Volume 6, Issue 8, August – 2021 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165

Quality Perception -
A Hierarchical Modelling Approach
* Dr Juhi Kamakoty, Dr. Sandeep Singh
* Acropolis Faculty of Management & Research, Indore

Abstract:- SERVQUAL has been widely studied across may be accrued to the systematic evolution of research in
various industries including health care sector by this area is the well established link between service quality
researchers world wide. Most of these studies have used and business performance
general linear model to draw their findings. However
using GLM to study service quality in various hospitals II. LITERATURE REVIEW
amounts to shear violation of the basic assumption of
independence of observations as the data are nested and Service quality is an assessment of how well the
clustering in data could be easily observed which was service delivered meets the client’s expectation. It may
ignored by researchers. This study seeks to plug this means the quality of deliverables to some, while to some
loophole by using Hierarchical modeling on R software. others it may means the human interactions involved at the
Data from 250 patients was collected from Big and service encounter yet to others it may mean the experience
Small hospitals from the commercial capital of central involved at the time of availing the service. The quality of
India and analyzed. They showed the existence of 5 service is thus a relative concept having different meaning at
factors quiet in accordance with Parasuraman (1988) different times.With the advent of globalization service
findings. There after hierarchical modeling was applied quality has all of a sudden caught attention of both
with these factors as dependent variables. Findings researchers as well as practitioners (Abdullah, 2005). In
showed that on Tangibility there is no significant order to survive in business, it is service quality alone that
difference in perception across gender, age, annual can give a competitive edge to business and thus ensure
income and education qualification. Findings further survival (Bitner, 1993). Services have its own idiosyncratic
shows that Reliability, Assurance, Responsiveness and characteristics and are therefore difficult to design, measure
Empathy varies across gender. and maintain. Services are heterogeneity,perishable
intangibles and production & delivery of the services are
Keywords:- Servqual, Service, Quality, Health Care, concurrent. Moreover the outcome quality is contingent to
Regression, Clustering , Hierarchical Modeling. the situational inputs. The quality of service is thus an
abstruse concept that takes on different meaning depending
I. INTRODUCTION on the inputs and the service partners.

The gradual shift of economy from manufacturing to The latent concept of service quality may be measured
service base and the fight for survival in razor sharp by measuring the factors or determinants of service quality.
competition all of a sudden brought quality of service in The determinants may be technical, functional, process
focus . In the seminal works of Parsuraman et. related, human related, behavioral constructs , image , IT
al.(1985,1988) in the decades of 1980’s service quality has based , experiential and so on (Parsuraman et al., 1985;
been conceptualized, defined and measured using the gap Groonos, 1984; Cronin & Taylor, 1992; Mattson, 1992;
model. Service quality gap is computed as the difference Teas, 1993). The literature on service quality is flooded with
between percepti on and expectation. Servperf (Cronin & various frameworks and models to asses service quality of
Taylor, 1992) is another such universal construct that various services like telecom, healthcare, hospitality, public
measures the service quality in various service settings but services, aviation, education , tourism, banking
unlike the Servqual gap model, this captures service quality etc.(Nwabueze and Mileski,2008; Rhee and Rha,2009;
perception only.Servqual and Servperf the two scales that Shamdasani et al.,2008; Sunindijo et al., 2014;Kamakoty et
dominate service quality researches, as they are simple and al., 2015 ). These models largely measure the service
easy to administer and requires minor semantic modification quality from the service user’s perspective or service
to customize it to industry specific settings. Service quality providers perspective and offers useful insights about
has always been a fancy amongst the researchers and the enhancing service quality by judiciously using business
literature has been replete with service quality frameworks, resources. The most popular instruments used for
models, factors, constructs, antecedents, consequents, measuring the service quality are Servqual and Servperf
determinants etc. Research in the area of service quality has scale or customized variants of these scales . Table 1
given important insights to managers for improving quality given below summarizes Servqual based research on
by judiciously using their resources. An obvious reason that service quality.

IJISRT21AUG694 www.ijisrt.com 471

Volume 6, Issue 8, August – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
Table 1 : Research in Service Quality ( Kamakoty & Sohani, 2011)
Industry SERVQUAL based Research work
Banking Rahman, 2005; Balaji and Ganesan,2005 ; Davila et al.,2010 , Newman, K. (2001)
Restaurant/Fast Food/ Balaji and Ganesan, 2005 , Chowdhary and Prakash,2007
Hospitality
Hospital/Health (Nwabueze and Mileski,2008; Chowdhary and Prakash,2007
Care/Fitness Centres
Consultancy Viadiu et al., 2002
Public Services Foster and Newman,1998
Education Balaji and Ganesan 2005

Airlines Balaji and Ganesan 2005

IT/Telecomm. Balaji and Ganesan 2005, Zhou et al.(2002)

Service quality is as important in health care industry questionnaire and the scaled were easily understood by
as it is any where else. Quality of health services may be responders who were either patients or their attendants.
defined as the value that is being offered to the customer in
terms of quality of life, life expectancy , cure, prevention The data was personally collected by the researchers
etc. While quantitatively it can be measured in terms of with the help of well structured questionnaire after a detailed
counts, expectancies, reduction, risk factor etc ., debriefing of the respondent patients/attendants’ in
researchers have occupied themselves in developing hospitals. Five hospitals were selected for data collectiobn.
constructs to measure the qualitative aspects of services. The criteria for selection was as medical facilities offered,
There is a rich repository of literature where in one can find target segment and accessibility. A sample of 250 responses
various scholarly work to frame, measure, factorize was collected using convenience sampling. Data analysis
quality of services. The universal gap based Servqual model was done using R s/w. Intially the data was checked for
is by and large the most popular tool elicited in literature to anomalitie and missing values. Once the data was clean, it
measure the service quality of medical services, health was put to factor analysis. The extracted factors were
centres, hospitals across various countries/ communities by analyzed across demographic variables. The hospitals were
students, medical staff, research scholars etc. Griffith classified into two categories Big and Small on the basis of
(2002), Lehtinen and Lehtinen (1985),. John (1989), Bed capacity. Two hospitals qualified in Big category out of
Reidenbasch and Sandifer (1990) Babakus and Glynn 5. ‘t’ test and anova analysis was applied to study the
(1992), Bowers et. al. (1994), Youssef & Nel (1996), and perception expectation gap in the five extracted service
Lim and Tang (2000) Jabnoun and Chaker (2003). quality factors across gender, age, education, income . The
findings were then reported along with managerial
This paper attempts to study the service quality of implications.
hospitals of commercial capital of central India on the basis
of size. The data has been collected from 250 pateints IV. DATA ANALYSIS
dispersed across big abd small hospital. The patients are
nested in the hospital so hierarchical modeling had beed Total sample size was 250, out of which 46 were
used with patients at level at level 1 and hospital at level 2. missing values which were eventually imputed using
Servqual perception only scale is used to capture the data regression method. Out of 250 respondents 52.8% were
across the dichotomy of big and small hospitals. The data is male and 47.2% were female. About 40% of the respondents
then subjected to analysis to develop important insights on were below 30 years, 17% between 30-40 years, 14.4%
health care services. between 40-50 years and 28% were above 50 years.
Qualification wise 62.4% were undergraduates, 31.6% were
III. RESEARCH METHODOLOGY graduates and 6% were post graduates. On Income basis
58.8% were below 1 lacs, 36.4 were between 1 – 2.5 lacs,
A through secondary research had been conducted to and only 4.8% were above 2.5 lacs per year.
identify the gap in literature. Various journals, news,
reports, conference papers had been studied to get useful Gender is coded as 1 for male and 2 for female. Age is
insights on the existing body of knowledge on quality of coded as 1 for patients below 30 years of age, 2 for patients
medical services. Once the gap had been identified , It was between 30 – 40 years, 3 for patients between 40 -50 years
decided to measure service quality across hospitals by and 4 for patients above 50 years of age. Qualification is
employing Servqual perception only scale. Sevqual scale coded as 1 for undergraduates, 2 for Graduates and 3 for
had been customized across health care setting to capture the post graduates. Coding of annual income is 1 for income
data from patients. or their attendants on a seven point less than 1 lacs, 2 for income between 1 – 2.5 lacs and 3 for
Likert scale (strongly disagree to strongly agree) along with annual income above 2.5 lacs.
demographic variables like age, income, gender etc . A pilot
study was undertaken in order to determine whether the Before subjecting the data for Factor Analysis, Bartlett
test was carried out to assess the factorability of the data.

IJISRT21AUG694 www.ijisrt.com 472

Volume 6, Issue 8, August – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
The Kaiser-Meyer-Olkin Measure of Sampling Adequacy is Adequacy of 0.85 (Chi-Square of 2756.602) clearly
a statistic that indicates the proportion of variance in your indicating that data if fit for factorization.
variables that might be caused by underlying factors. : KMO
values between 0.8 and 1 indicate the sampling is adequate. Scree Plot shown below highlights that 5 factors
KMO values less than 0.6 indicate the sampling is not explain most of the variance, which is in accordance with
adequate and that remedial action should be taken. Output Parasuraman findings. These 5 factors accounts for 69.44%
showed Kaiser-Meyer-Olkin Measure of Sampling of variance in data.

Fig.1 . Scree Plot

Table 2 :Variance of Factors Extracted

S. No Factor Extracted Variance %
1 Tangibility 25.8%
2 Reliability 22.5%
3 Responsiveness 8.09%
4 Assurance 7.34%
5 Empathy 5.60%
Cummulative Variance Explained 69.44%

Factor Analysis showed that Item 1 through 4 loaded on Factor 1 referred to as Tangibility, items 5 through 9 loaded on
Factor 2 referred to as Reliability, items 10 through 13 loaded on factor 3 referred as Responsiveness, item 14 through 17 loaded
on factor 4 referred as Assurance and lastly items 18 through 22 loaded on 5th factor called as Empathy. Reliability analysis
carried out showed all factors have Cronbach Alpha of above .70, which is quiet in keeping with international standards, therefore
a summated scale was created for all Factors and named accordingly for both Perception and Expectation.

Table 3: Reliability of Factors Extracted

S. No Factor Cronbach Alpha
1 Tangibility 0.849
2 Reliability 0.829
3 Responsiveness 0.699
4 Assurance 0.840
5 Empathy 0.85

Gap was studied by taking the difference between Perception and Expectation summated scale. Positive value indicated
patients perception about hospital services are higher than what they expected and negative value showed just the opposite.
Thereafter descriptive statistics was carried out for the acquired factors as shown below in table. Output in the table clearly shows
that data is normally distributed with Skewness near zero and Kurtosis well below 3.0 (with low standard errors)

IJISRT21AUG694 www.ijisrt.com 473

Volume 6, Issue 8, August – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165

K
u
r
t
o
s
i
s

Square of Skewness

Fig 2: Showing distribution of dependent varaible

Above graph shows that the distribution of dependent y suggesting female are relatively more disgruntled with the
variables (RATER) show Beta distribution hospital empathy than male.

For data analysis, we have used Hierarchical Modeling On running the script for Tangibility as a response
using Generalized Linear Mixed Effect Model using R variable the output showed the evidence of Inter Class
software using packages like ‘lmer4’, ‘merTools’ and Correlation of 0.03 which is evidence of clustering (though
‘fitdistrplus’. The last on is used to asses distribution of weak) in the data for Big and Small hospitals, thereby
dependent variables. lending justification to the application of Hierarchical
Modeling. The output for Random Effect is shown as below
We have used various factors of SERVQUAL
perception scale as dependent variable and demographic Random effects:
descriptors – age, gender, annual income, and education Groups Name Variance Std.Dev.
qualification as predictors. Further we have divided the Big- Small (Intercept) 0.0069 0.0861
hospitals as Big – Small depending on bed capacity and this Residual 0.855 0.92
has been used for capturing random effect.
For Fixed Effect none of the values are significant at 5
On running the script for Empathy as a response % significance level. Findings shows that cutting across pati
variable the output showed the evidence of Inter Class ents of all age, gender, education qualification and annual in
Correlation of 0.07 which is evidence of clustering (though come there is no difference in the hospitals tangibility aspect
weak) in the data for Big and Small hospitals, thereby .
lending justification to the application of Hierarchical
Modeling. The output for Random Effect is shown as below On running the script for Responsiveness as a response
variable the output showed the evidence of Inter Class
Random effects: Correlation of 0.11 which is strong evidence of clustering in
Groups Name Variance Std.Dev. the data for Big and Small hospitals, thereby lending full
Big- Small (Intercept) 0.01853 0.1361 justification to the application of Hierarchical Modeling.
Residual 1.74854 1.3223 The output for Random Effect is shown as below’

For Fixed Effect Age (β= -0.165, t-value = -2.32) and Random effects:
Gender (β= -0.368, t-value = -2.214) have significant effect Groups Name Variance Std. Dev.
on empathy. Highly significant Intercept ((β= 5.95, t-value = Big- Small (Intercept) 0.07853 0.280
13.72). Findings show that as the patients’ age goes up his p Residual 0.299 1.731
erception goes down by 16% thus patients with higher age a
re more dissatisfied with the empathy of the hospital. On ge For Fixed Effect Gender (β= -0.553, t-value = -2.54) a
nder female perception is about 36% lesser than male, clearl nd annual income (β= -0.892, t-value = -3.62) have signific
ant effect on Responsiveness and age in nearly significant A

IJISRT21AUG694 www.ijisrt.com 474

Volume 6, Issue 8, August – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
ge (β= -0.187, t-value = -2.00). Findings show that female p ents age goes up his perception about hospitals assurance de
erception of responsiveness is 55% lower than male. Also as creases by 14%. This shows that old patients find hospitals l
the patients income goes up his perception on hospitals resp ess assuring than younger counterparts.
onsiveness goes down by 89%. dissatisfied with the empath
y of the hospital. On gender female perception is about 18% On running the script for Reliability as a response
lesser than male (though marginally significant), clearly sug variable the output showed the evidence of Inter Class
gesting female are relatively more disgruntled with the hospi Correlation of 0.22 which is strong evidence of clustering in
tal responsiveness than male. the data for Big and Small hospitals, thereby lending full
justification to the application of Hierarchical Modeling.
On running the script for Assurance as a response The output for Random Effect is shown as below
variable the output showed the evidence of Inter Class
Correlation of 0.12 which is strong evidence of clustering in Random effects:
the data for Big and Small hospitals, thereby lending full Groups Name Variance Std. Dev.
justification to the application of Hierarchical Modeling. Big- Small (Intercept) 0.643 0.801
The output for Random Effect is shown as below’ Residual 4.914 2.216

Random effects: For Fixed Effect Age (β= 0.311, t-value = -2.59) other
Groups Name Variance Std. Dev. predictors are insignificant. Findings show that as the patient
Big- Small (Intercept) 0.08 0.295 s age goes up his perception about hospitals reliability increa
Residual 1.129 1.062 ses by 31%. This shows that young patients find hospitals le
ss reliable than their younger counterparts.
For Fixed Effect Age (β= -0.144, t-value = -2.517) oth
er predictors are insignificant. Findings show that as the pati

Table 4: Depicting Model fit

S. No Dependent AIC BIC LogLik Deviance χ2 P Value
Variable
1 Empathy 864.84 889.49 -425.42 850.84 - -
2 Tangibility 685.2 710.37 -335.86 671.72 179.11 0
3 Responsiveness 1000.58 1025.23 -493.29 986.58 0.000 0
4 Assurance 758.59 783.24 -372.29 744.59 241.99 0
5 Responsiveness 1127.17 1151.82 -556.58 1113.17 0.00 0

Table above shows the model fit for various SERVQUAL factors. It can be clearly seen that model No 2 with Tangibility as
dependent variable has lowest AIC (Akaike's Information Criteria) and BIC (Bayesian Information Criteria).

R
E
S
I
D
U
A
L
S

FITTED VALUES
Fig 3: Residual Plot

Figure above shows the residual plot against the fitted values a random pattern which is clearly indicative of a good fit.

IJISRT21AUG694 www.ijisrt.com 475

Volume 6, Issue 8, August – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
V. DISCUSSION [5]. Bowers, M.R. Swan, J.E, Hoehler & Willaim.F.
(1994) “What attributes determine quality and
Tangibility aspect of service quality is insignificant satisfaction with healthcare services?”, Healthcare
across age, income, qualification and gender implies that the Management Review, Vol.19, pp. 40-49
customers look for higher level of tangibility aspects then [6]. Chowdhary, N and Prakash, M. (2007), “Prioritizing
what is offered by the hospitals. They know look beyond the service quality dimensions, Managing Service Quality,
basic medical infrastructure and expect improved Vol.17 No. 5, pp.493-509
hospitality. Increase in perception of Reliability with age [7]. Cronin, J. J. Jr. and Taylor, S. A. (1992), “Measuring
implies that hospitals need to craft focused services for the service quality: a re-examination and extension”,
patients of younger age. Assurance is an aspect of service Journal of Marketing, Vol. 6, pp.-55-68
quality to be built in the senior generations. This can be [8]. Davila, J. A. M., Garcia, L.C., Valdunciel, L. and
done by robust selection, training, induction of hospital Florez, M. (2010), “Operations in Banking: the service
staff and medicos. Additionally Assurance can be built quality and effects on satisfaction and banking “, The
through proper follow ups and courtesy calls. Empathy is Service Industries Journal, Vol. 30 No. 13, pp. 2163-
seen to be effected by age. Senior patients are not very 2182
happy in this context. Healthcare providers should take a [9]. Foster, G. and Newman, K. (1998), “What is service
serious note of they represent the major chunk of their quality when service equals regulation?” The Service
clientele. Empathy can be imbibed through proper Industries Journal , Vol. 18 No. 4, pp. 51-65
communication and quick response to patients ailment & [10]. Griffith, J. R, Alexander, J. A; Warden; G. L. (2002) ,
complains. It calls for sensitization of staff towards handling “ Measuring comparative hospital performance /
patients. This also applies for handling the female gender Practitioner response”) , Journal of Healthcare
patients with more of sensitivity and patience as the Management; Vol. 47.No . 1, pp. 41-57.
evidences show that females rate the hospitals lower on this [11]. Groonos, C. (1984), “A Service Quality Model and it’s
factor. Females are not happy on Responsiveness aspect of Marketing implications”, European journal of
service quality of hospitals. They look for timely responses Marketing, Vol. 28 No. 4, pp. 36-44
and prompt actions. The hospitals need to work on this [12]. Jabnoun, N. and Chaker, M. (2003), "Comparing the
aspect by ensuring the availability of staff, facilities and quality of private and public hospitals", Managing
services as an when required. Similarly people of higher Service Quality: An International Journal, Vol. 13 No.
income brackets also have an urge for more 4, pp. 290-299.
Repsonsiveness. [13]. John, Joby   1989 . Perceived quality in health care
service consumption: What are the structural
VI. CONCLUSION dimensions? In Jon M. Hawes and John Thanopoulos
(eds), Developments in marketing services, (pp. 518–
It can be concluded from this study that hospital lack 521). Orlando, FL: Academy of Marketing Science
on certain aspects of service quality. Specially the senior [14]. Kamakoty, J., Sohani, N. and Sohani, N. (2015)
patients and the female patients look for more of empathy, ‘Determinants of service quality in education: service
responsiveness and assurance. The hospitals should provider’s perspective and academician’s perspective’,
seriously craft solutions to cater to their requirements as Int. J. of Service and Operations Management, Vol.
these constitute the chunk of hospital clientele. Tangibility is 20, No. 2, pp. 141-164)
another aspect that the hospitals have to look beyond the [15]. Lim,C. P. and Tang, N.K.H. (2000), "A study of
normal. Apart from robust medical infrastructure, the patients’ expectations and satisfaction in Singapore
patients look for tangiblization of the service delivery hospitals", International Journal of Health Care
process. Quality Assurance, Vol. 13 No. 7, pp. 290-299
[16]. Mattsson, J. (1992), “A service quality model based on
REFERENCES ideal value standard”, International journal of service
industry, Vol. 3 No. 3, pp. 18-33
[17]. Newman, K. (2001), “Interrogating SERVQUAL: a
[1]. Abdullah, F. (2005), “Hedperf versus Serfperf”, critical assessment of service quality measurement in
Quality Assurance in Education, Vol 13 No 4, pp. 305- high street retail bank”, International Journal of Bank
328 Marketing, Vol. 19 No. 3, pp. 129-139
[2]. Babakus.E and Glynn M.W. (1992) “Adopting the [18]. Nwabueze, U. and Mileski, J.P. (2008), “The three
SERVQUAL scale to hospital services: An empirical dimensions of Quality Service; The case of service
investigation”, Health Services Research, Vol. 26 quality gaps in UK National Health Service”,
No.6, pp. 767-786. International Journal of Public Administration, Vol.
[3]. Balaji, B. and Ganesan, S. (2005), “Integrating people, 31, pp. 1328-1353
technology and strategy: quality dimensions in select [19]. Parsuraman, A., Zeithmal, V. A. and Berry, L. L.
services”, ICFAI Journal of Marketing Management, (1985), “A conceptual model of service quality and its
Feb.,2005,pp. 07-23 implications for future research”, Journal of
[4]. Bitner, M.J. (1993), “Tracking the evolution of the Marketing, Vol .49 No. 3, pp. 41-50
service marketing literature”, Journal of Retailing,
Vol. 69, pp. 61-103

IJISRT21AUG694 www.ijisrt.com 476

Volume 6, Issue 8, August – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
[20]. Parsuraman, A., Zeithmal, V. A. and Berry, L. L.
(1988), “SERVQUAL: A multiple item scale for
measuring consumer perception of service quality”,
Journal of Retailing, Vol 6.4 No. 1, pp. 12-37
[21]. Rahman, Z. (2005), “Service quality: Gaps in the
Indian Banking Industry”, The ICFAI journal of
Marketing Management, Feb., 2005, pp. 37-47
[22]. Redienbach,E and Sandifer, S.B. (1990) “Exploring
perceptions of hospital operations by a modified
SEVQUAL approach” Journal of Healthcare
marketing, Vol. 49 pp.441-450.
[23]. Rhee, S. and Rha, J. 2009, “Public service quality and
customer satisfaction: exploring the attributes of
service quality in the public sector”, The Service
Industries Journal, Vol. 29 No. 11, pp. 1491-1512
[24]. Shamdasani, P., Mukherjee, A., Malhotra, N. (2008),
“Antecedents and consequences of service quality in
consumer evaluation of self-service internet
technologies”, The service Industries Journal, Vol. 28
No. 1, pp. 117-138
[25]. Teas, K. R., (1993), “Expectations, performance
evaluation and consumers’ perception of quality”,
Journal of Marketing, Vol. 57, pp. 18-34
[26]. Viadiu, F. M., Fa, M. C. & Saizarbitoria, I. H. 2002),
“Do quality consultants offer a quality service”, Total
Quality Management, Vol. 13 No. 6, pp. 797-811
[27]. Youssef, F. and Nel, D. (1996), “Healthcare quality in
NHS Hospitals” International Journal of healthcare
Quality assurance, Vol. 9 No. 1, pp. 15-28.
[28]. Zhou, T., Lu, Y., Wang, B.,(2002) “The relative
importance of website design quality and service
quality in determining consumers’ online repurchase
behavior”, Information Systems Management, Vol.
26, pp. 327-337

IJISRT21AUG694 www.ijisrt.com 477