Does National

Embed Size (px)

Citation preview

  • 8/3/2019 Does National

    1/10

    Does national culture

    influence consumers'

    evaluation of travel

    services? A test ofHofstede's model of

    cross-cultural differences

    John C. Crotts and

    Ron Erdmann

    Introduction

    The role of customer satisfaction in

    influencing repeat patronage and positive

    word of mouth is well-documented (Crotts,

    1999; Augustyn and Ho, 1998; Kotler et al.,

    1998; Oppermann, 1998; Heskett et al.,

    1997). Without meeting or even exceeding

    customer expectations, a tourism enterprise

    as well as an entire destination should not

    expect to find loyal patrons who not only

    repeat purchase but also ``clone'' themselves

    among their friends and relatives (Ford and

    Heaton, 2000; Pine and Gilmore, 1998).

    Therefore, it has become a critical strategic

    task for management to systematically gain

    feedback from their guests as to their service's

    ability to satisfy needs and meet expectations.The task becomes more problematic where

    guests come from different national cultures.

    Do national cultural differences influence

    consumer evaluations? Cultural psychologists

    suggest that they do. According to Hofstede

    (1991), a prominent dimension of culture is

    the masculinity versus femininity dimension

    (i.e. competitiveness vs. cooperation) where

    certain cultures find desirable assertive

    behavior, while others emphasize behavior

    that is more modest. The degree to whichconsumers from a specific culture are willing

    to provide assertive and critical evaluative

    responses to consumer satisfaction surveys is

    a visible sign of this cultural dimension.

    According to Hofstede, individuals from

    societies where masculine values prevail more

    frequently evoke behaviors that are assertive,

    judgmental and have less concern for the

    feelings of others, which in turn should be

    reflected in their consumer satisfaction scores.

    On the other hand, guests from societieswhere there is more tenderness and sympathy

    for others and where assertiveness is not a

    desirable characteristic should be overtly

    moderate as to the extent they are willing to

    provide criticism in their evaluations. If this is

    true, not accounting for cultural differences in

    such measures may lead to over-estimating

    customer satisfaction, customer loyalty and

    positive word of mouth.

    This research is organized around the

    following two questions:(1) Do international visitors from masculine

    cultures evaluate tourist services more

    negatively than visitors from more

    feminine cultures?

    The authors

    John C. Crotts is at the Research Centre of Bornholm,

    Denmark and is Director and Associate Professor at the

    School of Business and Economics, College of Charleston,Charleston, South Carolina, USA.

    Ron Erdmann is Senior Market Research Analyst/Deputy

    Director, Tourism Industries, US Department of

    Commerce, International Trade Administration,

    Washington, DC, USA.

    Keywords

    Customer satisfaction, Measurement, Tourism, USA,

    National cultures

    Abstract

    The influence of national culture on consumer evaluations

    of travel services was the focus of this study. Drawing

    from a representative sample of overseas visitors to the

    USA, and controlling for socio-economic and trip

    characteristics, results provide a limited indication that

    national culture influences how customers evaluate travel

    services and their willingness to repeat purchase and

    recommend a service to others. The implications for

    researchers are that national cultural differences are one

    of many forces influencing consumer decision making. It

    is a measurable construct, like gender and socio-economicclass, that conditions how consumers interact with others

    and should be taken into account in our attempts to

    better understand consumers needs and expectations.

    Electronic access

    The research register for this journal is available at

    http://www.mcbup.com/research_registers/

    quality.asp

    The current issue and full text archive of this journal is

    available at

    http://www.emerald-library.com

    410

    Managing Service Quality

    Volume 10 . Number 6 . 2000 . pp. 410419

    # MCB University Press . ISSN 0960-4529

    http://www.mcbup.com/research_registers/quality.asphttp://www.mcbup.com/research_registers/quality.asphttp://www.emerald-library.com/http://www.emerald-library.com/http://www.mcbup.com/research_registers/quality.asphttp://www.emerald-library.com/
  • 8/3/2019 Does National

    2/10

    (2) What influence does national culture have

    in predicting repeat patronage and the

    probability of positive word of mouth?

    The research rests on the widely-held

    assumption that customer satisfaction is

    linked to customer loyalty and customerloyalty is linked to a firm's profits and future

    growth potential (Crotts, 1999; Augustyn and

    Ho, 1998; Kotler et al., 1998; Oppermann,

    1998; Heskett et al., 1997). The significance

    of the study rests on determining if

    satisfaction measures alone are the best

    predictors of the probability of repeat

    patronage and third-party endorsements, or

    should other social-cultural should be

    included.

    International visitors contribute much to

    the export earnings of tourist receiving

    nations, generating sales, labor earnings and

    employment. However, the marketplace is

    highly competitive. We contend that those

    destinations that can appreciate the relevant

    subtleties of cultural differences when gaining

    feedback from its visitors will have strategic

    advantages over those which do not.

    Literature review

    Customer satisfaction has become the

    buzzword of the 1990s, generating much

    interest and research among academic

    researchers. Today, leading hospitality and

    tourism firms recognize that profits hinge on

    the perceived level of customer satisfaction

    and the ability to achieve some level of

    consumer loyalty. Loyal customers have often

    been touted to be the profit engines of

    businesses, generating as much as 80 percent

    of a firm's total sales and profits through theirrepeat patronage (Noe, 1998). Even this

    percentage may be low, in that high-repeat

    customers also become a firm's willing

    apostles generating the positive word of

    mouth that drives new customers to a

    particular firm (Heskett et al., 1997).

    Systematically listening to customers allows

    management to measure how well they are

    doing in serving their customers and when

    benchmarked with other firms gain some

    understanding of their competitive position inthe marketplace. Gaining feedback also helps

    management to identify customer segments

    who are most predisposed to value their

    service, as well as the almost satisfied and

    dissatisfied who need extra attention in an

    effort to elevate their user status (Heskett et

    al., 1997). Performance measures that are

    simple, consistent and elicit customer

    feedback on dimensions that are important to

    the consumer allows management to track

    performance and predict service outcomes.

    Only by measuring can a firm become

    confident in its ability to retain customers and

    generate positive word of mouth.

    There are many methods used in gaining

    customer feedback that are generally accepted

    as valid and reliable. Fairfield Inns employs

    touch screen computers asking guests who are

    checking out to rate or score their

    performance. The Promus hotel group

    employs a 100 percent satisfaction guarantee

    to encourage customer complaints, which areinputted into a reporting system for

    identifying the root causes of service failures.

    Arguably the most common customer

    feedback mechanism is the self-administer

    survey (Parasuraman et al., 1990). The

    question arises as to the reliability and validity

    of these feedback measures particularly in a

    cross-cultural environment.

    Cultural values is an umbrella concept that

    includes such elements as shared values,

    beliefs and norms that collectively distinguish

    a particular group of people from others

    (Pizam et al., 1997). These widely shared

    values are programmed into individuals in

    subtle ways from quite an early age (Otaki et

    al., 1986), are resistant to change (Hofstede,

    1991) and remain evident when at home or

    while traveling abroad (Pizam and Reichel,

    1996; Pizam and Sussmann, 1995). Cultural

    differences have often been purported as the

    basis for specific ``stereotypes'' given to

    tourists from specific national origins. For

    example, residents have been reported tostereotype French and Italians as excessively

    demanding (Boissevian and Inglott, 1979),

    English as stiff, socially-conscious and honest

    (Pi-Sunyer, 1977), and American as cautious,

    calculating and purposeful (Brewer, 1978).

    Though there is considerable debate as to the

    accuracy of such national stereotypes (Dann,

    1993), there is little doubt that culture is one

    of many forces influencing consumer decision

    making (You et al., 2000; Pizam and

    Sussmann, 1995; Assael, 1987). According toPizam and colleagues, the focus of tourism

    research in this area should be to identify and

    understand the cultural differences that are

    not only scientifically valid but provide

    411

    Does national culture influence consumers' evaluation of travel services?

    John C. Crotts and Ron Erdmann

    Managing Service Quality

    Volume 10 . Number 6 . 2000 . 410419

  • 8/3/2019 Does National

    3/10

    information useful to practitioners. We

    contend that determining the extent to which

    national culture influences a consumer's

    willingness to report dissatisfaction and its

    subsequent influence on the likelihood that

    the customer will recommend the service to

    others are issues worthy of investigation.

    National culture has been defined in

    hundreds of ways (Erez and Early, 1993).

    Arguably, the most widely utilized dimensions

    of culture are the five presented by Hofstede

    (1980) and his colleagues (Hofstede and

    Bond, 1988) from their instrument called the

    Values Survey Module or VSM. Briefly,

    they are:

    (1) power distance (a tolerance for class

    differentials in society);

    (2) individualism (the degree to whichwelfare of the individual is valued more

    than the group);

    (3) masculinity (achievement orientation,

    competition, and materialism);

    (4) uncertainty avoidance (intolerance of

    risk; and later,

    (5) the Confucian dynamic, or long-term

    orientation (stability, thrift, respect for

    tradition and the future).

    Drawing from Hofstede's (1991) research,

    Asian societies tend to score high in long-term

    orientation, collectivism and power distance,

    but mixed in terms of masculinity and

    uncertainty avoidance characteristics. On the

    other hand, Western societies tend to score

    low long-term orientation, collectivism,

    power distance and uncertainty avoidance,

    but mixed in terms of masculinity. The

    masculinity index was selected since it most

    clearly articulates the cultural traits that are

    assertive, judgmental and have less concern

    for the feelings of others, which in turn shouldbe reflected in their consumer satisfaction

    scores.

    Method

    The data for the study were from the 1996,

    1997 and 1998 Inflight Survey of Overseas

    Visitors to the United States. Annually, the

    US Department of Commerce's Tourism

    Industries surveys more than 80,000 overseasvisitors on regularly scheduled flights with

    self-administered questionnaires distributed

    in-flight by airline flight crew personnel after

    departure from the USA. Employed is a

    random cluster sampling procedure, where all

    passengers on randomly selected flights. on

    randomly selected days, are administered the

    self-administrated survey instrument.

    Currently there are 61 US and foreign flag

    carriers cooperating with the survey and

    response rates are generally quite high,

    averaging 58 percent (Tourism Industries,

    1999).

    The survey itself is composed of 29

    questions and is available in 11 languages.

    Each language version of the survey has been

    forward and back translated to insure

    uniformity among the versions. The purpose of

    the survey is to elicit visitor feedback and travel

    party and trip characteristics of overseas

    visitors that firms can use to better understand

    and serve the overseas markets. Two of thesequestions ask respondents to rate the quality of

    the airport in which they had just departed, as

    well as the airline they were using, on a variety

    of attributes along a five point scale (where 5 =

    excellent; 1 = poor). A third question attempts

    to estimate customer loyalty by asking

    respondents if they would ``choose or

    recommend the airline they were using on their

    next trip on this route''. Responses are

    recorded along a five point scale where

    1 = definitely would, 2 = probably would,

    3 = probably would not, 4 = definitely would

    not and 5 = unsure. These three ratings were

    used as dependent measures in the subsequent

    analysis. The principle limitation of the data is

    that the name of the airline the respondent

    evaluates is considered proprietary information

    that is not released to researchers.

    For purposes of the analysis, the dataset

    was limited to those respondents who

    indicated their country of birth was the UK,

    Germany, Japan, Brazil and Taiwan. These

    countries were selected for their significanceto the USA's overseas market (i.e. in 1998,

    55.1 percent of overseas arrivals resided in

    these six nations) but also due to their unique

    scores along Hofstede's (1980, 1991) national

    culture measures. In addition, the selection of

    country of birth was made over country of

    citizenship since research has shown that

    virtually all of the cultural dimensions of

    interest are learned by the age of ten years and

    remain relatively immune to change over the

    remaining life span (Hofstede, 1991).The second phase in the sample selection

    process delimited the final sample to those

    visitors who had not visited the USA in the

    previous five years; whose primary trip

    412

    Does national culture influence consumers' evaluation of travel services?

    John C. Crotts and Ron Erdmann

    Managing Service Quality

    Volume 10 . Number 6 . 2000 . 410419

  • 8/3/2019 Does National

    4/10

    purposes were deemed discretionary (i.e.

    holidays, leisure, visiting friends and

    relatives); who were between the ages of 45 to

    60 years of age; and were employed in

    managerial or professional occupations.

    Hofstede (1991) warned that comparisons of

    countries should always be based upon people

    of the same socio-economic groupings since

    social class also influences behavior.

    Controlling for the number of previous visits

    (e.g. no previous visits to the USA in past five

    years) and trip purpose (e.g. discretionary)

    was also deemed appropriate since the trip

    planning characteristics can be quite different

    between these groups (Oppermann, 1998;

    Guy et al., 1990, Gitelson and Crompton,

    1984) and not controlling for their effects may

    have biased the results. The specific age andoccupation categories were selected by way of

    the frequencies of responses to preserve the

    sample sizes while still controlling as much as

    possible the influence on the study's

    dependent measures. The net effect of

    employing the five criteria reduced the entire

    database to 983 respondents, or less than

    1 percent of the available database.

    Subjects were next assigned to one of three

    groups based upon Hofstede's (1980, 1991)

    national cultural rankings along the

    masculinity index. Specifically, subjects from

    Japan were assigned to the high masculine

    group (Japan ranked first of 53 nations),

    subjects from the UK and Germany (ranks

    tied at ninth/tenth of 53 nations) were

    assigned to the middle masculinity group and

    subjects from Brazil and Taiwan (27th and

    32nd of 53 nations) to the low masculinity

    group. Though these measures of national

    culture are dated, Hofstede notes that national

    cultures change very slowly and when:

    Cultures shift,F F F

    they shift together, so that thedifferences between them remain intact (1991,

    p. 77).

    Given that the six nations selected in this

    study were chosen for their unique

    differences, the relative national cultural

    differences between them should have

    remained unchanged. In all, three matched

    samples were derived where 582 (59.2

    percent) subjects were assigned to the high

    masculine group, 283 (28.8 percent) to the

    middle group and 118 (12.0 percent) to thelow masculine group again for a total of

    983 respondents.

    These national culture measures and their

    subsequent groupings provided a means in

    which to explore if national culture indeed

    predicts tourist behavior. Specifically,

    individuals born in a cultural setting where

    modest behavior is the norm (i.e. femininity)

    will be less critical in the evaluations of tourist

    services when compared to respondents born

    in cultural settings where aggressive (i.e.

    masculine) behaviors are the norm.

    Furthermore, national cultural differences

    will be evident in explaining differences in

    customer loyalty as reflected by visitors' self

    reported intentions to make a repeat purchase

    and recommend the service to others.

    Results

    Description of sub groupsDescribing the sub samples' social

    demographic characteristics would be

    meaningless since the sampling criteria

    eliminated most differences. However, a

    number of issues are interesting from a

    methodological perspective. First, self reports

    of country of birth differed from country of

    residence in only nine (0.9 percent) of the 983

    cases. This was somewhat disappointing in

    that a priori we wished to explore the effects

    of culture of residents on culture of birth

    values. Though an intriguing issue, such an

    analysis is not often possible since one

    measure tends to mirror the other even in

    large datasets of this nature.

    Even with the sampling criteria being

    employed, further steps were warranted to

    test for differences between the sub sample

    groups. When asked to indicate their gender,

    approximately three-quarters of each sample's

    respondents indicated they were female (see

    Table I). However, the proportion of

    respondents who were female wassignificantly higher in the high masculine

    group at 84.4 percent. Gender, like

    nationality of birth, is an important factor to

    account for in analyses of this nature since

    men, on average, are programmed with

    tougher values than women, though the gap

    between them varies by country. According to

    Hofstede (1991), the gap between genders

    rises along with masculinity scores, implying

    that the higher representation of females in

    the high masculine group should producemore moderate or less harsh evaluations in

    contrast to what was predicted. Though

    gender differences should be anticipated,

    Hofstede has warned that they are:

    413

    Does national culture influence consumers' evaluation of travel services?

    John C. Crotts and Ron Erdmann

    Managing Service Quality

    Volume 10 . Number 6 . 2000 . 410419

  • 8/3/2019 Does National

    5/10

    F F F statistical rather than absolute, [implying]

    that there is an overlap between the values of

    men and women so that any given value may be

    found among men and women (Hofstede, 1991,

    p. 85).

    To further determine, where possible, that the

    sub groups were matched in all important

    areas, crosstabs were produced on where the

    respondents were sitting on the aircraft.

    Approximately nine out of ten respondents

    indicated they were sitting in the economy or

    coach section of the airplane. Though the

    distribution across the three groupings wassimilar, respondents in the high masculine

    group as a whole reported sitting in business

    class significantly more often than the other

    two groupings (see Table I). Predicting the

    potential effect of such differences is difficult.

    On the one hand, the vast majority of

    published complaints in newspapers indicate

    that passengers in the economy class

    experience the most frustration due to the

    frugal level of service. On the other hand,

    perceived service quality is often defined asthe value received for the price paid (Murphy

    and Pritchard, 1997; Chang and Wildt, 1994)

    so first-class and business-class passengers

    may be more discerning when it comes to

    evaluating service quality when compared to

    their economy-class counterparts. However,

    such differences should be revealed in the

    level of service and not so much in the airport

    or airline facilities that made up the majority

    of dimensions respondents could evaluate.

    Does national culture influence customer

    evaluations?

    Respondents ratings of airport facilities, on

    average, were in the average to good range.

    Only the ratings of concessionary prices

    dipped into the poor range (Table II). As

    predicted, respondents assigned to the low

    masculinity national culture group reported

    on average more positive evaluation of airport

    facilities. Moreover, the between group

    differences were most pronounced when

    respondents in the low to middle masculinity

    groups were compared to respondents in the

    high masculinity (i.e. Japan) group. Both

    Sceffe and Donnett T3 post hoc tests revealed

    significant differences between the high and

    low masculinity groups on all ten criteria in

    the directions predicted. On the other hand,

    differences between the medium masculinity

    group with either the low or high masculinity

    groups were few. Though the average ratings

    across all three groups were in the direction

    predicted, respondents assigned to the

    medium group reached statistical significance

    only in ratings of quality concession goods

    and airport security.

    Similar differences were revealed

    comparing the responses of the lower tomiddle masculinity groups with the high

    masculine group on their evaluations of the

    airflight (Table III). Respondents in the high

    masculine grouping reported on average

    lower ratings of airline service quality than

    their lower masculine counterparts on 14 of

    the 15 evaluative criteria. However, no

    statistical differences were found between the

    lower and middle masculinity groups.

    It is worth noting that the three groups

    virtually mirrored one another in terms oftheir average rating of airline ability to have an

    on-time departure. On a relative basis, this

    dimension of service quality was ranked low

    by all three groups, suggesting that when

    Table I Description of the samples national cultural groupings by gender and where seated on aircraft

    Low Medium High

    Gender

    Male (percent) 23.3 26.0 16.0

    Female (percent) 76.7 74.0 84.4

    100 100 100Chi 13.60, p < 0.001

    Where are you sitting on the airplane?

    1st class (percent) 0.8 0.3 1.5

    Business class (percent) 4.2 2.6 10.4

    Economy (percent) 95.0 97.0 88.1

    100 100 100

    Chi 22.63, p < 0.000

    Sample sizes 118 283 582

    414

    Does national culture influence consumers' evaluation of travel services?

    John C. Crotts and Ron Erdmann

    Managing Service Quality

    Volume 10 . Number 6 . 2000 . 410419

  • 8/3/2019 Does National

    6/10

    delay occurs it has a negative influence on

    ratings, regardless of cultural norms.

    What influence does national culture

    have in predicting customer loyalty?

    Asked ``Would you choose or recommend this

    airline for your next trip on this route?'', 18.8

    percent indicated that they definitely would,

    59.6 percent said they probably would, 5.0

    percent they probably would not, 1.7 percent

    definitely would not and 12.0 percent were

    unsure. Responses to this question were used

    to classify into two separate groups, which are

    similar to Heskett et al.'s (1997) classification

    strategy. They were the definite repeat

    purchasers and third party endorsers

    (hereafter refered to as loyals) comprised of

    216 respondents and the unlikely to repeat

    purchase or recommend to others (hereafter

    refered to as potential defectors) comprised of

    190 respondents. The large 59.6 percent of

    the sample who indicated they probably

    would repeat purchase was dropped from the

    analysis in an effort to create the greatest

    possible distances between the bivariate

    loyalty groups.

    A canonical discriminant analysis with the

    stepwise procedure was conducted with the

    Table II ANOVA table mean differences between national cultural groups in evaluation of US airport facilities

    Low Medium High F p-value

    Dimension rated

    Airport access 3.92a 3.90b 3.62a,b 12.31 0.000

    Ground transportation 3.84a 3.78b 3.31a,b 30.5 0.000

    Terminal convenience 3.82a 3.74b 3.26a,b 33.03 0.000Cleanliness 3.96a 3.92b 3.47a,b 36.89 0.000

    Concession goods 3.68a,b 3.20a,c 3.07a,b,c 17.58 0.000

    Concession prices 2.97a 2.60b 2.52a,b 5.31 0.005

    Terminal seating 3.73a 3.48b 3.05a,b 31.67 0.000

    International facilities 3.53a 3.36b 3.00a,b 22.15 0.000

    Security 3.93a,b 3.59a,c 3.33a,b,c 28.15 0.000

    Overall evaluation 3.74a 3.54b 3.32a,b 15.45 0.000

    Notes: Scored on a five point scale where 5 = excellent and 1 = poorMean values with the same subscripts significantly differ from one another employing both Scheffe and DonnettT3 tests at p < 0.01

    Table III ANOVA table mean differences between national cultural groups in evaluation of airlines

    Low Medium High F p-value

    Dimension rated

    Convenient schedules 3.99a 3.90b 3.57a,b 17.12 0.000

    Ticket price 3.34a 3.76b 3.23a,b 26.01 0.000

    Reservation service 3.75a 3.81b 3.43a,b 15.63 0.000

    Check-in waiting time 3.66a 3.55b 3.14a,b 26.60 0.000

    Check-in personnel 3.79a 3.95b 3.41a,b 34.50 0.000

    Departure time 3.54 3.58 3.57 0.044 0.957

    Food and beverage 3.55a 3.65b 3.33a,b 10.14 0.000

    Flight attendants 3.98a 4.07b 3.78a,b 10.77 0.000

    Cabin cleanliness 3.88a 4.02b 3.58a,b 28.74 0.000

    Cabin noise level 3.36a 3.29b 2.97a,b 18.64 0.000

    Seat comfort 3.13 3.13a 2.86a 8.22 0.000

    Cabin layout 3.54a 3.47b 3.07a,b 28.28 0.000

    Carry on storage 3.68a 3.63b 3.14a,b 42.11 0.000

    Overall evaluation of aircraft 3.84a 3.80b 3.46a,b 19.88 0.000

    Overall evaluation of airline 3.85a 3.86b 3.50a,b 24.61 0.000

    Notes: Scored on a five-point scale where 5 = excellent and 1= poorMean values with the same subscripts significantly differ from one another employing both Scheffe and DonnettT3 tests at p < 0.01

    415

    Does national culture influence consumers' evaluation of travel services?

    John C. Crotts and Ron Erdmann

    Managing Service Quality

    Volume 10 . Number 6 . 2000 . 410419

  • 8/3/2019 Does National

    7/10

    loyals and defectors variable as the group

    membership variable (1, 0). The purpose of

    the disciminant analysis was to identify the

    relative importance of national culture in

    combination with satisfaction measures in

    discriminating between loyalty and defectors

    customers. In addition, the two dummy

    variables describing the compartment of the

    aircraft in which the respondent was sitting

    (i.e. first class, business class, economy class)

    and gender were included to test other

    competing hypotheses simultaneously. The

    stepwise procedure provided a means to

    determine the relative importance of each of

    the variables to significantly increase the

    loyalty prediction. Again the analysis rests on

    the assumption that not accounting for social-

    cultural differences in satisfaction measures

    may lead to less accurate loyalty measures.

    A test of equality of group means revealed

    significant differences between those

    respondents in the loyal and defector groups

    on five of the six independent variables. Table

    IV shows the means and statistical differences

    of the independent variables relative to the

    group membership. As predicted, loyalty

    groups differed significantly on satisfaction

    measures. Subjects indicating high loyalty

    tendencies rated the aircraft and flight in the

    good to excellent range while subjects

    revealing terrorist tendencies rated the same

    attributes in the fair to average range. In

    addition, subjects who indicated that they

    were born in the less masculine societies a

    measure of national culture reported greater

    loyalty tendencies than those born in more

    masculine societies. Conversely, those

    subjects reporting their country of birth in

    masculine societies were more likely to report

    defector attitudes. Significant differencesbetween males and females were also found

    where females were most likely to report

    loyalty than males. Surprisingly, no significant

    differences were found between

    compartments in the aircraft in which the

    passenger was sitting.

    Results of the canonical discriminate

    function, shown in Table V, reveals the

    discriminant function was statistically

    significant, as measured by the chi-square

    statistic. With an eigenvalue of 0.124 and a

    canonical correlation value of 0.744, the

    variables accounted for a significant amount

    of the variance. The Wilks lambda which is

    the ratio of the within-group sum of squares

    to the total sum of squares for the model

    indicated that both groups were significantly

    different from one another.

    Of particular importance in this study was

    the need to identify which variables would be

    retained in the stepwise procedure as well as

    their relative importance in discriminating

    between loyal and defector groups. Step 1

    retained respondents' overall evaluation of theflight, step 2 the national culture measure and

    step 3 respondents' overall evaluation of the

    aircraft (Table VI). Gender and where

    respondents were sitting on the aircraft were

    not significant as to their ability to reduce the

    Wilks lambda values and account for the

    residual variance.

    The final step in the discriminant analysis

    was to test the function's ability to correctly

    classify respondents into the appropriate

    group. The classification accuracy of the

    model achieved a relatively high accuracy hit

    rate at 88.4 percent where 80.2 percent of the

    defector group and 94.9 percent of the loyal

    group could be correctly classified.

    Discussions and implications

    Before summarizing the results of this study

    and discussing their implications, it is

    important to review this study's limitations.

    First, the results should not be construed to

    be representative of all tourists from Japan,

    the UK, Germany, France, Brazil and

    Taiwan, let alone those who visit the USA,

    due to the highly delimited nature of the

    sample. Without additional research it is far

    too premature to make such generalizations.

    Table IV Test of equality of group means: loyalty group membership

    Independent variables Loyal Defectors F p