A multivariate analysis of hotel benefit bundles and choice trade-offs

ht. J. Hospitality Management Vol. 7 No. 1, pp. 29-41,1988 02X-4319/88 53.00 + 0.00 Printed in Great Britain @ Pergamon Press plc

A multivariate analysis of hotel benefit bundles and choice trade-offs

Lance Wilensky and Francis Buttle Department ofHotel, Restaurant and Travel Administration, University ofMassachuserts

at Amherst, Amherst, MA 01003, U.S.A.

itiarketing research in hotels is generally limited to the analysis of guest related darn. collected either during registration or from guest comment cards. The data obtained in this fashion have little merit as predictors of customer behaviour (Lewis, 1981) yet techniques of multir~ariate analysis are available which enable management not only to predict customer choice with some degree of confidence but to develop more approp- riate segmentation, positioning and marketing mix strategies. This paper explores the application of these techniques in the competitive hotel market at the world’s busiest airport-London Heathrow. With the cooperation of Holiday inn management, a tn’o stage programnle of research was ilndertaken.

Key words: hotel benefit bundles multivariate analysis choice trade-offs

Objectives

Objectives for the research were expressed as follovvs:

Stage 1 (1) To determine which attributes in a hotel the Holiday Inn guest considers important

when selecting a hotef at which to stay. (2) To isolate the underlying factors (benetit dimensions) in these attributes. (3) To assess the relative significance of these benefit dimensions to the hotel guest.

Stage 2 (1) To establish a scale of perceived instrumentality for the hotels competing in the

Heathrow market, using the benefit dimensions identified above. (2) To plot the perceived profile of these units in order to establish the relative

competitive advantage of the Holiday Inn vis ci vis the other properties. (3) To assess the perceived level of substitutability between the Heathrow hotel

competitors.

29

30 Lance Wilensky and Francis Buttie

Theoretical foundations

Consumer behaviour theorists and investigators have struggled for decades to understand, and therefore be in position to influence. the decision-making processes of consumers. One of the most fruitful constructs to emerge in recent years is that of in\.olvement. Complex buying behaviour is a class of high involvement decision-making in which the buyer passes through a learn-feel-do process during purchasing (Buttle, 1986). Firstly, customers acquire information about alternative products. then they evaluate the alterna- tives, so developing preferences before finally making a buying decision. The consumer can be characterised as passing through a learning process in which attitudes (i.e. brand evaluations based upon the acquired information) piay a significant role in determining purchase intention. Other models of buyer behaviour are also depicted in Fig. 1.

On most occasions. there are many hotels competing for the guest’s custom. If merely seeking shelter the guest may be brand-indifferent and hotels can become perfect substitutes for each other, but if aiming to satisfy a higher-order need it is probable that the consumer will attempt to discriminate between alternatives and make an informed choice based upon each property’s attributes and therefore its ability to solve his accommodation problem, This is the learn-feel-do process, which is a fundamental assumption of this research.

Two people’s seIection of different hotels can thus be explained in three ways (Lewis, 1983):

(1) They may be seeking different attributes in a hotel. (2) They may perceive each hotel’s attributes differently. because they base different

information. (3) Because they ptace different evaluations on each property’s attributes they are

willing to make different trade-offs. For example, one may be prepared to pay more for a hotel with room service whereas the other may sacrifice room service for a better rate. The process of hotel choice described by Lewis above involves the application of compensatory decision rules to evaluated brand information.

T/re concepts of attitude nnd benefit are central to this process Attitudes are learned predispositions to respond in a consistently favourable or unfavour- able way towards an object or class of objects (Allport. 1935). An individual’s attitude towards a hotel brand comprises his beliefs or knowledge about the brand (the cogniril*e

Involvement

High LOW

otternatIves Dlssononce -

LOW reducing buytng Inertia behawour

Fig. 1. Types of buying behaviour.

HoteI benefit bundles and choice trade-offs 31

component of attitude) derived from his perception of the brand’s attributes, his like or dislike of the brand (the ~~~~~~~~ component) and his tendency to behave in a particular way towards the brand (the be~avi~l~rul component of attitude).

Benefits provide the satisfactions which are the sine qua non of marketing. Guests do not choose to stay in a hate1 because it possesses attributes such as a swimming pool or haute cuisine restaurant but because of the benefits they expect to enjoy from experiencing these attributes. Hotel attributes, therefore, are the features which produce the benefits which in turn lead to satisfied customers.

The research design was as fotlows:

Stage I A self-completion questionnaire under a covering letter was handed by front-desk staff to independent guests (Le. not members of parties such as aircrew or conferences) checking into the Holiday Inn, Heathrow, with a spoken request to return the completed instrument to the desk. When checking-out, guests were asked whether this had been done. Of the 500 questionnaires distributed. 130 (26%) were returned. The questionnaire asked the guest to rate each of -40 hotel attributes, for its importance in the hotef choice decision, on a scale of one (of no importance) to six (most important). The list of attributes had been culled from other studies of a similar nature as well as from unstructured interviews with industry personnel. The results were subjected to factor analysis using the principal component method.

Stage 2 A self-completion questionnaire, administered as above was distributed, meeting with a similar number of responses and response rate. This questionnaire asked guests to rate on a Spoint scale (1 = very bad, 5 = very good) their perceptions of seven hotels servicing the Heathrow market (Table I). The attributes rated were the factors which had been extracted from the analysis of the stage 1 survey. The results were subjected to t\vo forms of analysis: firstly. non-metric multidimensional scaling was used to establish how Holiday Inn and the other hotels were positioned in the guest’s mind; secondly, cluster analysis was used to find out which hotels were most similar and therefore competitors. In addition, the questionnaire asked guests to provide an overall opinion of each of the seven hotels on a scale ranging from 1 (very bad) to 9 (very good), and some demographic data.

Results

siuge 1 Attributes mentioned by at Ieast 85% of the respondents as important (i.e. rated either 5 or 6 on the b-point scale) in the hotel selection decision were:

Lance Wtlensky and Francis Burtie

Room and bath cleanliness (92%) Professionalism of staff (89%) Friendliness and courtesy of staff (86%).

A National Opinion Polls survey in 1981 produced these comparable results; picked out as particularly important were the following: ‘private bathrooms, friendly staff, an appear- ance of caring about guests, efficient running of the hotel and ample car parking’.

Attributes mentioned by at least 50% of the respondents as being unimportant (i.e. rated either 1 or 2 on the &point scale) to the selection decision were:

Entertainment for children (74%) Hotel entertainment (58%) Sauna, steambath and gymnasium (55%) Sporting facilities (55%) Conference facilities (53%).

The rating data were analysed in order to establish the underlying factors in the data. The initial unrotated factor matrix identified eight factors which explained 60% of the total variance in the 40 variables. One factor which was dominated by the attribute ‘convenience of location’ was excluded from further analysis as the project was restricted to the hotels in the Heathrow vicinity only.

In order to improve interpretability of the factors, a varimax rotation was performed on the matrix. Seven relatively clean factors (see Table 2). in which the variables loading heavily on one factor do not load heavily on others, were found to account for almost 50% of the variance. Cleanliness indicates that the factors are relatively self-contained. The factor titles assigned are the arbitrary decision of the researchers and are intended to reflect the type of customer benefit implied by the high-loading attributes

Table 7. Major hotels at Heathrow Airport

Fivestarhotels Operator Skyline Sheraton Mngt Corp

Fourstarhotels Excelsior Trusthouse Forte Penta Penta Hotels Holiday Inn Holiday Inns Sheraton Heathrow Sheraton Mngt Corp

Threestarhotels Crest Crest Hotels Post House Trusthouse Forte Ariel* Trusthouse Forte

Skyways* Trusthouse Forte

*Hotels in the Heathrow market but excluded from this analysis.

Hotel benefit bundles and choice trade-offs 33

Table 2. Critical factors in hotel selection

Factor No.

Factor title

Percentage variance

accountedfor High loadings on

Opportunities for relaxation 11%

Value for money 7%

Standard of personal service 7%

Physical attractiveness

Appealing image

Standard of services

Suitability for business guests

7%

6%

6%

5%

Sauna, steambath and gymnasium Sporting facilities Swimming pool Videochannel room ente~ainment Extras (toiletries, mints, bathrobe) Entertainmentforchildren Hotel entertainment Price and value Actual room rate Pricediscountsand sales promotions Friendliness and courtesy of staff Front-deskefficiencyandservice Room/bath cleanliness Quick registration/checkout Reliable reservation system interior appearance Room and bath decorand furnishings Room and bath size Hotel personality and prestige Reputation of hotel Security of hotel and hotel room Conveniences (hair dryer, suit press) Services (messages, porter, laundry) Meeting roomsand business

facilities (telex, secretary) Conferencefacilities

The rating scales on Holiday Inns‘ own guest comment card bear a resemblance to ours. Guests are asked to rate seven dimensions of the hotel product: the room, food quality,

service (catering facilities). value for money. front desk staff. service. and other facilities. In order to establish the relative importance of these factors to the guest a benefit vafrre

index (BVI) was constructed. The mean rating for each of the attributes was multiplied by the attribute’s loading on the seven factors: the product was then divided by the total number of attributes (JO), producing the results shown in Table 3. The figures show that the most valued benefits for the Holiday Inn guest are, in order of importance: standard of personal service, physical attractiveness, opportunities for relaxation. standard of services, appealing image, same for money and suitability for business guests.

So far then, we have reported the findings with respect to stage one objectives. Let us now move to stage two.

3-l Lance Wllensky and Francis Buttle

Table 3. Benefit value index

Factor No.”

Sum of attribute means Benefitvalue multiplied by factor loadings index

28.14 0.7035 21.78 0.5447 32.11 0.8028 28.29 0.7072 23.35 0.5839 24.65 0.6164 12.72 0.3181

*Refer to Table 2 for explanation. I I

Stage 2 The degree to which a brand has the attributes wanted by a consumer is called its perceived instrumentality (Xssael, 19S-l). To measure the perceived instrumentality (for the Holiday Inn guest) of hotels competing in the Heathrow market a second questionnaire, con- structed as described in the Methodology Section, was compiled and administered.

blean ratings of the competing hotels on the five-point benefit scale are detailed in Table 4. Not surprisingly. since the sample comprised Holiday Inn guests, Holiday Inn was rated

highest on all factors. To establish the estent to which the competing hotels were perceived as capable of

delivering the benefits sought by Holiday Inn guests an ir1tie.r ofperceirved itzstrutt~entnli!\ (IPI) was constructed by aggregating the products of the mean ratings for each factor multiplied by the betlefit tzhe inde_~. For example, the index was calculated thus for the Sheraton HeathroIv:

IPI = (3.09 x 0.7035) + (2.78 x 0.5147) + (3.31 x O.SO?S) + (3.35 x 0.7077) - (3.75 x 0.5339) + (3.55 x 0.6161) + (3.37 x 0.3lSl)

= 11.1989.

Equivalent calculations for all seven hotels appear in Table 5. The same process can be used to calculate the IPI of the ‘perfect’ hotel. simply by

substituting the highest possible point on the rating scale (5) for the mean rating in the above calculation: the perfect score is 21.3530. This statistic can be used for comparison purposes in two vvays: firstly. by dividing the instrumentality score for each factor by the ideal for the factor. management could calculate how perfect the hotel is at delivering each of the seven benefits: secondly, the overall IPI for each hotel divided by the ideal tells management how close it is to delivering the ideal product. This statistic we have called the percentage perfecr score. Table 6 shows the percentage perfect scores for the seven hotels rated by Holiday Inn guests.

Hotel benefit bundles and &we trade-offs 35

Table 4. Mean benefit scores for Heathrow hotels

Hotel No.t Factor

No.” 1 2 3 4 5 6 7

1 3.09 3.94 2.44 2.60 2.80 3.43 2.48

2 2.78 3.74 2.87 2.76 2.65 2.81 2.63

3 3.31 3.99 2.95 3.09 3.02 3.17 2.82

4 3.35 4.37 2.81 3.17 3.22 3.54 2.84

5 3.75 3.97 2.51 3.00 3.20 3.77 2.70

6 3.59 4.05 2.67 2.98 3.02 3.69 2.80

7 3.37 3.69 2.89 3.21 3.38 3.40 3.05

*Refer to Table 2 for explanation. *Where hotel numbers are as follows: 1 = Sheraton Heathrow, 2 = Holiday Inn,

3 = Crest Heathrow, 4 = THF Posthouse, 5 = Penta Heathrow, 6 = Sheraton Skyline, 7 = THF Excelsior.

This form of analysis shows how well a hotel is performing in fulfilling the guests’ needs on each specific benefit, as well as in the overall trade-off. Table 6 shows that the other hotels are perceived as offering substantially less than 70% of what Holiday Inn guests want. It is now clear why guests chose to stay at the Holiday Inn. The hotel is perceived as delivering better on each benefit, as well as overall in the benefit trade-off.

It is possible to use the data output from these caIculations to graphically plot the competitive profiles of the Heathrow hotels. The two key statistics are benefir vnlue index and the percentage perfect Scores for each hotel on each factor. Figure 2 gives an example of the competitive profiles of two of the hotels: Sheraton Heathrow and Holiday Inn. The

Table 5. Perceived instrumentality Table 6. Percen rage perfect scores scores for Heathrow hotels for Heathrow hotels

Hotel Index of perceived

No.’ instrumefltai~ty

1 14.198 2 17.088 3 11.663 4 12.665 5 12.919 6 14.546 7 11.722

*RefertoTabie4forexpianation.

Hotel Percentage No.* perfectscore

1 66.40 2 79.91 3 54.54 4 59.23 5 60.41 6 68.02 7 54.82

*RefertoTabie4forexplanation.

36

0.80+ Al 81

0.70+ A3 A2 83 B2

A4 B4 O.bO+

A5 05 VALUE - INDEX - A6 86

0.50+

0.40+

A7 97 0*30+

+___~-_~~~+~~~~-ll~~+-~ -------+-~---~~~~+~~~11-~~~~~+ 0.560 0.630 0.700 0.770 0.840 0.910

PERCENTAGE PERFECT PREFERENCE

where A = Sheraton Heathrow 'I? = Holi",ay Inn 1 = Standard of Personal Service 2 = Physical Attractiveness 3 = Opportunities for relaxation 4 = Standard of Services 5 = Appeailng Imqe 6 = Value for Money 7 = Suitabiiity for Business Guests

Fig. 2. Competitive profile: Sheraton Heathrow vs Hofiday Inn.

greater the interpoint distances between benefits higher up the left hand scale the stronger the competitive position; ho\veser the greatest Iveakness of this method is th;\t it allows only two hotels to be compared at any one time.

For strategic planning purposes. a technique u hich quantitatively identifies the benefits vvhich are important to the target market. and then picks out how that market perceives a particular property and its competition. can be esrremel> valuable (Lewis. 192). Dravving comparisons betxveen two properties has its place in this process but it fails to provide a truly compact index of the overall (dis)similarity of ail properties competing in the market

Hotel benefit bundles and chom trade-offs 37

place. Such an indes can be derived by mathematically combining the benefits in a

multidimensional scaling procedure called re&ced space positioning nnn/ysis (Coxon,

1952).

The index of(dis)sitnilarify is derived thus: (1) Using only questionnaires in which respondents have been able to rate all hotels in

the study, calculate the mean score for all benefits and all hotels (in this survey. despite exclusion of invalid questionnaires. mean benefit scores were very similar to those in Table

1). (3) Calculate the difference in mean scores between hotels. With seven hotels there are

31 pairings of hotels for which comparisons must be made. Given seven benefits, there is a total of 117 calculations.

(3) For each hotel pair comparison, sum the total difference between means. ignoring signs. This provides an aggregate index ofsimilnrit~ between pairs of hotels.

(4) An irlciex of&similarity is calculated by dividing this aggregate by ttvo. The resulting statistic shows how much change must take place in the guests’ perceptions for the hotels to be judged identical. The it2cie.y of (tiis)sirnilarir_v thus constructed is an a\.erage representation of how guests perceive the seven hotels.

The indices for this study are shown in Table 7. The matrix shows that the Sheraton Heathrow and Holiday Inn are less dissimilar (1.906) than the Holiday Inn and Crest Heathrow (4.177).

Use of an appropriate non-metric multi-dimensional scaling procedure can graphical11 present the data in a \vay which simplifies simultaneous analysis. These data ivere input to the Gutman-Lingoes Small Scale Analysis algorithm. A key question concerns the number of dimensions to represent. Since there are seven hotels, 6-dimensional space would be needed to account for all the differences between them. Since conceptualising 6-D space is near impossible for the human mind the Gutman-Lingoes algorithm begins with a one dimensional representation of the data. The number of dimensions can then be increased at the discretion of the analyst. Computer analysis suggested that a two-dimen- sional representation of the data provided a satisfactory summary interpretation of yuests'

Table 7. Matrix of indices of (dis)similarity between Heathrow hotels

Hotel Hotel Hotel Hotel Hotel Hotel 1 2 3 4 5 6

Hotel 2 1.9062 Hotel 3 2.3541 4.1770 Hotel 4 1.5312 3.4375 0.8645 Hotel 5 1.0148 2.9211 1.3867 0.5416 Hotel 6 0.3229 1.8750 2.4062 1.5625 1.0540 Hotel 7 2.1041 4.0104 0.4583 0.5729 1.0893 2.1354

Refer to Table 4 for explanation of hotel numbers.

Lance Wilensky and Francis Buttle

G_lTT’lAN-LI~:G~lES’ ZOEFFICIENT 3F ALIEJA,ION : (3.00000 I:4 3 ITERATIONS. KRJSKAL'S SlRESS : 0.00000 _____________________~~~~~___________________________

4 I , I 4

1 6

5

4

,

Fig. 3. Reduced space perceptual plot. (Refer to Table 4 for explanation of hotel numbers.)

perceptions. Figure 3 contains the computer print-out. Since the axes may be rotated at will they are not dra\vn in. As expected. since the respondents vvere all Holiday Inn guests. the distances betvveen the position of Holiday Inn and the other hotels in the market are large. A survev sampling guests from other hotels would be unlikely to achieve this result.

This multidimensional analysis has rested on the assumption that Holiday Inn competes in a market segment shared with the other operators. There are many ways of segmenting the market for hotel accommodation (Buttle. 1956) and depending upon which method is chosen the assumption may be held valid or rejected. For example. if segmented by travel motive. all the hotels in this study would very probably identify the business traveller as the

Hotef benefit bundles and choice trade-offs

HOTELS 1

I I i 6

I f

Fig. 4. Hotels clustered by benefit bundle.

39

prime target. If segmenting by income, it is probable that the hotels are somewhat differently targeted.

Since it is axiomatic in marketing that customers buy products because of the benefits they are capable of deliverin g, benefit segmentation has become widely applied (Calotone and Johar, 195-t). Cluster analysis of the data collected in the benefit questionnaire enables the validity of the assumption that the hoteIs compete in the same segment to be tested. This technique allows management to determine which hotels are most similar in the benefit bundle trade-off. The greater the similarity, the more the guest views them as substitutes.

The output of this procedure is in the form of a dendrogram as shown in Fig. 3. This shows that the Holiday Inn guest perceives four main clusters of hotels in the Heathrou market:

Cluster 1. Sheraton Heathrow, Sheraton Skyline Cluster 2. Holiday Inn Cluster 3. Cresta Heathrow and THF Excelsior Cluster 4. Penta Heathrow and THF Posthouse. At first appearance it would seem that the Holiday Inn stands alone, however, at the

second level. clusters 1 and 2 join together, as do clusters 3 and 4, thus indicating that Holiday Inn competes in the same market segment, defined in terms of benefit bundles, as the two Sheratons. Reference back to Fig. 3 shows how these three hotels are grouped to the right of the plot.

Summary and comment

This paper reports research which explores the applicability of multivariate techniques to hotel marketing management problems. The main findings can be summarised as follows:

l hfultivariate analysis can contribute towards unravelling the complexities of consumer behaviour.

l The 40 attributes rated by Holiday Inn guests in the stage 1 survey were condensed by

the use the principaf component method into seven factors: standard of personal service, physical attractiveness, opportunities for reiasation. standard of services, appealing image, value for money. suitability for business guests. Convenience of location was also identified as a key factor but was excluded from further analysis on the grounds that all hotels in the research were competing in the same geographic area.

* These factors were deemed to be the fundamental benefits which respondents sought in a hotel. and were then used to campile a second questionnaire.

* From the resuits a benefir vnlrre itzde.~ (BVI) was constructed lvhich picked out the relative importance of these factors in the hotel choice decision, In order of importance they were standard of personal service. physical attractiveness, opportunities for relaxation, standard of services. appeafing image, value for money and suitability for business guests.

* An inde.x ofpo-eeiwd iirsirrrmePrtirlitv (PI) w.vas constructrd which measured the extent to which hotels in the Heathrow market were perceived as capable of delivering the benefits sought by HoIiday inn guests.

* The IPI of the ‘perfect’ hotel for Holiday Inn guests was constructed. The percerrtnge perftm scores (i.e. the prosimity of a hotel’s IPI to the ideal IPI) can provide the basis for competitive comparisons and intraunit management control.

* The BVI and percentage per@ct scores can be plotted graphically to provide visual guidance to management on the relative perceived ability of tlvo competing hotels to deliver perfectly on the customer-satisfying benefits.

* A more compact it~&x of ~~~s)~i~~l~I~r~t~ of a!1 competing properties is provided by a muitidimensiona~ scaling procedure called r~~~{~~~~~~~c~ ~~s~i~oFzi~z~ rznafysis.

* ‘fn order to assess which properties were competing to provide more or less the same bundle of benefits, cluster analysis was applied to the stage 2 data. The rechnique revealed that Holiday Inn was competing against the txo Sheraton units.

Despite the insights gained into the value of the methodology to hotel marketing management. the results of the research should be treated cautiously for a number of reasons:

* The project assumes a complex-buying behaviour model of consumer choice. Some guests may follow other than a learn-feel-do process.

* The project assumes that the guest is responsible for making the choice to stay in the hotel. This is unlikely to be universally true.

l The titles assigned to the facrai-s are arhitrar~. m The surveys coftected data from Holiday Inn guests only, and the results therefore

should not be regarded as typic& of ail guests staying in competing properties. l The sample size was restricted for budgetary reasons. An acceptable number of vafid

questionnaires for this study to be considered statisticaliy significant woutd be in the region of 3000; given a response rate of 26%. some 12000 questionnaires \vould need to be distributed!

* We have no idea whether the 26% returned are typical of all Holiday Inn guests.

References

Hotel benefit bundles and chorce trade-offs 41

Buttle, F. A. (1986) Hotel and Foodservice Marketing: A Managerial Approach. Casseil. London. Calatone, R. and Johar. J. (1981) Segmentation of the tourism market using a benefit segmentation

framework. Journal of Travel Research, Fall. Coxon, A. (1982) A User’s Guide to Multidimensional Scaling. Heinemann. London. Lewis, R. C. and Pizam, A. (1981) Guest surveys: a missed opportunity. Cornell Hotel and

Restaurant Administration Quarterly 22(3), November, 3234. Lewis, R. C. (1982) Positioning analysis for hospitality firms. International Journal of Hospitality

Management l(l), 115-118. Lewis, R. C. (1983) Getting the most from marketing research. Cornell Hotel and Restaurant

Administration Quarterly 24(3), November, Sl-85. National Opinion Polls (1981) Usage and attitudes towards hotels, unpublished research data.