The current issue and full text archive of this journal is available at

www.emeraldinsight.com/0263-4503.htm

Measuring
Measuring service quality in service quality in
higher education: HEdPERF higher education
versus SERVPERF
31
Firdaus Abdullah
MARA University of Technology, Jalan Meranek, Malaysia Received May 2004
Revised October 2005
Accepted October 2005
Abstract
Purpose – This paper aims to test and compare the relative efficacy of three measuring instruments
of service quality (namely Higher Education PERFormance (HEdPERF), SERVPERF and the
moderating scale of HEdPERF-SERVPERF) within a higher education setting. The objective was to
determine which instrument had the superior measuring capability in terms of unidimensionality,
reliability, validity and explained variance.
Design/methodology/approach – After a pilot test, data were collected from students in two
public universities, one private university and three private colleges in Malaysia between January and
March 2004, by the “contact person” route. From a total of 560 questionnaires, 381 were usable: a
response rate of 68.0 per cent. This sample of nearly 400,000 students in Malaysian tertiary institutions
was in line with the generalized scientific guideline for sample size decisions. Data were subjected to
regression analysis.
Findings – A modified five-factor structure of HEdPERF is put forward as the most appropriate
scale for the higher education sector.
Research limitations/implications – Since this study only examined the respective utilities of
each instrument within a single industry, any suggestion that the HEdPERF is generally superior
would still be premature. Nonetheless, the current findings do provide some important insights into
how these instruments of service quality compare with one another.
Practical implications – The single dominant factor on this study is “access”, which has clear
implications for institutions’ marketing strategies.
Originality/value – This is believed to be the first study of its kind carried out among consumers of
the higher education service.
Keywords Service quality assurance, Higher education, Measuring instruments
Paper type Research paper

Introduction
Service industries are playing an increasingly important role in the economy of many
nations. In today’s world of global competition, rendering quality service is a key for
success, and many experts concur that the most powerful competitive trend currently
shaping marketing and business strategy is service quality. Since the 1980s service
quality has been linked with increased profitability, and it is seen as providing an
important competitive advantage by generating repeat sales, positive word-of-mouth
feedback, customer loyalty and competitive product differentiation. As Zeithaml and Marketing Intelligence & Planning
Bitner (1996, p. 76) point out: Vol. 24 No. 1, 2006
pp. 31-47
q Emerald Group Publishing Limited
. . . the issue of highest priority today involves understanding the impact of service quality on 0263-4503
profit and other financial outcomes of the organisation. DOI 10.1108/02634500610641543
MIP Service quality has since emerged as a pervasive strategic force and a key strategic
24,1 issue on management’s agenda. It is no surprise that practitioners and academics alike
are keen on accurately measuring service quality in order better to understand its
essential antecedents and consequences, and ultimately, establish methods for
improving quality to achieve competitive advantage and build customer loyalty.
The pressures driving successful organisations toward top quality services make the
32 measurement of service quality and its subsequent management of utmost importance.
Interest in the measurement of service quality is thus understandably high. However,
the problem inherent in the implementation of such a strategy has been compounded
by the elusive nature of service quality construct, rendering it extremely difficult to
define and measure. Although researchers have devoted a great deal of attention
to service quality, there are still some unresolved issues that need to be addressed, and
the most controversial one refers to the measurement instrument.
An attempt to define the evaluation standard independent of any particular service
context has stimulated the setting up of several methodologies. In the last decade, the
emergence of diverse instruments of measurement such as SERVQUAL (Parasuraman
et al., 1988), SERVPERF (Cronin and Taylor, 1992) and evaluated performance (EP)
(Teas, 1993a, b) has contributed enormously to the development in the study of service
quality. SERVQUAL operationalises service quality by comparing the perceptions of
the service received with expectations, while SERVPERF maintains only the
perceptions of service quality. On the other hand, EP scale measures the gap between
perceived performance and the ideal amount of a feature rather than the customer’s
expectations. Diverse studies using these scales have demonstrated the existence of
difficulties resulting from the conceptual or theoretical component as much as from the
empirical component.
Nevertheless, many authors concur that customers’ assessments of continuously
provided services may depend solely on performance, thereby suggesting that
performance-based measure explains more of the variance in an overall measure of
service quality (Oliver, 1989; Bolton and Drew, 1991a, b; Cronin and Taylor, 1992;
Boulding et al., 1993; Quester et al., 1995). These findings are consistent with other
research that have compared these methods in the scope of service activities, thus
confirming that SERVPERF (performance-only) results in more reliable estimations,
greater convergent and discriminant validity, greater explained variance, and
consequently less bias than the SERVQUAL and EP scales (Cronin and Taylor, 1992;
Parasuraman et al., 1994; Quester et al., 1995; Llusar and Zornoza, 2000). Whilst its
impact in the service quality domain is undeniable, SERVPERF being a generic
measure of service quality may not be a totally adequate instrument by which to assess
the perceived quality in higher education.
Nowadays, higher education is being driven towards commercial competition
imposed by economic forces resulting from the development of global education
markets and the reduction of government funds that forces tertiary institutions to seek
other financial sources. Tertiary institutions had to be concerned with not only what
the society values in the skills and abilities of their graduates (Ginsberg, 1991; Lawson,
1992), but also how their students feel about their educational experience (Bemowski,
1991). These new perspectives call attention to the management processes within the
institutions as an alternative to the traditional areas of academic standards,
accreditation and performance indicators of teaching and research.
 Tertiary educators are being called to account for the quality of education that they Measuring
provide. While more accountability in tertiary education is probably desirable, the service quality in
mechanisms for its achievement are being hotly debated. Hattie (1990) and Soutar and
McNeil (1996) oppose the current system of centralised control, in which the higher education
government sets up a number of performance indicators that are linked to funding
decisions. There are a number of problems in developing performance indicators in
tertiary education. One such problem is that performance indicators tend to become 33
measures of activity rather than true measures of the quality of students’ educational
service (Soutar and McNeil, 1996). These performance indicators may have something
to do with the provision of tertiary education, but they certainly fail to measure the
quality of education provided in any comprehensive way.
A survey conducted by Owlia and Aspinwall (1997) examined the views of different
professionals and practitioners on the quality in higher education and concluded that
customer-orientation in higher education is a generally accepted principle. They
construed that from the different customers of higher education, students were given
the highest rank. Student experience in a tertiary education institution should be a key
issue of which performance indicators need to address. Thus it becomes important to
identify determinants or critical factors of service quality from the standpoint of
students being the primary customers.
In view of that, Firdaus (2005) proposed HEdPERF (Higher Education
PERFormance-only), a new and more comprehensive performance-based measuring
scale that attempts to capture the authentic determinants of service quality within higher
education sector. The 41-item instrument has been empirically tested for
unidimensionality, reliability and validity using both exploratory and confirmatory
factor analysis. Therefore, the primary question is directed at the measurement of service
quality construct within a single, empirical study utilising customers of a single industry,
namely higher education. Specifically, the ability of the more concise HEdPERF scale is
compared with that of two alternatives namely SERVPERF instrument and the merged
HEdPERF-SERVPERF as moderating scale. The goal is to assess the relative strengths
and weaknesses of each instrument in order to determine which instrument had the
superior measurement capability in terms of unidimensionality, reliability, validity and
explained variance of service quality. Eventually, the results of this comparative study
were used to refine the HEdPERF scale, transforming it into an ideal measuring
instrument of service quality for higher education sector.

Research foundations
Many researchers (Parasuraman et al., 1985; Carman, 1990; Bolton and Drew, 1991a, b)
concur that service quality is an elusive concept, and there is considerable debate about
how best to conceptualise this phenomenon. Lewis and Booms (1983, p. 100) were
perhaps the first to define service quality as a “. . . measure of how well the service level
delivered matches the customer’s expectations”. Thereafter, there seems to be a broad
consensus that service quality is an attitude of overall judgement about service
superiority, although the exact nature of this attitude is still hazy. Some suggest
that it stems from a comparison of performance perceptions with expectations
(Parasuraman et al., 1988), while others argue that it is derived from a comparison of
performance with ideal standards (Teas, 1993a, b) or from perceptions of performance
alone (Cronin and Taylor, 1992).
MIP In terms of measurement methodologies, a review of literature provides plenty of
24,1 service quality evaluation scales. Some stem from the realisation of conceptual models
produced to understand the evaluation process (Parasuraman et al., 1985), and others
come from empirical analysis and experimentation on different service sectors (Cronin
and Taylor, 1992; Franceschini and Rossetto, 1997b; Parasuraman et al., 1988). The
most widely used methods applied to measure perceived quality can be characterised
34 as primarily quantitative multi-attribute measurements. Within the attribute-based
methods, a great number of variants exist and among these variants, the SERVQUAL
and SERVPERF instruments have attracted the greatest attention.
Generally, most researchers acknowledge that customers have expectations and
these serve as standards or reference points to evaluate the performance of an
organisation. However, the unresolved issues of expectations as a determinant of
perceived service quality have resulted in two conflicting measurement paradigms: the
disconfirmation paradigm (SERVQUAL) which compares the perceptions of the
service received with expectations, and the perception paradigm (SERVPERF) which
maintains only the perceptions of service quality. These instruments share the same
concept of perceived quality. The main difference between these scales lies in the
formulation adopted for their calculation, and more concretely, the utilisation of
expectations and the type of expectations that should be used.
Most research studies do not support the five-factor structure of SERVQUAL
posited by Parasuraman et al. (1988), and administering expectation items is also
considered unnecessary (Carman, 1990; Parasuraman et al., 1991a, b; Babakus and
Boller, 1992). Cronin and Taylor (1992) were particularly vociferous in their critiques,
thus developing their own performance-based measure, dubbed SERVPERF. In fact,
the SERVPERF scale is the unweighted perceptions components of SERVQUAL,
which consists of 22 perception items thus excluding any consideration of
expectations. In their empirical work in four industries, Cronin and Taylor (1992)
found that unweighted SERVPERF measure (performance-only) performs better that
any other measure of service quality, and that it has greater predictive power (ability to
provide an accurate service quality score) than SERVQUAL. They argue that current
performance best reflects a customer’s perception of service quality, and that
expectations are not part of this concept.
Likewise, Boulding et al. (1993) reject the value of an expectations-based
SERVQUAL, and concur that service quality is only influenced by perceptions. Quester
et al. (1995) perform similar analysis to Cronin and Taylor in the Australian
advertising industry, and their empirical tests show that SERVPERF performs best,
while SERVQUAL performs worst, although the differences are small. Teas (1993a) on
the other hand, discusses the conceptual and operational difficulties of using the
“expectations minus performance” approach, with a particular emphasis on
expectations. His empirical test subsequently produces two alternatives of perceived
service quality measures namely EP and normed quality (NQ). He concludes that the
EP instrument, which measures the gap between perceived performance and the ideal
amount of a feature rather than the customer’s expectations, outperforms both
SERVQUAL and NQ.
A review of service quality literature brings forward diverse arguments in relation
to the advantages and disadvantages in the use of these instruments. In general, the
arguments make reference to aspects related to the characteristics of these
scales notably their reliability and validity. Recently, Llusar and Zornoza (2000) concur Measuring
that SERVPERF results in more reliable estimations, greater convergent and service quality in
discriminant validity, greater explained variance, and consequently less bias than the
EP scale. These results are consistent with earlier research that had compared these higher education
methods in the scope of service activities (Cronin and Taylor, 1992; Parasuraman et al.,
1994). In fact, the marketing literature appears to offer considerable support for
the superiority of simple performance-based measures of service quality (Mazis et al., 35
1975; Churchill and Surprenant, 1982; Carman, 1990; Bolton and Drew, 1991a, b;
Boulding et al., 1993; Teas, 1993a; Quester et al., 1995).

Research methodology
Research objectives
On the basis of the conceptual and operational concerns associated with the generic
measures of service quality, the present research attempts to compare and contrast
empirically the HEdPERF scale against two alternatives namely the SERVPERF and
the merged HEdPERF-SERVPERF scales. The primary goal is to assess the relative
strengths and weaknesses of each instrument in order to determine which instrument
had the superior measurement capability in terms of unidimensionality, reliability,
validity and explained variance of service quality. The findings were eventually used
in transforming HEdPERF into an ideal measuring instrument of service quality for
higher education sector. The various steps involved in this comparative study are
shown by means of flow chart in Figure 1.

Research design
Data were collected by means of a structured questionnaire comprising of four sections
namely A, B, C and D. Section A contained nine questions pertaining to student
respondent profile. While sections B and C required respondents to evaluate the service
components of their tertiary institution, in which only perceptions data were collected
and analysed. Specifically, section B consisted of 22 perception-items extracted from
the original SERVPERF scale (Cronin and Taylor, 1992), and modified to fit into higher
education context.
Section C on the other hand is composed of 41 items extracted from the original
HEdPERF (Firdaus, 2005), a scale uniquely developed to embrace different aspects of
tertiary institution’s service offering. As the items were generated and validated within
higher education context, no modification was required. All the items in sections B and
C were presented as statements on the questionnaire, with the same rating scale used
throughout, and measured on a 7-point, Likert-type scale that varied from 1 ¼ strongly
disagree to 7 ¼ strongly agree. In addition to the main scale addressing individual
items, respondents were asked in section D to provide an overall rating of the service
quality, satisfaction level and future visits. There were also three open-ended questions
allowing respondents to give their personal views on how any aspect of the service
could be improved.
The draft questionnaire was eventually subjected to pilot testing with a total of
30 students, and they were asked to comment on any perceived ambiguities, omissions
or errors concerning the draft questionnaire. The feedback received was rather
ambiguous thus only minor changes were made accordingly, for instance, technical
jargons were rephrased to ensure clarity and simplicity. The revised questionnaire was
MIP
24,1

36

Figure 1.
Comparing HEdPERF,
SERVPERF and
HEdPERF-SERVPERF
subsequently submitted to three experts (an academician, a researcher and a Measuring
practitioner) for feedback before being administered for a full-scale survey. These service quality in
experts viewed that the draft questionnaire was rather lengthy, which in fact coincided
with the preliminary feedback from students. Nevertheless, in terms of number of items higher education
in the questionnaire, current study somewhat conforms with similar research works
(Cronin and Taylor, 1992; Teas, 1993a, b; Lassar et al., 2000; Mehta et al., 2000; Robledo,
2001) that attempted to compare various measuring instruments of service quality. 37
In the subsequent full-scale survey, data were collected from students of six higher
learning institutions (two public universities, one private university and three private
colleges) in Malaysia for the period between January and March 2004. Data had been
collected using the “personal-contact” approach as suggested by Sureshchandar et al.
(2002) whereby “contact persons” (registrar or assistant registrar) have been approached
personally, and the survey explained in detail. The final questionnaire together with a
cover letter was then handed personally or mailed to the “contact persons”, who in turn
distributed it randomly to students within their respective institutions.
A total of 560 questionnaires were distributed to six tertiary institutions, of these 390
were returned and nine discarded due to incomplete responses, thus leading to a response
rate of 68.0 per cent. The number of usable sample size of 381 for a population size of nearly
400,000 students in Malaysian tertiary institutions was in line with the generalized
scientific guideline for sample size decisions as proposed by Krejcie and Morgan (1970).
In order to determine which instrument had the superior measurement capability, a
new scale was developed by merging the two measuring instruments of HEdPERF and
SERVPERF. The scope of the empirical investigation notably the methodology utilised
for the development of the merged scale was defined. Next, the results obtained from
the three instruments were computed, comparing them based on the widely-used
criteria of unidimensionality, reliability, validity and their ability to predict service
quality. The results of this comparative study were subsequently used to refine the
HEdPERF scale, transforming it into an ideal measuring instrument of service quality
for higher education sector.

Results and discussion

Developing the merged HEdPERF-SERVPERF scale
The literature appears to offer considerable support for the superiority of SERVPERF
in comparison with other generic instruments. HEdPERF on the other hand, has been
empirically tested as the more comprehensive and industry-specific scale, which was
uniquely designed for higher education. Thus, it would seem rational to combine the
two finest scales in developing possibly the pre-eminent one, and subsequently to
determine which of these three instruments had the superior measurement capability
in terms of unidimensionality, reliability, validity and explained variance of service
quality. In developing the new scale, factor analysis was used to determine a new
dimensional structure of service quality by merging HEdPERF and SERVPERF items.
Specifically, this technique allowed reduction of a large number of overlapping
variables to a much smaller set of factors.
One critical assumption underlying the appropriateness of factor analysis is to
assess the overall significance of the correlation matrix with Bartlett test of sphericity,
which provides the statistical probability that the correlation matrix has significant
correlations among at least some of the variables. The results were significant,
MIP x 2 ð50; N ¼ 381Þ ¼ 13; 073; ( p ¼ 0.01), a clear indication of suitability for factor
24,1 analysis. Next, Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy was
computed to quantify the degree of intercorrelations among the variables, and the
results indicate an index of 0.89, a “meritorious” sign of adequacy for factor analysis
(Kaiser, 1970). As for the adequacy of the sample size, there is a 7-1 ratio of
observations to variables in this study, which falls within acceptable limits.
38 HEdPERF’s proposed measure of service quality was a 41-item scale, consisting of 13
items adapted from SERVPERF, and 28 items generated from literature review and
various qualitative research inputs namely focus groups, pilot test and expert validation
(Firdaus, 2005). The 41-item instrument has been empirically tested for unidimensionality,
reliability and validity using both exploratory and confirmatory factor analysis. Therefore,
all the 50 items (28 HEdPERF’s and 22 SERVPERF’s items) were included in the factor
analysis utilising the maximum likelihood method, which was followed by a varimax
rotation. The decision to include a variable in a factor was based on factor loadings greater
than ^0.3 (Hair et al., 1995). The choice regarding factor loadings greater than ^0.3 was
not based on any mathematical proposition but relates more to practical significance.
According to Hair et al. (1995, p. 385), factor loadings of 0.3 and above are considered
significant at p ¼ 0.05 with a sample size of 350 respondents (N ¼ 381 in this study).
Scree test was used to identify the optimum number of factors that can be extracted
before the amount of unique variance begins to dominate the common variance structure
(Cattell, 1966). The scree plot provided four factors, and these factors were subsequently
rotated using a varimax procedure. The variable’s communality, which represents the
amount of variance accounted for by the factor solution for each variable was also
assessed to ensure acceptable levels of explanation. The results show that communalities
in 13 variables were below 0.50, “. . . too low for having sufficient explanation” (Hair et al.,
1995, p. 387). Consequently, a new factor solution was derived with the non-loading
variables eliminated and the results yielded four factors, which accounted for 41.0 per cent
of the variation in the data (compared to 37.2 per cent of the variance explained in the first
factor solution). Table I shows the results of the factor analysis in terms of factor name,
the variables loading on each factor and the variance explained by each factor. The four
factors identified in Table I can be described as follows:
(1) Non-academic aspects. This factor contains variables that are essential to enable
students fulfil their study obligations, and it relates to duties and
responsibilities carried out by non-academic staff.
(2) Academic aspects. This factor represents the responsibilities of academics, and
it highlights key attributes such as having positive attitude, good
communication skill, allowing sufficient consultation, and being able to
provide regular feedback to students.
(3) Reliability. This factor consists of items that put emphasis on the ability
to provide the pledged service on time, accurately and dependably.
(4) Empathy. This factor relates to the provision of individualized and personalized
attention to students with clear understanding of their specific and growing
needs while keeping their best interest at heart.

It is important to note that the four factors identified did not conform exactly
with neither the six-factor structure of HEdPERF nor the five-factor structure of
 Measuring
Factor 1: Factor 2:
non-academic academic Factor 3: Factor 4: service quality in
Variables aspects aspects reliability empathy higher education
Promises kept 0.65
Sympathetic and reassuring in solving problems 0.38 0.57
Dependability 0.52 39
On-time service provision 0.74
Responding to request promptly 0.51
Trust 0.44
Feeling secured with the transaction 0.47 0.31
Politeness 0.45 0.32
Individualised attention 0.68
Giving personalized attention 0.74
Knowing student needs 0.66
Keeping student interests at heart 0.55
Knowledge in course content 0.33 0.57
Showing positive attitude 0.66
Good communication 0.75
Feedback on progress 0.68
Sufficient and convenient consultation time 0.56 0.32
Excellent quality programmes 0.62
Variety of programmes/specialisations 0.43
Flexible syllabus and structure 0.31 0.60
Reputable academic programmes 0.31 0.56
Educated and experience academicians 0.50
Efficient/prompt dealing with complaints 0.51
Good communication 0.51 0.37
Positive work attitude 0.53 0.36
Knowledge of systems/procedures 0.50 0.36
Providing service within reasonable time 0.51
Equal treatment and respect 0.75
Fair amount of freedom 0.56
Confidentiality of information 0.65
Easily contacted by telephone 0.57
Counselling services 0.53 0.32
Student’s union 0.33
Feedback to improve service performance 0.36 0.32 0.33
Standardised and simple delivery procedures 0.42 0.35
Eigenvalues 10.29 2.79 2.68 1.72 Table I.
Percentage of variance 26.2 5.9 5.8 3.0 Results of factor analysis
Cummulative percentage of variance 26.2 32.2 38.0 41.0 (factor loadings)

SERVPERF. In fact, the new dimensions extracted were the result of the amalgamation
between HEdPERF and SERVPERF scales, in which two factors (non-academic
aspects and academic aspects) were found in HEdPERF and the other two (reliability
and empathy) were identified in SERVPERF.

Comparative test of unidimensionality

A highly mandatory condition for construct validity and reliability checking is the
unidimensionality of the measure, which is referred to the existence of a single
construct/trait underlying a set of measures (Hattie, 1985; Anderson and Gerbing, 1991).
MIP In order to perform a comparative check of unidimensionality, a measurement model is
24,1 specified for each scale and confirmatory factor analysis is run by means of structural
equation modelling within LISREL framework (Joreskog and Sorbom, 1978).
Specifically, LISREL 8.3 (Scientific Software International, Chicago, IL) for windows
was used to compare the three measuring instruments where individual items in the
model are examined to see how closely they represent the same concept.
40 Table II presents the measures of model fit for all the three scales. The overall fit of the
model to the data was evaluated in various ways. Specifically, an exact fit of a model is
indicated when the p for chi-square (x 2) is above a certain value (usually set to p . 0.05)
as well as indicated by other goodness-of-fit measures. While chi-square is sensitive to
sample size and tends to be significant in large samples, a relative likelihood ratio
between a chi-square and its degrees of freedom was used. According to Eisen et al.
(1999), a relative likelihood ratio of five or less was considered an acceptable fit, a
prerequisite attained by all the three scales. A number of goodness-of-fit measures were
proposed to eliminate or reduce the dependence on sample size. These indices have
values ranging between 0 and 1, with higher values indicating a better fit. Table II shows
the indices for the three scales and all the values are close to 1, indicating that there is an
evidence of unidimensionality for the scales (Bryne, 1994).
The next measure to consider is the root mean square error of approximation
(RMSEA), which is the measure of the discrepancy per degree of freedom. The RMSEA
is generally regarded as one of the most informative fit indices (Brown and Cudeck,
1993; Diamantopoulos and Siguaw, 2000). As illustrated in Table II, the RMSEA value
for the HEdPERF was 0.07 (dimension Understanding removed), an evidence of fair fit
to the data. While both the SERVPERF and the moderating HEdPERF-SERVPERF
scales showed a poor fit of 0.08. Therefore, it was concluded that the modified
HEdPERF model fits fairly and represents a reasonably close approximation in the
population. Nevertheless, Bryne (1998, p. 119) cautions that:
. . . fit indices provide no guarantee whatsoever that a model is useful. . . Fit indices yield
information bearing only on the model’s lack of fit . . . they can in no way reflect the extent to
which the model is plausible; this judgement rests squarely on the shoulders of the researcher.

Measures of fit HEdPERF a SERVPERF HEdPERF-SERVPERF

2
Chi-square (x ) at p ¼ 0.01 2404.57 613.89 1938.79
Degree of freedom (df) 726 200 540
Relative likelihood ratio (x 2/df) 3.31 3.07 3.59
Goodness-of-fit index (GFI) 0.75 0.87 0.76
Adjusted goodness-of-fit index (AGFI) 0.71 0.83 0.72
Comparative fit index (CFI) 0.95 0.91 0.92
Non-normed fit index (NNFI) 0.95 0.89 0.92
Incremental fit index (IFI) 0.95 0.91 0.92
Root mean squared error of
approximation (RMSEA) 0.07 0.08 0.08
a
Table II. Note: The overall fit assessment was computed with the exclusion of dimension understanding due
Unidimensionality check to its poor fit, RMSEA ¼ 0.08
Comparative test of reliability Measuring
Unidimensionality alone is not sufficient to ensure the usefulness of a scale. The service quality in
reliability of the composite score should be assessed after unidimensionality has been
acceptably established. In this study coefficient a or Cronbach’s a was computed for higher education
the service quality dimensions of all the three instruments. As a guideline, an a-value
of 0.70 and above is considered to be the criteria for demonstrating internal consistency
of new scales and established scales, respectively (Nunnally, 1988). In general, there 41
was relatively good internal consistency in all the dimensions representing the
three scales.
The Cronbach’s a for HEdPERF dimensions ranged from 0.81 to 0.92, with the
exception of the dimension understanding (a ¼ 0.63). Owing to its low reliability score,
the dimension understanding was removed as part of the scale modification process.
The results, as shown in Table III, indicated that the reliability scores of the modified
HEdPERF were comparatively superior than SERVPERF’s range of 0.68-0.76, and
slightly better than HEdPERF-SERVPERF’s range of 0.77-0.91. In fact, previous
replication studies (Carman, 1990; Babakus and Boller, 1992; Finn and Lamb, 1991)
that compared various measuring instruments within a single setting were unable to
demonstrate such superiority.

Comparative test of validity

While internal consistency estimates of reliability show higher values for the modified
HEdPERF scale, the next comparative test involves assessing the validity of the three
instruments. For the purpose of this study, two validity tests were conducted namely
criterion validity and construct validity. The criterion variable used to compare the
three scales was the respondents’ global assessment of service quality (Bollen, 1989,
p. 186). Whereas the variable used in the construct validity tests was the global
preference, which was measured by the summation of the overall satisfaction and
future visit intentions (Bollen, 1989, p. 188).
The degree of criterion and construct validity was subsequently computed using the
pairwise correlations between the global assessment of service quality, the global
preference and each of HEdPERF, SERVPERF and SERVPERF-HEdPERF measures.
The validity coefficients for the three scales are all significant at p ¼ 0.01 level.
The criterion and construct validity coefficients were 0.58 and 0.57, respectively, for the
modified HEdPERF scale, 0.27 and 0.34, respectively, for SERVPERF scale, and 0.53
and 0.57, respectively, for the moderating SERVPERF-HEdPERF scale. The results

Cronbach Cronbach Cronbach

alpha SERVPERF alpha HEdPERF-SERVPERF alpha
HEdPERF dimensions (a ) dimensions (a ) dimensions (a )

Non-academic aspects 0.92 Responsiveness 0.68 Non-academic aspects 0.91

Academic aspects 0.89 Assurance 0.75 Academic aspects 0.87
Reputation 0.85 Empathy 0.76 Reliability 0.88
Access 0.88 Tangible 0.73 Empathy 0.77
Programme issues 0.81 Reliability 0.74
Understanding a 0.63
Table III.
Note: a Dimension discarded Reliability coefficients
MIP indicated that the validity coefficients for the modified HEdPERF scale were
24,1 significantly greater than both the SERVPERF and SERVPERF-HEdPERF scales.
In other words, these findings demonstrated yet again the superiority of the modified
HEdPERF in terms of criterion and construct validity.

Comparative regression analysis

42 The effect size. The regression model considered the global assessment of service
quality as a dependent variable and the service quality scores for the individual
dimensions of HEdPERF, SERVPERF and SERVPERF-HEdPERF as the independent
variables. A multiple regression analysis was subsequently conducted to evaluate how
well these scales predicted service quality level (Table IV). The linear combination of
the modified five-dimension HEdPERF (dimension understanding dropped) was
significantly related to the service quality level, R 2 ¼ 0:35; adjusted R 2 ¼ 0:34;
Fð5; 354Þ ¼ 38:4; p ¼ 0:01: The sample multiple correlation coefficient was 0.59,
indicating that approximately 34.8 per cent of the variance of the service quality level
in the sample can be accounted for by the linear combination of the five dimensions of
HEdPERF scale.
Likewise, the five dimensions of SERVPERF was also significantly related to the
service quality level, R 2 ¼ 0:24; adjusted R 2 ¼ 0:23; Fð5; 354Þ ¼ 22:1; p ¼ 0:01:
However, the sample multiple correlation coefficient of 0.49 was much lower than
HEdPERF’s, thus indicating that only 24.0 per cent of the variance of the service
quality level can be accounted for by the linear combination of the five dimensions of
SERVPERF scale. As for the moderating SERVPERF-HEdPERF scale, the linear
combination of the four dimensions was also significantly related to the service quality
level, R 2 ¼ 0:31; adjusted R 2 ¼ 0:30; Fð4; 355Þ ¼ 28:92; p ¼ 0.01. The sample
multiple correlation coefficient was 0.55, slightly lower than HEdPERF’s, thus
indicating that approximately 30.5 per cent of the variance of service quality level can

Measuring scales Standardised coefficients (b) Significant ( p)

a 2
HEdPERF (R ¼ 0.34)
Non-academic aspects 20.11 0.39
Academic aspects 20.02 0.85
Reputation 0.29 0.03
Access 0.41 0.01
Programme issues 0.17 0.12
SERVPERF (R 2 ¼ 0.23)
Responsiveness 20.06 0.27
Assurance 0.28 0.01
Empathy 20.10 0.06
Tangible 0.21 0.01
Reliability 0.09 0.12
HEdPERF-SERVPERF (R 2 ¼ 0.30)
Non-academic aspects 0.26 0.03
Academic aspects 0.22 0.01
Table IV. Reliability 0.11 0.32
Effect size and relative Empathy 20.07 0.10
importance of the
individual dimensions Note: a The modified version with dimension understanding removed
be explained by the four dimensions. The findings from the regression analysis once Measuring
again demonstrated that the modified HEdPERF scale was better in explaining the service quality in
variance of service quality level.
The relative influence. Table IV shows the results of the relative influence of the higher education
individual dimensions of the three scales. The dependent variable was the global
assessment of service quality level. The resultant output of HEdPERF had an adjusted
R 2 of 0.34 ( p ¼ 0.01) and yielded only one significant dimension namely “Access”, 43
which concurred with the findings by Firdaus (2005). It alone accounted for 13.7 per
cent ð0:372 ¼ 0:14Þ of the variance of service quality level, while the other dimensions
contribute an additional 21.1 per cent (34.8-13.7 per cent). This implied that the
dimensions “Non-academic aspects”, “Academic aspects”, “Reputation”, and “Program
issues” did not contribute significantly towards explaining the variance in the overall
rating.
SERVPERF scale on the other hand had an adjusted R 2 of 0.23 ( p ¼ 0.01) and
yielded two significant dimensions namely “Tangible” and “Assurance”, which
explained almost all the variance in the service quality level. The other dimensions
namely “Responsiveness”, “Reliability” and “Empathy” did not contribute significantly
towards explaining the variance in the overall rating. As for the
SERVPERF-HEdPERF scale, the adjusted R 2 was 0.30, and yielding two significant
dimensions namely non-academic aspects and academic aspects. The two dimensions
accounted for 23.0 per cent ((0.26 þ 0.22)2 ¼ 0.23) of the variance of service quality
level, while the other two dimensions contribute an additional 7.5 per cent (23.0-30.5 per
cent). This implied that the dimensions reliability and empathy did not contribute
significantly towards explaining the variance in the overall rating.

The modified HEdPERF scale

The empirical analysis indicated that the modified five-factor structure with 38 items
resulted in more reliable estimations, greater criterion and construct validity, greater
explained variance, and consequently a better fit (Table V). Besides the better
quantitative results, the modified HEdPERF scale also had the advantage of being
more specific in areas that are important in evaluating service quality within higher
education sector. Hence, service quality in higher education can be considered as a
five-factor structure with conceptually clear and distinct dimensions namely
non-academic aspects, academic aspects, reputation, access and programme issues.
The current study also showed that SERVPERF performed miserably. Although
SERVPERF was developed and subsequently proven as the superior generic scale to
measure service quality in wide-ranging service industry, it did not provide a better
perspective for the higher education setting.

Criteria Modified HEdPERF SERVPERF HEdPERF-SERVPERF

Cronbach alpha (a) range 0.81-0.92 0.68-0.76 0.77-0.91

Criterion validity 0.58 0.27 0.53
Construct validity 0.57 0.34 0.57 Table V.
Adjusted R 2 0.34 0.23 0.30 Comparison of the
RMSEA 0.07 0.08 0.08 three scales
MIP Conclusions and implications
24,1 The objectives of this research were twofold. The primary goal was to test and
compare the relative efficacy of the three conceptualisations of service quality in order
to determine which instrument had the superior measurement capability in terms of
unidimensionality, reliability, validity and explained variance. The other objective is
concerned with enhancing the HEdPERF scale, thus transforming it into an ideal
44 measuring instrument of service quality for higher education sector. The tests were
conducted utilizing students sample from Malaysian tertiary institutions, and the
findings indicated that the three measuring scales did not perform equivalently in this
particular setting. In fact, the results led us to conclude that the measurement of service
quality by means of the HEdPERF method resulted in more reliable estimations,
greater criterion and construct validity, greater explained variance, and consequently
better fit than the other two instruments namely SERVPERF and
HEdPERF-SERVPERF. In short, the findings demonstrated an apparent superiority
of the modified five-factor structure of HEdPERF scale.
In what may be the first of its kind within higher education setting, the regression
analyses compared the three scales so as to determine how well they predicted service
quality level. Although the modified five-factor structure of HEdPERF clearly
outperformed the SERVPERF and the moderating HEdPERF-SERVPERF dimensions
in terms of explaining the variances in service quality level, the implications of these
findings are less clear. Since this study only examined the respective utilities of each
instrument within a single industry, any suggestion that the HEdPERF is generally
superior would still be premature. Nonetheless, the current findings do provide some
important insights into how these instruments of service quality compare with one
another.
The current results also suggest that the dimension access is the most important
determinant of service quality in higher education, thus reinforcing the
recommendation made by Firdaus (2005). In other words, students perceived access
to be more important than other dimensions in determining the quality of service they
received. As the only HEdPERF dimension to achieve significance, access is concerned
with such elements as approachability, ease of contact and availability of both the
academics and non-academics staff (Firdaus, 2005). Tertiary institutions should,
therefore, concentrate their efforts on the dimension perceived to be important rather
than focusing their energies on a number of different attributes, which they feel are
important determinants of service quality. While the idea of providing adequate
service on all dimensions may seem attractive to most service marketers and
managers, failure to prioritise these attributes may result in inefficient allocation of
resources. In conclusion, the current study provides empirical support in favour
of the idea that the modified five-factor structure of HEdPERF with 38 items
may be the superior instrument in measuring service quality within higher education
setting.

Limitations and suggestions for future research

The current study allows us to understand how three measuring instruments of service
quality compare to one another. To date, these scales have not been compared and
contrasted empirically thus making this research a unique contribution to the services
marketing literature. However, this work may pose more questions than it provides
answers. The present findings suggest that the modified HEdPERF scale is better Measuring
suited in higher education service settings. Caution is necessary in generalizing the service quality in
findings although considerable evidence of relative efficacy was found in the modified
HEdPERF scale. Given that the current study is limited to one service industry, this higher education
assertion would need to be validated by further research. Future studies should apply
the measurement instrument in other countries, in other industries, and with different
types of tertiary institutions in order to test whether the results obtained are general 45
and consistent across different samples.

References
Anderson, J.C. and Gerbing, D.W. (1991), “Predicting the performance of measures in a
confirmatory factor analysis with a pretest assessment of their substantive validities”,
Journal of Applied Psychology, Vol. 76 No. 5, pp. 732-40.
Babakus, E. and Boller, G.W. (1992), “An empirical assessment of the SERVQUAL scale”, Journal
of Business Research, Vol. 24 No. 3, pp. 253-68.
Bemowski, K. (1991), “Restoring the pillars of higher education”, Quality Progress, October,
pp. 37-42.
Bollen, K.A. (1989), Structural Equations with Latent Variables, Wiley, New York, NY.
Bolton, R.N. and Drew, J.H. (1991a), “A longitudinal analysis of the impact of service changes on
customer attitudes”, Journal of Marketing, Vol. 55, pp. 1-9.
Bolton, R.N. and Drew, J.H. (1991b), “A multi stage model of customer’s assessments of service
quality and value”, Journal of Consumer Research, Vol. 17, pp. 375-84.
Boulding, W., Kalra, A., Staelin, R. and Zeithaml, V.A. (1993), “A dynamic process model of
service quality: from expectations to behavioural intentions”, Journal of Marketing
Research, Vol. 30, pp. 7-27.
Brown, M.W. and Cudeck, R. (1993), “Alternative ways of assessing model fit”, in Bollen, K.A.
and Long, J.S. (Eds), Testing Structural Equation Models, Sage, Newbury Park, CA.
Bryne, B.M. (1994), Structural Equation Modelling with EQS and EQS/Windows-Basic Concepts,
Applications and Programming, Sage, Thousand Oaks, CA.
Bryne, B.M. (1998), Structural Equation Modeling with LISREL, PRELIS and SIMPLIS: Basic
Concepts, Applications and Programming, Lawrence Erlbaum Associates, Mahwah, NJ.
Carman, J.M. (1990), “Consumer perceptions of service quality: an assessment of the SERVQUAL
dimensions”, Journal of Retailing, Vol. 66, pp. 33-55.
Cattell, R.N. (1966), “The scree test for the number of factors”, Multivariate Behavioural Research,
Vol. 1, pp. 245-76.
Churchill, G.A. and Surprenant, C. (1982), “An investigation into the determinants of customer
satisfaction”, Journal of Marketing Research, Vol. 19, pp. 491-504.
Cronin, J.J. and Taylor, S.A. (1992), “Measuring service quality: reexamination and extension”,
Journal of Marketing, Vol. 56, pp. 55-68.
Diamantopoulos, A. and Siguaw, J.A. (2000), Introducing LISREL, Sage, London.
Eisen, S.V., Wilcox, M. and Leff, H.S. (1999), “Assessing behavioural health outcomes in
outpatient programs: reliability and validity of the BASIS-32”, Journal of Behavioural
Health Sciences & Research, Vol. 26 No. 4, pp. 5-17.
Finn, D.W. and Lamb, C.W. (1991), “An evaluation of the SERVQUAL scale in a retailing
setting”, in Holman, R. and Solomon, M.R. (Eds), Advances in Consumer Research,
Association for Consumer Research, Provo, UT, pp. 483-90.
MIP Firdaus, A. (2005), “The development of HEdPERF: a new measuring instrument of service
quality for higher education”, International Journal of Consumer Studies, online
24,1 publication, 20 October.
Franceschini, F. and Rossetto, S. (1997b), “On-line service quality control: the ‘Qualitometro’
method”, De Qualitac, Vol. 6 No. 1, pp. 43-57.
Ginsberg, M.B. (1991), Understanding Educational Reforms in Global Context: Economy, Ideology
46 and the State, Garland, New York, NY.
Hair, J.F. Jr, Anderson, R.E., Tatham, R.L. and Black, W.C. (1995), Multivariate Data Analysis with
Readings, Prentice-Hall, Englewood Cliffs, NJ.
Hattie, J. (1985), “Methodology review: assessing unidimensionality of tests and items”, Applied
Psychological Measurement, Vol. 9, pp. 139-64.
Hattie, J. (1990), “Performance indicators in education”, Australian Journal of Education, No. 3,
pp. 249-76.
Joreskog, K.G. and Sorbom, D. (1978), Analysis of Linear Structural Relationships by Method of
Maximum Likelihood, National Educational Resources, Chicago, IL.
Kaiser, H.F. (1970), “A second-generation little jiffy”, Psychometrika, Vol. 35, pp. 401-15.
Krejcie, R. and Morgan, D. (1970), “Determining sample size for research activities”, Educational
and Psychological Measurement, Vol. 30, pp. 607-10.
Lassar, W.M., Manolis, C. and Winsor, R.D. (2000), “Service quality perspective and satisfaction
in private banking”, Journal of Services Marketing, Vol. 14 No. 3, pp. 244-71.
Lawson, S.B. (1992), “Why restructure? An international survey of the roots of reform”, Journal of
Education Policy, Vol. 7, pp. 139-54.
Lewis, R.C. and Booms, B.H. (1983), “The marketing aspects of service quality”, in Berry, L.,
Shostack, G. and Upah, G. (Eds), Emerging Perspectives on Services Marketing, American
Marketing, Chicago, IL, pp. 99-107.
Llusar, J.C.B. and Zornoza, C.C. (2000), “Validity and reliability in perceived quality measurement
models: an empirical investigation in Spanish ceramic companies”, International Journal of
Quality & Reliability Management, Vol. 17 No. 8, pp. 899-918.
Mazis, M.B., Ahtola, O.T. and Klippel, R.E. (1975), “A comparison of four multi-attribute models
in the prediction of consumer attitudes”, Journal of Consumer Research, Vol. 2, pp. 38-52.
Mehta, S.C., Lalwani, A.K. and Han, S.L. (2000), “Service quality in retailing: relative efficiency of
alternative measurement scales for different product-service environments”, International
Journal of Retail & Distribution Management, Vol. 28 No. 2, pp. 62-72.
Nunnally, J.C. (1988), Psychometric Theory, McGraw-Hill, Englewood-Cliffs, NJ.
Oliver, R.L. (1989), “Processing of the satisfaction response in consumption: a suggested
framework and research propositions”, Journal of Consumer Satisfaction, Dissatisfaction,
and Complaining Behaviour, No. 2, pp. 1-16.
Owlia, M.S. and Aspinwall, E.M. (1997), “TQM in higher education – a review”, International
Journal of Quality & Reliability Management, Vol. 14 No. 5, pp. 527-43.
Parasuraman, A., Zeithaml, V.A. and Berry, L.L. (1985), “A conceptual model of service quality
and its implications for future research”, Journal of Marketing, No. 49, pp. 41-50.
Parasuraman, A., Zeithaml, V.A. and Berry, L.L. (1988), “SERVQUAL: a multiple-item scale for
measuring consumer perceptions of service quality”, Journal of Retailing, Vol. 64 No. 1,
pp. 12-40.
Parasuraman, A., Berry, L.L. and Zeithaml, V.A. (1991a), “Refinement and reassessment of the
SERVQUAL scale”, Journal of Retailing, Vol. 67 No. 4, pp. 420-50.
Parasuraman, A., Berry, L.L. and Zeithaml, V.A. (1991b), “More on improving service quality Measuring
measurement”, Journal of Retailing, Vol. 69 No. 1, pp. 140-7.
Parasuraman, A., Zeithaml, V.A. and Berry, L.L. (1994), “Reassessment of expectations as a
service quality in
comparison standard in measuring service quality: implications for future research”, higher education
Journal of Marketing, Vol. 58, pp. 111-24.
Quester, P., Wilkinson, J.W. and Romaniuk, S. (1995), “A test of four service quality measurement
scales: the case of the Australian advertising industry”, Working Paper 39, Centre de 47
Recherche et d’Etudes Appliquees, Group esc Nantes Atlantique, Graduate School of
Management, Nantes.
Robledo, M.A. (2001), “Measuring and managing service quality: integrating customer
expectations”, Managing Service Quality, Vol. 11 No. 1, pp. 22-31.
Soutar, G. and McNeil, M. (1996), “Measuring service quality in a tertiary institution”, Journal of
Educational Administration, Vol. 34 No. 1, pp. 72-82.
Sureshchandar, G.S., Rajendran, C. and Anantharaman, R.N. (2002), “Determinants of
customer-perceived service quality: a confirmatory factor analysis approach”, Journal of
Services Marketing, Vol. 16 No. 1, pp. 9-34.
Teas, R.K. (1993a), “Expectations, performance evaluation, and consumers’ perceptions of
quality”, Journal of Marketing, Vol. 57 No. 4, pp. 18-34.
Teas, R.K. (1993b), “Consumer expectations and the measurement of perceived service quality”,
Journal of Professional Services Marketing, Vol. 8 No. 2, pp. 33-54.
Zeithaml, V.A. and Bitner, M.J. (1996), Services Marketing, McGraw-Hill, Singapore.

Further reading
Crawford, F. (1991), Total Quality Management, Committee of Vice-Chancellors and Principals
Occasional Paper, London, December.

To purchase reprints of this article please e-mail: [email protected]

Or visit our web site for further details: www.emeraldinsight.com/reprints