The human element in airline service quality

job performance

The human element in airline service quality: contact personnel and the customer

Sunil Babbar Department of Information Technology & Operations Management, Barry Kaye College of Business, Florida Atlantic University,

Boca Raton, Florida, USA, and Xenophon Koufteros Information & Operations Management, Mays Business School, Texas A&M University, College Station, Texas, USA

AbstractPurpose – The purpose of this paper is to examine empirically the dimension of personal touch and its elements of individual attention, helpfulness, courtesy, and promptness as determinants of customer satisfaction for passenger airlines.

Design/methodology/approach – Survey data from 437 individuals and a hierarchical approach to structural equation modeling are used to systematically evaluate four alternative measurement models A second-order measurement model of personal touch appeared to represent the data very well and can be supported from a theoretical point of view.

Findings – Personal touch is found to statistically and substantively affect passenger satisfaction explaining about 54 percent of the variance In other words, collectively, individual attention, helpfulness, courtesy, and promptness are found to have a significant effect on airline passenger satisfaction Research limitations/implications – This research examines the satisfaction of customers of US passenger airlines Future research should overcome this limitation by extending the survey to include experiences on international flights and with non-US airlines The results are biased more towards the responses of “younger” passengers and those who flew primarily in the economy level of service Practical implications – The findings of this research have important strategic and managerial implications for passenger airlines and serve to validate the corporate culture and customer service quality driven models of exemplary airlines such Southwest Airlines, JetBlue Airways, and Scandinavian Air Systems.

Originality/value – The paper provides a very thorough review of the literature and is the first to examine empirically the affect of personal touch displayed by contact employees on the satisfaction of customers of passenger airlines.

Keywords Customer relations, Customer satisfaction, Customer services quality, Airlines, Interpersonal skills, United States of America

Paper type Research paper

IntroductionThe number of passenger boardings on flights of US airlines has risen from about

452 million in 1991 to more than 738 million in 2005 ( Bureau of TransportationStatistics, 2006) US passenger airlines currently employ a combined 405,000 full-timeequivalent employees Many of these employees are customer contact employees whoshape the experience of airline passengers through the quality of service they provide

Interactions between customers and the employees of an airline probably influence the

customers’ perceptions of the airline (Gursoy et al., 2005) Interestingly, because service

quality is more visible, passengers may use service quality as a basis for judging

the overall quality of an airline (Rhoades and Waguespack, 1999) Most visible to

service customers is the service that contact employees provide

Since the 1980s, customer contact, generally defined as the interface between the

customer and the service provider, has been recognized as a key difference between

services and manufacturing operations (Chase, 1981; Soteriou and Chase, 1998) In

keeping with the recent definition of Zomerdijk and Vries (2007), we define customer

contact as a direct encounter between a customer and a service provider that takes

place at the same time but not necessarily in the same place, and has the potential for

interaction between the two A contact employee is someone who engages in such

encounters with customers as part of the service delivery process

An astute understanding of the critical role customer contact employees play in

shaping the experience of customers can be invaluable to service providers In this study,

we empirically examine the impact of service provided by customer contact personnel of

US passenger airlines on customer satisfaction with the airline In particular, we identify

human- or people-related factors; individual attention, helpfulness, courtesy, and

promptness that shape the element of “personal touch” in the service provided by contact

employees and examine their role as determinants of customer satisfaction with the

airline

In the following sections, we review the literature, state our hypothesis, outline

systematically the research methods we employ, present our findings, discuss

managerial implications, and identify limitations of our research

Literature review

There are indeed fundamental differences between goods and services With goods

being tangible, their designers are able to utilize well-established physical principles to

understand and shape their performance (Taura and Yoshikawa, 1994) Services, on the

other hand, are typically intangible and involve relatively greater customer contact

(Soteriou and Chase, 1998) Given the intangible nature of services, it becomes hard to set

standards (Harvey, 1998) While the question of how best to delight customers continues

to challenge both manufacturing and service organizations, the building blocks of

service and their relationship to service quality are less obvious than for manufactured

goods (Soteriou and Chase, 1998) It is generally recognized that the “softer” or intangible

service component is harder to design and manage (Vandermerwe, 1994) Yet, more often

than not, it is this soft component that tends to carry a heavier weight in the customers’

assessment of their quality of experience with a firm’s product Because of the intangible

nature of service, how a service is delivered becomes an important determinant of

service quality (Gro¨nroos, 1984) This attests to the importance of the role of contact

employees and the need to better understand and measure the extent of their influence on

customer satisfaction

Customer contact is a key consideration in the design and delivery of

services including airline service The roots of thinking on customer contact in the

context of service delivery can be traced back to Whyte’s (1946) published essay

“When workers and customers meet” Since that time, many have noted the

relevance of customer contact as a key differentiator of services and manufacturing

Airline service

quality

805

(Chase, 1978, 1981; Parasuraman et al., 1985; Kellogg and Chase, 1995; Harvey, 1998;Prajogo, 2006) and it is now an important dimension in most service taxonomies(Mersha, 1990; Soteriou and Chase, 1998).

Mills et al (1983) consider the interface between the firm and the customer to be aseminal element of consideration for service firms Such an interface has also beenreferred to as “contact theory” in the literature (Kellogg and Chase, 1995) with anumber of frameworks having incorporated the concept of customer contact (Mills andMargulies, 1980; Schmenner, 1986) Contact with the customer providesservice-delivering firms with valuable opportunities for responding to the needs ofcustomers while simultaneously marketing their products to them (Chase et al., 1984;Delene and Lyth, 1989; Chase and Hayes, 1991) It also gives contact employees theopportunity to build customers’ trust and confidence and to help sustain a relationshipwith them well into the future (Mills and Margulies, 1980)

A review of the literature reveals a progression of thinking on what constitutescustomer contact Early on, customer contact was defined simply as the physicalpresence of the customer in the system (Chase, 1978) Later, Schmenner (1986) definedcustomer contact as an interaction between the customer and service provider andcustomization of the service that places demands on the design of the service deliverysystem More recently, Mersha (1990) has defined customer contact as customerpresence, either face-to-face or else mediated through technology, with low- orhigh-levels of interaction between the customer and service provider Zomerdijk andVries (2007) echo Mersha’s sentiment in defining customer contact as a directencounter between a customer and a service provider that takes place at the same timebut not necessarily in the same place, and has the opportunity for interaction.The importance of the role of customer contact employees in creating and providinghigh-quality service has been underscored by numerous researchers (Zeithaml et al.,1985; Hartline and Ferrell, 1996; Cook et al., 2002) Jan Carlzon, responsible for turningScandinavian Air Systems (SAS) into a customer-driven company, first introduced theconcept he referred to as “moment of truth” (Peters and Austin, 1985) Carlzon defines amoment of truth as “an episode in which a customer comes into contact with any aspect

of the company, however remote, and thereby has an opportunity to form animpression” (Collier, 1994) In his book, Moments of Truth, Carlzon (1987, p 3) writes,

“Last year each of our ten million customers came in contact with approximately fiveSAS employees”:

These 50 million “moments of truth” are the moments that ultimately determine whether SASwill succeed or fail as a company They are the moments when we must prove to ourcustomers that SAS is their best alternative

By homing in on these moments of truth and changing the company’s outlook from one

of flying “airplanes” to one of flying “people”, he brought to the forefront the criticalrole which contact employees play in determining customer satisfaction Themanagement of moments of truth can indeed shape the quality of service rendered(Gro¨nroos, 1990) Behaviors and attitudes of contact employees can significantly shapethe customers’ experience with the service provided and their assessment of its quality(Parasuraman et al., 1985, 1988; Haywood-Farmer, 1988; King and Garey, 1997;Winsted, 2000; Yoon et al., 2001) It has been suggested that, in the case of services,employees represent the firm and define its product (Shostack, 1977), and actions

IJOPM

28,9

806

which the employees take can shape the customers’ perceptions of the firm itself

(Zeithaml and Bitner, 2000)

The affect theory of social exchange (Lawler, 2001), which posits relationships as a

source of emotions, attests to the important role which contact employees play in

shaping the customers’ assessment of the service firm itself As Sierra and McQuitty

(2005) explain, this theory purports that emotions are directed at the group context and

not limited to the service agent Accordingly, a favorable encounter with a service

employee positively impacts a customer’s impression of the entire service firm A

customer’s emotional response to a service can often be attributed to the customer’s

experience with the service employees (van Dolen et al., 2001) and can influence the

customer’s future purchase intentions (Berry, 2000) Service encounters resulting in

positive customer emotions increase the probability of customer loyalty and repeat

purchase (Czepiel, 1990; Lawler et al., 2000)

Customers have been found to evaluate quality of service based on the level of

concern and civility (Winsted, 2000) and listening and understanding demonstrated by

employees (Chandon et al., 1997) Further, individual attention (Vandermerwe, 1994),

cheerfulness (Harvey, 1998), friendliness (Brown and Sulzer-Azaroff, 1994)

and courtesy (Goodwin and Smith, 1990) have also been recognized as determinants

of customer satisfaction with the service experience King and Garey (1997) have

examined the concept of relational quality in service encounters According to them,

relational quality refers to customers’ perceptions and evaluations of service

employees’ communications and behaviors such as respect, courtesy, warmth,

empathy and helpfulness, and involves customers’ feelings and emotional states

through interactions with employees

Research hypothesis

Despite the significance of the service the airline industry provides, there is limited

empirical research and theory development in the area of airline service quality Often,

data from the US Department of Transportation is used as a basis for commentary

about service quality and the relative positions of airlines (Rhoades et al., 1998;

Rhoades and Waguespack, 1999; Gursoy et al., 2005; Bowen and Headley, 2006)

A number of authors have presented conceptual commentary on the culture of

particular airlines (Ekdahl et al., 1999; Laszlo, 1999; Ford, 2004; Smith, 2004) Others

have examined service breakdowns through critical incidents (Edvardsson, 1992) and

service failure (Bejou and Palmer, 1998) in airlines and presented frameworks for

managing the air travel experience (Carlzon, 1987; Le Bel, 2005)

In general, US airlines are found wanting in the quality of service they render to

their customers (Oneal, 1991; McClenahen, 1991; Rhoades et al., 1998) It thus comes as

no surprise that customer satisfaction with US passenger airlines is extremely low

even today (Yu, 2007) As noted earlier, customer contact employees can play an

important role in shaping customer satisfaction Despite this, the nature of the

relationship between service elements of individual attention, helpfulness, courtesy,

and promptness of contact employees and the satisfaction of airline customers remains

to be empirically examined

Our extensive review of the literature provides the theoretical underpinning for

the relationship we hypothesize and empirically examine Accordingly, we state the

following:

Airline service

quality

807

H1 Individual attention, helpfulness, courtesy, and promptness of contactemployees constitute elements of “personal touch” and positively affect thelevel of satisfaction of airline customers.

Research methods

In order to test our hypothesis, we sought empirical evidence that supports the positedrelationship To gain such empirical evidence we first took measurements for eachconstruct

Instrument development

To develop our measurement instrument, we reviewed the extensive literature on servicequality Such literature spans many disciplines such as marketing, managementinformation systems, operations management, hospitality management, and health caremanagement It is quite clear that the literature is influenced by the seminal work onSERVQUAL and its subsequent refinements The SERVQUAL instrument (Parasuraman

et al., 1988, 1991) separately measures the expected level of service and the experiencedlevel of service Service quality scores are based on the difference between the twomeasures While the SERVQUAL instrument has been used extensively, its tenure hasbeen controversial and has lead to modifications and the emergence of other competinginstruments (SERVPERF – Cronin and Taylor, 1992; TOPSIS – Mukherjee and Nath,2005) van Dyke et al (1997) and van Dyke et al (1999) have done an excellent job insummarizing some of the difficulties surrounding SERVQUAL and so we refrain fromengaging in a similar debate of the consequential issues pertaining to SERVQUAL’sperformance Given some of the insights the SERVQUAL instrument provides, we used it

as one basis to draw and adapt items from, in formulating our constructs We concentratehere on the interpersonal interactions that occur during service delivery Such interactionscan have the greatest impact on service quality perceptions (Brady and Cronin, 2001).According to Brady and Cronin, three distinct factors constitute customer perceptions ofinteraction quality They include perceptions of employee attitudes, employee behaviors,and employee expertise We sought measures that would capture such factors

Based on the literature, constructs of interest were identified and defined Itemswere adopted, adapted, and created to reflect each construct We asked each of oursurvey respondents to identify the airline he/she uses most often for domestic (USA)flights and then to rate that airline relative to what their expectations were for thatspecific airline Subsequent to item generation, to refine our instrument, we subjectedthe constructs and items to a formal pre-test study We provided each construct(i.e individual attention, helpfulness, courtesy, promptness, and satisfaction) with itsdefinition and list of items to 12 college of business faculty members from several largeuniversities in the USA representing multiple disciplines such as marketing, serviceoperations, information systems, and strategy The participants evaluated each itembased on the definition for each construct and provided additional comments as theyrelate to the coverage of the content domain of each construct The feedback wereceived was useful and helped us reword several items and simplify the language.Pilot study

The issue of proper measurement is important and should not be taken lightly.Before embarking on confirmatory factor analytic techniques it is prudent to examine

IJOPM

28,9

808

the measurement model through exploratory methods All of the methods employed

are shown in Figure 1

The items were entered in a survey that was administered to 170 respondents who

were mostly business students Each construct with its block of items was first factor

analyzed separately This was done to assure the internal rule of unidimensionality

Only one factor emerged for each case and the loadings appeared to be adequate We

specified principal axis factoring as the method of extraction and direct oblimin

(oblique rotation) as the choice of rotation In other words, five separate analyses were

used, one for each construct For each case, one clear factor emerged and the loadings

were fairly substantive; the lowest was 0.595 while the overwhelming majority were

individual attention to 0.946 for satisfaction Next, we subjected the entire set of items

across all the constructs to an exploratory factor analysis using maximum likelihood

estimation as the method of extraction (to be as compatible with our confirmatory

approach as possible) and oblique rotation, as we would expect the factors to be

correlated Subjecting all the items across all the constructs to exploratory factor

analysis is more stringent than analyzing each construct separately as both the

internal and external rules of unidimensionality are addressed simultaneously

An attempt to run exploratory factor analysis across constructs (excluding

satisfaction which is treated as a dependent variable in our model) failed to discriminate

between constructs and produce four distinct factors In fact, one general factor emerged

that explained 55 percent of the variance This was not totally unexpected as prior

empirical research has reached similar findings A significant number of studies

(Arnau and Thompson, 2000; Brady and Cronin, 2001; Kilbourne et al., 2004; Lai, 2006)

have posited and tested a second-order factor structure where service quality is

conceptualized at the second-order latent level and the individual dimensions as

first-order latent variables Overall, after the pilot study we added three items and

modified two items to improve the coverage of content domain and readability

Description of large-scale sample

The final survey was administered to students (n ¼ 437) at a large university in the

Southeastern USA responses for the large-scale study were collected throughout the

course of 2006 (spring, summer and fall semesters, respectively) Respondents were all

students majoring in business Although we acknowledge that our respondents may

not be representative of the US “population”, students do form an eligible respondent

group as they too fly on airlines just like the many other residents of the USA As we

will present below, the respondents had a fairly decent level of flying experience

Students also come from all walks of life with quite a bit of variation in reported

household incomes A great number of respondents are considered “non-traditional”

students as they work and go to school at the same time

As is depicted in Table I, about a third of the respondents typically fly four or more

times a year and 60 percent of respondents at least twice a year About 7.1 percent fly

more than once a month These results suggest that overall, respondents have a decent

level of flying experience Respondents were quite evenly split between male and female

Female respondents accounted for 56.4 percent of the total sample and male respondents

43.6 percent In general, the respondents were relatively young with about 90.7 percent

being 40 years or less in age The majority of respondents were white (52.4 percent),

Airline service

quality

809

Figure 1.

Measurement and

structural model methods

Examining Unidimensionality: Internal Perspective Exploratory Factor Analysis for Each Construct

- Principal Axis Factoring

Model 4: Four Latent First-Order Factors & one Second-Order Factor

Selection of Measurement Model

- Four Latent First-Order Factors & one Second-Order Factor

- Convergent Validity (Magnitude of Coefficients & T-values)

Examination of Separate Measurement Model for each Construct

- Model Fit (Fit Indices)

- Convergent Validity (Magnitude of Coefficients & T-values)

- Discriminant Validity (AVE vs Squared Correlation)

Airline service

quality

811

followed by hispanic (22.0 percent), and black (14.8 percent) The respondents andperhaps their immediate family appeared to be relatively affluent; over 38.7 percentreported annual household incomes of over $60,000 and 14.9 percent reported incomes ofover $100,000 The overwhelming majority of respondents (86.9 percent) reported thatthey fly primarily for leisure whereas 13.1 percent reported flying primarily for business.The respondents typically fly in economy class of service (94 percent).

There are no substantive differences between our respondents and the non-respondentuniversity student body in terms of several demographics The average age of students atthe university is 25 for undergraduates and 34 for graduate students The university has apopulation that includes 61 percent female students and 39 percent male students.About 57 percent of all students indicate “White” as their ethnicity, 17 percent black, and

16 percent hispanic

Confirmatory methods

We followed Anderson and Gerbing’s (1988) two-step approach for structural equationmodeling, which suggests that researchers should first obtain an adequate measurementmodel, and then test the structural model in a second step Confirmatory factor analysiswith a covariance matrix as input and maximum likelihood estimation was firstperformed on each of the five constructs separately (Figure 1)

We first calculated a covariance matrix of the 30 items of interest in this study usingPRELIS 2.0 (Jo¨reskog and So¨rbom, 2002), and screened the data for possible violations

of statistical assumptions, such as skewness, kurtosis (peakedness), and possibleoutliers No violations of the assumptions were identified

To assess convergent validity, the individual item loadings and their respectivet-tests can be examined t-Values greater than j2j are considered to be significant at the0.05 level, and t-values greater than j2.576j suggest significance at the 0.01 level(Koufteros, 1999) It is generally recognized that to support model fit, a consensus

a comparative fit index (CFI) 0.90; and a standardized root mean square residual(RMR) below 0.05 (Byrne, 1998; Hu and Bentler, 1999; Koufteros and Marcoulides,2006) Reliability of a given construct within the context of structural equationmodeling can be assessed through the composite reliability estimate (Koufteros, 1999).The estimate of average variance extracted (AVE) (Fornell and Larcker, 1981) iscomplementary to the measure of composite reliability Estimates of AVE greater than0.50 and composite reliabilities above 0.80 are acceptable (Koufteros, 1999)

Discriminant validity can be assessed by comparing the AVE with the squaredcorrelation between constructs (Fornell and Larcker, 1981) Discriminant validity isexhibited when the AVE is higher than the squared correlation between twoconstructs In other words, this occurs when the “within variance” is higher than the

“between variance.”

First- and second-order measurement models

A hierarchical approach was used for further measurement model development Thehierarchical approach was deemed necessary given the findings from our pilot studyand prior research that posited the existence of a higher-order structure for servicequality related constructs (Kettinger and Lee, 1994; Kettinger et al., 1995; Myerscough,2002; Somers et al., 2003; Rodgers et al., 2005; Anitsal and Paige, 2006; Lai, 2006)

IJOPM

28,9

812

A hierarchical approach is a systematic process that evaluates alternative models that

can potentially describe relationships between observed and latent variables This

process includes the construction of four models (Figure 1) The first model is

hypothesized to include one first-order latent variable with 21 observed indicators

Model 2 hypothesizes four first-order uncorrelated (orthogonal) factors Model 3 is

similar to Model 2 except that factor correlations are specified Finally, Model 4

(Figure 2) includes one second-order factor and four first-order factors with

corresponding observed variables Our methods are reflective of the body of literature

that posits and tests higher-order models (Rindskopf and Rose, 1988; Arnau and

Thompson, 2000; Somers et al., 2003; Lai, 2006) Gerbing et al (1994) suggest that we

should recognize that the first-order factors are the constituent facets of the constructs

of interest In other words, they are the building blocks of the constructs In our specific

case, individual attention, helpfulness, courtesy, and promptness are the facets of

personal touch which is operationalized as a second-order factor

A measurement model is selected based on fit indices and theoretical considerations

An examination of convergent validity in the context of the selected measurement

model is also due (tests for discriminant validity tests and assessment of composite

reliability and AVE remain the same as reported in the previous analysis above)

Structural model

Once a measurement model has been selected, the next step is to test the substantive

hypothesis A structural model is evaluated and if the model fits the data adequately,

research hypothesis The magnitudes of the structural coefficients as well as the

statistical significance associated with such coefficients are useful in the assessment of

potential relationships between variables The analysis can be complemented by an

Results

The five posited measurement models appear to be supported by various fit indices

There is evidence of potent relationships between observed and latent variables

(t-values) and fit indices are suggestive of well-fitting models (Table II) Descriptive

statistics are presented in Table III Each construct had a composite reliability and an

AVE that met generally acceptable criteria (Table IV)

While the constructs appear to enjoy convergent validity and manifest strong model fit,

it is apparent from Table IV that there is lack of evidence suggestive of discriminant

validity between the four antecedents to passenger satisfaction The correlations between

constructs are strong, leading to high-squared correlations (i.e between variance) Such

squared correlations are higher than AVEs (within variance) for all comparisons For

discriminant validity support to be tenable, AVE for each construct has to be higher than

the squared correlation between two constructs under comparison Our findings provide

credence to the arguments advanced by other researchers that the model ought to be

specified at a higher level, i.e a second-order model can represent the data more effectively

Although other researchers in the service quality field have posited second-order

measurement models and attest to respectable model fit, it is imperative that alternative

models be specified, tested, and compared as there is quite a bit of disparity in findings

across the literature The first model (Model 1) (Table V) specifies that all 21 items are

Airline service

quality

813

Personal Touch

Airline service

quality

815

reflective of one latent variable Model fit was moderate Thex per degree of freedom

was above 5 and other fit indices were indicative of moderate model fit

The second model (Model 2) (Table V) posits four uncorrelated latent variables that

are related to their respective observed variables Given the strong correlations

between the latent variables as mentioned previously, an orthogonal specification for

the relationships between latent variables would be expected to produce poor model fit

Indeed, such was the case where all fit indices failed to pass muster

Model 3 (Table V) was similar to Model 2 except that the latent variables were free

to correlate Model fit was quite acceptable as all fit indices met respective criteria

However, retaining such a model would not resolve the issue of unidimensionality and

Table IV Correlations, reliability,

and AVE

Measurement models

Goodness of fit indices for

alternative models of factor

structure (n ¼ 437)

Model 1: one first-order factor

Model 2:

four first-order factors-uncorrelated

Model 3: four first-order factors-correlated

Model 4: four first-order factors, one second-order factor

Standardized root mean

square residual (RMR)

Table V Alternative measurement model structures

Airline service

quality

817