55.
Centra, J. (1993). Reflective faculty evaluation. San Francisco: Jossey-Bass.
Centra, J. (2003). Will teachers receive higher student evaluations by giving higher
grades and less coursework? Research in Higher Education, 44, 495–518.
Cronin, J., & Taylor, S. (1992). Measuring service quality: A reexamination and
extension. Journal of Marketing, 56(3), 55–68.
Cronin, J., & Taylor, S. (1994). SERVPERF versus SERVQUAL: Reconciling
performance based and perceptions minus performance measurements of
service quality. Journal of Marketing, 58(1), 125–131.
Cuthbert, P. (1996). Managing service quality in HE: is SERVQUAL the answer?
Part 1. Managing Service Quality, 6(2), 11–16.
Cuthbert, P. (1996). Managing service quality in HE: is SERVQUAL the answer?
Part 2. Managing Service Quality, 6(3), 31–35.
Eiszler, C. (2002). College students evaluations of teaching and grade inflation.
Research in Higher Education, 43, 483–501.
Feldman, K. (1989). The association between student ratings of specific instruc-
tional dimensions and student achievement. Research in Higher Education,
30, 583–645.
Gibbs, G. (1995). How can promoting excellent teachers promote excellent
teaching? Innovations in Education and International Training, 32(1), 74–
84.
Goldstein, G., & Benassi, V. (2006). Students’ and instructors’ beliefs about ex-
cellent lecturers and discussion leaders. Research in Higher Education, 47,
685–707.
Gronroos, C. (1984). A service quality model and its marketing implications. Eu-
ropean Journal of Marketing, 18(4), 36–44.
Hughey, D., Chawla, S., & Kahn, Z. (2003). Measuring the quality of university
computer labs using SERVQUAL: A longitudinal study. The Quality Man-
agement Journal, 10(3), 33–44.
Stodnick and Rogers 131

Jensen, J., & Artz, N. (2005). Using quality management tools to enhance feedback
from student evaluations. Decision Sciences Journal of Innovative Education,
3, 47–72.
Kamurka, W., Mittal, V., de Rosa, F., & Mazzon, J. (2002). Assessing the service
profit chain. Marketing Science, 21, 294–317.
Kerlinger, F. (1986). Foundations of behavioral research. New York: Holt, Rinehart
and Winston.
Lee, H., Lee, Y., & Yoo, D. (2000). The determinants of perceived service quality
and its relationship with satisfaction. Journal of Services Marketing, 14, 217–
231.
Malhotra, N., Kim, S., & Patil, A. (2006). Common method variance in IS research:
A comparison of alternative approaches and a reanalysis of past research.
Management Science, 52, 1865–1883.
Marks, N., & O’Connell, R. (2003). Using statistical control charts to analyze data
from student evaluations of teaching. Decision Sciences Journal of Innovative
Education, 1, 259–272.
Marley, K., Collier, D., & Goldstein, S. (2004). The role of clinical and process
quality in achieving patient satisfaction in hospitals. Decision Sciences, 35,
349–369.
McDougal, G., & Levesque, T. (2000). Customer satisfaction with services: Putting
perceived value into the equation. Journal of Services Marketing, 14, 392–
410.
Nedwek, B., & Neal, J. (1994). Performance indicators and regional management
tools. Research in Higher Education, 35, 75–104.
Nunnally, C. (1978). Psychometric theory. New York: McGraw-Hill.
Oldfield, B., & Baron, S. (2000). Student perceptions of service quality in a UK uni-
versity business and management faculty. Quality Assurance in Education,
8(2), 85–95.
O’Neill, M. (2003). The influence of time on student perceptions of service quality:
The need for longitudinal measures. Journal of Educational Administration,
41, 310–324.
Parasuraman, A., & Zeithmal, V. (2002). Understanding and improving service
quality: A literature review and research agenda. In B. Weitz & R. Wensley
(Eds.), Handbook of marketing. London: Sage, 339–367.
Parasuraman, A., Zeithaml, V., & Berry, L. (1985). A conceptual model of service
quality and its implications for future research. Journal of Marketing, 49(4),
41–50.
Parasuraman, A., Zeithaml, V., & Berry, L. (1988). SERVQUAL: A multiple item
scale for measuring customer perceptions of service quality. Journal of Re-
tailing, 64(1), 29–40.
Parasuraman, A., Zeithaml, V., & Berry, L. (1991). Refinement and reassessment
of the SERVQUAL instrument. Journal of Retailing, 67, 420–450.
132 Using SERVQUAL in Education

Petrick, J. (2002). Development of a multi-dimensional scale for measuring the

perceived value of a service. Journal of Leisure Research, 34, 119–134.
Petruzzellis, L., D’Uggento, A., & Romanazzi, S. (2006). Student satisfaction and
quality of services in Italian universities. Managing Service Quality, 16, 349–
364.
Pieter, J., Drago, W., & Schibrowsky, J. (2003). Virtual communities and the as-
sessment of online marketing education. Journal of Marketing Education,
25, 260–276.
Podsakoff, P., & Organ, D. (1986). Self-reports in organization research—Problems
and practices. Journal of Management, 12, 531–545.
Sahney, S., Banwet, D., & Karunes, S. (2004). A SERVQUAL and QFD approach
to total quality education. International Journal of Productivity and Perfor-
mance Management, 53, 143–166.
Silverstro, R., & Cross, S. (2000). Applying the service profit chain in a retail
environment: Challenging the satisfaction mirror. International Journal of
Service Industry Management, 11, 244–268.
Spreng, R., & Mackoy, R. (1996). An empirical evaluation of a model of perceived
service quality and satisfaction. Journal of Retailing, 72, 201–214.
Syler, R., Cegielski, C., Oswald, S., & Rainer, R. (2006). Examining drivers of
course performance. Decision Sciences Journal of Innovative Education,
4(1), 51–65.
Voss, G., Parasurman, A., & Grewal, D. (1998). The roles of price, performance,
and expectations in determining satisfaction in service exchanges. Journal of
Marketing, 62(4), 46–61.
Wilson, R. C. (1986). Improving faculty teaching: Effective use of student evalu-
ations and consultants. Journal of Higher Education, 57, 196–211.
Zeithaml, V., Berry, L., & Parasuraman, A. (1996). The behavioral consequences
of service quality. Journal of Marketing, 60(2), 31–46.

APPENDIX A—QUESTIONS USED TO MEASURE SERVQUAL

Empathy
1. The instructor is genuinely concerned about the students.
2. The instructor understands the individual needs of students.
3. The instructor has the student’s best long-term interests in mind.
4. The instructor encourages and motivates students to do their best.

Assurance
1. The instructor is knowledgeable in his/her field.
2. The instructor is fair and impartial in grading.
3. The instructor answers all questions thoroughly.
4. I am confident the instructor has an expert understanding of the material.
Stodnick and Rogers 133

Responsiveness
1. The instructor quickly and efficiently responds to student needs.
2. The instructor is willing to go out of his or her way to help students.
3. The instructor always welcomes student questions and comments.

Reliability
1. The instructor consistently provides good lectures.
2. The instructor is dependable.
3. The instructor reliably corrects information when needed.

Tangibles
1. The classroom is modern and updated.
2. The physical environment of the classroom aids learning.
3. The classroom is equipped with all the necessary equipment to aid learning.
4. The classroom is kept clean and free of distractions.

Michael Stodnick is an assistant professor of operations and supply chain man-

agement in the Department of Management at the University of North Texas. He
earned his PhD in management sciences from The Ohio State University. After re-
ceiving his BA, he worked as a materials manager for an axle company in Elkhart,
Indiana. His research interests include service operations management, continu-
ous improvement programs, and materials management. He is a member of and
has presented his research at the Decision Sciences Institute, Institute for Oper-
ations Research and the Management Sciences, and Production and Operations
Management Society.
Pamela Rogers is a doctoral candidate at the University of North Texas major-
ing in operations and supply chain management. She earned a BBA in Economics
and Marketing and an MS in Computer Education and Cognitive Systems. Her
business experience includes manufacturing, inventory control, computer-based
training, and Web-site development as well as military training. Her research in-
terests include manufacturing and service flexibility, supply chain flexibility, and
operations strategy formulation. She has presented her research at the annual meet-
ings of the Decision Sciences Institute, Southwest Decision Sciences Institute, and
Institute for Operations Research and the Management Sciences.