![[JIAHR logo]](http://scholar.lib.vt.edu/ejournals/JIAHR/JIAHR-logo.gif){kind=logo}
| April 1993 | Issue 6 | ISSN 1052-6099 |
|---|
Editor: Mahmood A. Khan
Department of Hospitality and Tourism Management
Virginia Polytechnic Institute and State UniversityManaging Editor: Eliza Tse
Virginia Polytechnic Institute
& State University
JIAHR publishes refereed papers on all aspects of hospitality and tourism research. When judged of sufficient quality, individual papers are sent electronically as a single issue of the Journal to members of the Academy of Hospitality Research and to subscribing individuals and libraries. The material is copyrighted. JIAHR is indexed/abstracted in "Lodging and Restaurant Index" and "Leisure, Recreation and Tourism Abstracts."
HOTEL RE-POSITIONING: AN ILLUSTRATION
Stowe Shoemaker
CONTENTS
- Editor's Note
- Abstract
- Key Words
- Introduction
- Methodology
- Results
- The Current Position
- Investigating Future Positions
- Positioning within a Realistic
Competitive Framework- Summary
- References
- EDITORIAL BOARD
- ARCHIVAL INFORMATION
- INSTRUCTIONS TO AUTHORS
EDITOR'S NOTE: This issue marks the first time that JIAHR, as an electronic journal, is publishing an article with graphics. To illustrate the author's thesis are six figures that can be accessed as Postscript files or through ftp (file transfer protocol). To obtain the figures electronically, type and enter the exact "user responses" in the right column, when given the prompts on the left.prompt user response -------------------------------------------------------Interested readers who do not have access to these resources may write the publisher to obtain hard copies of the graphics. Since this is an experiment in a new publishing medium, we will be interestsed in readers' constructive comments.* ftp borg.lib.vt.edu Name (): anonymous anonymous Password: + ftp> cd /pub/JIAHR/figures/jiahr6 ftp> dir ftp> get figure1.ps ftp> get figure2.ps ... ftp> get figure6.ps ftp> quit -------------------------------------------------------- * Contents of <> will vary depending on your local computer. + Replace with your local userid. These files can be printed on any Postscript compatible laser printer. Web-author's note: The forementioned is no longer necessary. Simply click the links when you come to the Figures.
ABSTRACT This article demonstrates how to perform a repositioning study. Specifically, using both multidimensional scaling (MDS), property fitting, and discriminant analysis, this article shows how to determine one's present position and then decide the new position to occupy. Data comes from a study conducted upon local residents in Las Vegas, Nevada.
KEY WORDS: Hotel repositioning, multi-dimensional scaling, perceptual maps, property fitting
INTRODUCTION In their book MARKETING LEADERSHIP IN HOSPITALITY, Lewis and Chambers (1989) state five critical steps needed to reposition a hotel or restaurant. These steps include determining the present position, deciding on the position to occupy, making sure the product is truly different for the repositioning, initiating the repositioning campaign and, finally, remeasuring to see if the position has significantly changed. Of the five, this author argues that the first two are the most critical. For if you do not know how the market currently perceives you, it is hard to determine which direction to move. After all, you already may be right where you belong.
Because the emphasis of their book was on marketing and not research, Lewis and Chambers did not delve into the specifics of how to perform repositioning, nor did they give an example of such a study. It is the goal here to address this issue. Specifically, using both multidimensional scaling (MDS), property fitting and discriminant analysis, this article will show how to determine one's present position and then decide the new position to occupy.
The example used here is that of a study done to provide the Las Vegas hotel and casino community with valuable information on their most captive audience -- local residents. Of specific interest was the development of perceptual maps to show how the hotels are "positioned" relative to each other in local consumers' minds.
THE STUDY Prior to beginning a positioning study, it is first necessary to determine the following:
The competitive framework (e.g. which brands also will be judged)The criteria (variables) on which the brands are judged
And, who is to do the judging
Since a change in any of these can affect the relative positions of the measured brands, it is important they be addressed at the outset. In the example presented, the competitive framework for the positioning was the 31 major hotels/casinos in Las Vegas, NV. In terms of the attributes of interest, all hotels were evaluated on a battery of 25 variables ranging from opinions about the gaming and atmosphere to the type of clientele and the suitability of the food service. In addition, each hotel was rated on the variables "good place overall" and "a place I now go frequently." Ratings were based on a ten-point scale, where "1" means "does not describe at all" and "10" means "describes perfectly."
Due to cost considerations, attributes were generated by looking at past research, discussion with colleagues, and approximately thirty unstructured interviews with Las Vegas residents. Ideally, however, it is also useful to conduct focus groups or other types of qualitative research after development of the formal list to ensure that all important items have been included.
METHODOLOGY Data for this article were taken from a major telephone study regarding the gambling habits of Las Vegas residents and their attitudes towards various local hotels. This project was conducted by I/H/R Research Group -- a full service marketing research firm headquartered in Southern California -- during the months of September and October, 1989.
Potential respondents were selected from Las Vegas and the surrounding suburban area using random digit telephone numbers. This type of sampling methodology ensured the inclusion of households with unlisted phone numbers -- estimated at approximately 57% in the Las Vegas area. The sampling supplier was Scientific Telephone Samples, a national supplier of computer generated random digit phone samples.
Upon the initial contact, all respondents were screened to ensure that they met the following qualification criteria:
At least 21 years of ageParticipate in any type of gaming -- including video poker, table games, sports book or slots in the Las Vegas area
And, gamble at least a few dollars per month
In addition, the sample was split approximately equally by sex. Altogether, 558 interviews were completed.
To meet the objectives of the study, the questionnaire addressed the following major issues:
Time spent gambling in an average monthTime spent gambling at one "session"
Amount of average bet
The hotel/casino where residents gamble most often -- and why
Other hotels/casinos which locals gamble at frequently
And, structured attribute ratings -- measuring such things as friendliness, type of atmosphere (elegant and sophisticated versus casual, relaxed), and gaming concerns (e.g. odds, cleanliness, etc.)
Due to the large number of hotels involved (31), the rating sequence was set up on a balanced "incomplete block" design. This design allowed respondents to rate a more manageable subset of hotels (5), while ensuring that all of the subsets could be aggregated, as necessary for analysis purposes. (For a complete discussion of balanced incomplete block designs, see Rink (1987) and Cochran and Cox (1957)).
Multidimensional scaling (MDS) was the statistical technique used to examine the relationship between hotels. In specific, the MDS algorithm KYST -- acronym for Kruskal, Young, Shepard and Torgerson, the developers of the model -- was used to create the information needed to visually represent consumers' current perceptions of the hotels.
The KYST algorithm "works" by taking the "distance" between the hotels -- created from the differences in mean scores between the hotels on the 25 attributes -- and constructing a set of coordinates that accurately represents these distances. These coordinates are then used to plot the hotels in perceptual space.
RESULTS The output (the coordinates) from the KYST program appears in Table I. The distance matrix is not shown due to its size, although it is available from the author.
TABLE I COORDINATES DERIVED VIA KYSTHotels Under Study
Sahara -.339 .018 Stardust -.243 -.092 Las Vegas Hilton -1.260 .358 Circus Circus -.650 -.882 Caesar's Palace -1.620 .513 Desert Inn -.784 .429 Tropicana -1.101 .210 Landmark 1.174 .380 Riviera -.475 .137 The Dunes -.167 .305 Bally's -.886 .692 Frontier -.061 .010 Imperial Palace .170 .397 The Sands -.190 .108 Vegas World 1.973 .754 Gold Coast -.344 -1.021 Show Boat .600 -.485 Continental 1.825 .229 Lady Luck 1.072 .006 Golden Nugget -1.590 .068 Flamingo -.806 .267 Maxim .274 .306 Hacienda .516 .285 Binion's Horseshoe .531 -.806 Holiday Inn/Strip .157 -.030 Fremont 1.220 .002 Barbary Coast .160 -.098 Four Queens .445 .062 Union Plaza .557 -.118 Sam's Town -.173 -.933 Palace Station 016 -1.069
Generally speaking, there are two methods of interpretation of an MDS perceptual map. The first is subjective in nature. This approach begins with looking at the properties of the hotels occupying extreme positions in the derived space. For these, one attempts to identify the attribute, or attributes, that can explain the relative positions.
The second method is termed "property fitting." Property fitting places the 25 attributes in the same perceptual space as the hotels, thus aiding in interpretation of the hotels' positioning. It does this using multiple regression. Essentially, the mean rating of each of the 27 measured attributes is taken individually and regressed on the derived space coordinates (from the KYST algorithm). One important by-product of the regressions, besides the R2 which shows how important each variable is in defining the perceptual space, is the beta weights -- one for each stimulus point. These beta weights are used to calculate the coordinates for the attribute vector. (The software program MDS PC (Smith 1986) automatically gives you directional cosines as part of its output. However, these can easily be computed by hand. See Dillon and Goldstein (1984), page 151, for an explanation on how this is done. ) The coordinates for the property fitting appear in Table II.
TABLE II PROPERTY FITTING COORDINATESAttributes
Friendly, courteous service -.794 -.608 Good food specials and buffets -.581 -.814 Safe place to go at night -.995 -.096 Games and tables are well-maintained -.998 -.069 Offers the kind of games or gambling I like to do -.541 -.841 Good place for the small-stakes player .304 -.953 Good place to take out-of-town guests -.962 -.274 Serve plenty of good free drinks -.421 -.907 Good atmosphere and decor -.993 .119 Good parking always available -.739 -.674 It's always clean and well-kept -.992 .128 A good place for "high rollers" -.898 .439 Popular with young people -.460 -.888 Fun and exciting atmosphere -.859 -.513 Popular with locals -.016 -.999 Better odds there -.169 -.986 Popular with older people -.249 -.969 Casual, relaxed atmosphere -.114 -.993 Good check cashing policy -.171 -.985 Realistically, a hotel where I would stay if I was visiting Las Vegas -.981 .194 Popular with tourists -1.000 -.006 Always have good entertainment in the bars or good lounge acts -1.000 -.010 Elegant and sophisticated atmosphere -.809 .587 Always have better luck -.335 -.942 A good place for foreign travelers -.835 .550 Good place overall -.922 -.388 A place I now go to frequently -.445 -.896
THE CURRENT POSITION The outcome of the MDS/KYST application is illustrated next, on two perceptual maps -- Figure 1 for the KYST results and Figure 2 for the KYST results and the property fitting. Both maps were constructed using a set of axes whose origin represents a theoretical average or neutral zone. The rating for the variable "good place overall" is labeled for reference purposes.
The first perceptual map reveals the relative positions of the hotels without any vectors. As can be seen, there appear to be four clusters of hotels. Those positioned in the upper left quadrant have the best ratings for "good place overall" (average score around 8.4), while those in the upper right have the lowest (4.5). In contrast, the two groups in the upper and lower center earned scores which were neither high nor low (approximately 6.5). As a means of understanding the relative positions, it is useful to think of the four clusters in terms of automobiles.
For example, based on prior knowledge about the hotels, it would appear that the hotels in the upper left represent Cadillacs/Mercedes, those in the upper center Oldsmobiles, while those in right quadrant are Chevrolets. Given the "down-home" feelings of those hotels in the lower center, one could say these are the "Recreational Vehicles/Pick Up Trucks" segment. Thus, we can hypothesize that hotels in the upper left cater to the high rollers or type of people who buy expensive cars and those attracted to the "glitter." On the other hand, those who would visit the "Chevrolets" would want to avoid the "glitter" and are looking more for the basics, while those in the lower center are looking more for "down home" feelings. Similar interpretations can be made for the other group.
A second way to interpret this map is to look at the distance between hotels. The closer the hotels are to each other, the more they are perceived to be similar. Conversely, the further apart they are, the more dissimilar. In this sense, from a local residents perspective, Caesar's Palace is in the same class as the Golden Nugget, Las Vegas Hilton, the Tropicana, and to a lesser extent Bally's, Desert Inn, and the Flamingo. It is NOT in the same class as Vegas World, the Landmark, the Continental, the Fremont or Lady Luck. Similarly, Sam's Town, Gold Coast, Circus Circus, Palace Station, Show Boat, and Binion's Horseshoe are more like each other.
With this idea of distance in mind, it is useful to use the created map to identify one's competition. After all, similar hotels are more likely to compete with each other than dissimilar ones -- especially if consumers have already decided to visit a certain class or style of hotel. Not surprisingly then, Lady Luck's competition is more likely to be the Fremont and/or the Continental and not necessarily Caesar's or the Golden Nugget.
It should be noted that Las Vegas's newest major hotels -- The Mirage, The Rio and Excalibur -- all opened just after interviewing for this study was finished. Therefore, they are not shown on any of the perceptual maps. It can be hypothesized, however, that The Mirage and, possibly, The Rio would be positioned in the upper left quadrant near Caesar's Palace, the Hilton, and Golden Nugget. Excalibur, on the other hand, might be positioned nearer to Circus Circus or, perhaps, in the mid-range cluster.
Additionally, the Riveria was in the processing of remodeling and upgrading during the time of interviewing. Thus, their position may be different now.
To further understand the positions of the hotels, it is necessary to look at Figure 2. Again, Figure 2 shows the results of the property fitting. The positioning of the attribute vectors allows a more meaningful interpretation of the perceptual map by showing which attributes were useful in separating the hotels.
Although the variables "a good place overall" and "a place I now go frequently" were not used in creating the distance matrix, they were included in the property fitting model. This is possible because the property fitting algorithm regresses each attribute individually.
The influence of these attribute vectors on the placement of the hotels is seen by drawing a perpendicular line from the hotel to each vector. The further from the origin this crossing occurs, the more the hotel(s) possess that attribute -- a check of the average scores reveals this to be true. While the vectors are shown pointing only in the positive direction, it is useful to remember that they continue in the negative direction, as well.
For example, looking at the upper left quadrant of the map, it can be seen that these hotels occupy this position because consumers believe them to:
Have elegant and sophisticated atmosphere (w)Be good places for foreign travelers (y)
Be good places for "high rollers" (l)
And, realistically would be hotels they would stay in if visiting Las Vegas (t)
In contrast, the hotels in the upper right quadrant are believed to possess LESS (i.e. lower ratings) of these attributes -- hence their opposite placement. And, the hotels clustered around the origin are thought to possess just average amounts of these attributes.
The hotels positioned in the lower center (Sam's Town, Gold Coast, Circus Circus, Palace Station, Show Boat, and Binion's Horseshoe) are there because, among other things, they are considered:
Popular with young people (m)Serve plenty of free drinks (h)
Popular with locals (o)
Have casual, relaxed atmosphere (r)
And, offer better odds (p)
However, within this cluster, Circus Circus is more likely to be considered to have "good food specials and buffets" (b) or "offer the kind or games or gaming I like" (e) than either Binion's Horseshoe or the Show Boat -- as evidenced by the direction of the attribute vector. Again, hotels positioned in the opposite direction are believed to offer less of these things.
INVESTIGATING FUTURE POSITIONS With the current positions defined, information on these maps can now be used to develop potential new positioning strategies. For instance, should any of the hotels positioned in the center cluster wish to be more like Caesar's, they need to convince local residents that they are a good place for "high rollers", a "good place for foreign travelers", and have "elegant and sophisticated atmosphere." Or, if they wish to separate themselves from both the center group and those in the upper left, they could try to position themselves between the Golden Nugget and Circus Circus. For this strategy, the hotels would need to develop programs geared to convince local residents that they have a "fun and exciting atmosphere," "good parking always available," and "friendly courteous service."
Another possible positioning strategy involves the hotels in the upper right quadrant (the Landmark, Lady Luck, etc.). Three viable options would seem to be available for any of these hotels. They can stay where they are, they can try to be more like the hotels in the center, or they can reposition to the lower right near the Show Boat and Binion's Horseshoe.
Should they decide to move more toward the center, they need to convince local residents that:
Their games and tables are well maintainedTheir hotel is a safe place to go at night
They always have good entertainment in the bars or good lounge acts
They have good atmosphere and decor
And, that their hotel is always clean and well-kept
On the other hand, should they wish to move toward the Show Boat and Binion's Horseshoe, they might consider plans that convince locals that they:
Serve plenty of good free drinksAre a good place for the small-stakes player
Are popular with young people
Have a good check cashing policy
Will have better luck there
Have good food specials and buffets
And, offer the kind of games local residents like to play
Following the same kind of logic, different repositioning strategies can be instituted for any of the hotels shown on Figures 1 and 2.
POSITIONING WITHIN A REALISTIC COMPETITIVE FRAMEWORK While it is useful to look at the market as a whole, it is important to remember that hotels tend to compete not so much with all hotels in a given market but, rather, with hotels in their same class. As a means of gaining further insight, a series of discriminant analyses were run on each hotel cluster. The specific goal of this analysis was to identify the strengths and weaknesses of each hotel property within the competitive framework defined via the MDS solution.
Discriminant maps are interpreted much the same way as MDS maps. The attributes are represented by vectors and their influence on each hotel is determined by drawing a perpendicular from the hotel to each vector. Again, these vectors can be thought of as having both positive and negative direction.
One difference with discriminant maps, however, is that the relative length of each vector is important. The longer the vector, the more important that variable is in separating the brands. Conversely, a relatively short line suggests that the hotels cannot be differentiated by that attribute.
In order to test how well these maps work -- that is, how different the hotels really are in terms of the measured attributes -- confusion matrices were used to summarize the number of correct and incorrect classifications made by the discriminant procedure. The Proportional Error Reduction (PER) method was then used to test whether the confusion matrices were good or bad. This formula is defined as :
PER = D-P/N-P
where: D=the number of correct assignments with the discriminant function N=the total number of subjects
P=the number of correct assignments under optimal prediction without the discriminant function (Optimal prediction is equal to the subjects in the largest group.)
The PER tells the proportional error reduction achieved by using the discriminant functions to classify groups. Clearly, the higher this is the better. The discriminant function classification should also outperform chance alone (Lewis 1984).
The power of the discriminant analysis was also tested a second way. Specifically, a formula developed by Tatsuoka (1970) shows the total discriminatory power residing in the discriminant functions or, equivalently, in the predictor battery as a whole. The higher this percentage, the better the independent variables are at separating the hotels. Tatsuoka's formula is defined as follows:
2 N w multi. = 1 - _______________________________ (N-k)(1+x1)(1+x2)....(1+xn) + 1
Where N= the sample size k= the number of groups x1, x2, .. xn are the eigenvalues found in computing the discriminant functions
The analysis for each hotel cluster, along with their perceptual map, will be discussed in separate detail.
THE RECREATIONAL VEHICLE/PICK UP TRUCK CLUSTER OF HOTELS The cluster with the greatest differences among the hotels was the "Recreational Vehicle/Pick Up Truck" segment. As shown in Table III, approximately 80% of the difference was explained by the measured attributes and 55% of the cases were classified correctly. (For perspective, the PER percentage was 46% compared to 17% by chance alone.) Figure 3 looks at this cluster of hotels.
Table III Discriminant Analysis Of Hotel Clusters
Discriminant Eigenvalue Canonical Wilk's Chi Sig. Function Correlation Lambda Sq
.221 792.5 .000
1 .923 .693 .424 449.5 .000 2 .603 .613 .680 202.2 .000
Centroids (group means) Function 1 Function 2
Showboat .279 .039 Binion's Horseshoe .047 -1.340 Palace Station .669 .435 Sam's Town .203 -.053 Gold Coast .628 .187 Circus Circus -1.827 .734
Significant Variables
Friendly, courteous service .155 -.120 Good food specials and buffets .052 .089 Safe place to go at night .083 -.011 Games and tables are well maintained .051 -.180 Good place for the small-stakes player .110 .064 Good place to take out-of-town guests -.097 -.052 Serve plenty of good free drinks .099 -.061 Good parking, always available .077 -.026 It's always clean and well-kept .198 -.145 A good place for "high rollers" .021 .644 Popular with young people -.285 .235 Fun and exciting atmosphere -.205 -.069 Popular with locals .307 .022 Better odds there .146 .096 Popular with older people .365 -.145 Casual, relaxed atmosphere .214 .033 Good check cashing policy .234 .217 Realistically, a hotel where I would stay, if I was visiting Las Vegas .107 .111 Popular with tourists -.466 .150 Always have good entertainment in the bars or good lounge acts .004 .166 Elegant and sophisticated atmosphere .168 .004 A good place for foreign travelers -.347 .129 A place I now go to frequently .194 .143
Classification matrix revealed that 55% of the cases were classified correctly. Proportional error reduction (PER) = 46% compared to 17% by chance alone. Tatsuoka's Explained Variance = 78%
The hotels with the most differentiation are Binion's Horseshoe, Circus Circus, Palace Station, and Gold Coast. As can be seen, Binion's Horseshoe is separated from the other hotels by the belief that it is a good place for "high rollers" (l). In relation to the other hotels, it is NOT considered to have good entertainment in the bars (v), be popular with foreign travelers (y), or young people (m). Circus Circus's strengths lay in the belief that it is popular with tourists (u), is a place for foreign travelers (y), popular with young people (m), and, to a lesser extent, has a fun and exciting atmosphere (n).
On the other hand, Palace Station and Gold Coast appear to have good check cashing policies (s), be popular with locals (o), are places respondents currently go (aa), and be popular with older people (q).
The Show Boat and Sam's Town, located near the origin, are not as differentiated from the other hotels. To separate themselves, one possible strategy would be to move further toward the lower right -- between Gold Coast and Binion's Horseshoe. To do this, the areas they need to work on include their check cashing policy, cleanliness, the perception that they are good for older people, and that they have friendly, courteous service. The result of that directional move would be to capture more business from those two competitors. However, it is likely that cross- over with Circus Circus would be diminished.
Within their own competitive framework, Figure 3 also suggests that each hotel -- with those two exceptions -- has a fairly clear position. Thus, if any wish to reposition, it would be best to do so within the framework of the whole market and not just within their own specific segment.
The remaining three hotel clusters did NOT exhibit the same strength of differentiation as the "Recreational Vehicles/Pick Up Truck" segment . Specifically, within each cluster only about 44% of the measured differences was explained by the 27 attributes. Similarly, PER percentages were in the 24% to 40% range -- higher than chance alone, but not terrific. Despite this, useful information can be gained from each of the following discriminant maps.
Table IV Discriminant Analysis Of Hotel ClustersDiscriminant Eigenvalue Canonical Wilk's Chi Sig. Function Correlation Lambda Sq
.490 437.0 .000 1 .291 .475 .633 280.7 .000 2 .164 .375 .736 187.6 .000
Centroids (group means) Function 1 Function 2
Caesar's Palace .659 -.131 Las Vegas Hilton -.035 .059 Tropicana -.354 .046 Golden Nugget .543 .779 Bally's .269 -.552 Desert Inn -.459 -.325 The Flamingo -.622 .122
Significant Variables Good food specials and buffets .063 .760 Safe place to go at night -.122 -.002 Games and tables are well maintained .212 .121 Offers the kind of games or gambling I like to do .051 .225 Good place for the small-stakes player -.105 .166 Good place to take out-of-town guests .291 .222 Serve plenty of good free drinks -.196 .181 Good atmosphere and decor .270 .159 Good parking, always available -.106 .038 It's always clean and well-kept .308 .017 A good place for "high rollers" .404 .005 Popular with young people .072 .305 Fun and exciting atmosphere .174 .227 Popular with locals .078 .454 Better odds there -.151 .436 Popular with older people -.226 .199 Casual, relaxed atmosphere -.325 .297 Good check cashing policy -.045 .403 Realistically, a hotel where I would stay, if I was visiting Las Vegas -.341 .206 Popular with tourists .250 .041 Always have good entertainment in the bars or good lounge acts .332 .217 Elegant and sophisticated atmosphere .445 .033 A good place for foreign travelers .182 .047 Good place overall .147 .073 A place I now go to frequently .092 .403
Classification matrix revealed that 40% of the cases were classified correctly. PER = 29% compared to 14% by chance alone. Tatsuoka's Explained Variance = 47%
Cadillac/Mercedes Cluster of Hotels
Figure 4 reveals how the "Cadillac/Mercedes" group of hotels (those in the upper left quadrant) are differentiated on the 27 measured attributes. Although these hotels are perceived to be similar when compared to all hotels in Las Vegas, this map suggests that differences between the brands DO exist.
For example, both Caesar's Palace and the Golden Nugget are more likely to be thought of as a place for "high rollers"(l) and as having "elegant and sophisticated atmosphere" (w). In contrast, the Flamingo and the Tropicana are more likely to be thought of as having "a casual, relaxed atmosphere" (r), being "popular with older people" (q), and "serving plenty of good free drinks" (h).
Unlike Caesar's Palace, the Golden Nugget is considered much more popular with locals (o). This may be because it is believed to have a good check cashing policy (s), good food specials and buffets (b), and better odds (p).
The Las Vegas Hilton's positioned in the center suggests that it scores equally well on all attributes. As such, it has no clear position of its own among local residents. To separate itself from the pack, Hilton might consider repositioning opposite the Golden Nugget in the upper letter center. The way to accomplish this -- as evidenced by the attribute vectors -- would be to improve the perception of the food specials and buffets (b), offer better odds (p), and offer a good check cashing policy (s). Of course, a similar strategy could be followed by the Tropicana.
Bally's and the Desert Inn are unique in the sense that no vectors point in their direction. However, Bally's is more associated with the attributes that define Caesar's, while the Desert Inn is more associated with the Tropicana and the Flamingo. Notice that both are perceived very different than the Golden Nugget. The lack of vectors pointing in their direction suggests that they are defined more by what they don't offer than by what they do. For instance, in relation to the other destinations in this cluster, it appears that Bally's is not considered to have good food specials, better odds, a good checking cashing policy, or be popular with locals. Another possible reason for the lack of vectors pointing toward Bally's and Desert Inn is that some variables which help distinguish these hotels were not included in the survey.
The Chevrolet Cluster of Hotels
Figure 5 examines the relationship between the "Chevrolet" group of hotels -- those positioned in the upper right quadrants of Figures 1 and 2.
Table V Discriminant Analysis Of Hotel ClustersDiscriminant Eigenvalue Canonical Wilk's Chi Sig. Function Correlation Lambda Sq
.521 284.9 .000 1 .384 .568 .721 142.9 .000 2 .148 .359 .828 82.6 .000
Centroids (group means) Function 1 Function 2
Landmark .198 .638 Lady Luck -.586 -.028 Fremont .004 -.706 Continental .890 .350 Vegas World -.506 -.254
Significant Variables
Friendly, courteous service -.124 .217 Good food specials and buffets -.044 .057 Safe place to go at night .275 .272 Good place for the small-stakes player .132 -.079 Good place to take out-of-town guests -.138 .209 Good atmosphere and decor -.268 .160 Good parking, always available .333 .406 A good place for "high rollers" -.201 -.035 Popular with young people -.123 -.169 Fun and exciting atmosphere -.369 .087 Popular with locals .159 -.138 Better odds there .042 -.304 Casual, relaxed atmosphere .236 -.127 Good check cashing policy -.204 .374 Popular with tourists -.266 -.040 Always have good entertainment in the bars or good lounge acts -.081 .027 Elegant and sophisticated atmosphere -.228 .265 Always have better luck there .024 -.142 A good place for foreign travelers -.290 .162
Classification matrix revealed that 52% of the cases were classified correctly. Proportional error reduction (PER) = 40% compared to 20% by chance alone. Tatsuoka's Explained Variance = 44%
Like Figures 3 and 4, this map shows that within the "Chevrolets" group of hotels significant differences between brands do exist. For instance, both the Landmark and Continental are thought to have the following attributes:
Good parking (j)Safe place to go at night (c)
Good check cashing policy (s)
Additionally, the Landmark and Lady Luck , but not the Continental, are considered to be more in line with:
Have elegant and sophisticated atmosphere (w)Be a place for foreign travelers (y)
Have friendly, courteous service (a)
And be a good place to take out-of-town guests (g)
As comparison, Vegas World and the Fremont are considered to be "popular with young people" (m) and a place "where you can get better odds" (p). Vegas World is also considered popular with tourists (u), while the Fremont is considered to have a casual relaxed atmosphere (r).
Figure 5 suggests that each hotel has a fairly defined position (i.e. there are relatively long vectors pointing in their direction.) Therefore, within this group of hotels, there is not an immediate need to reposition; unless, of course, they wish to reposition towards one of the other hotel clusters.
The Oldsmobile Cluster of Hotels
Figure 6 looks at the "Oldsmobile" group of hotels -- those positioned in the center of Figures 1 and 2.
Table VI Discriminant Analysis Of Hotel Clusters
Discriminant Eigenvalue Canonical Wilk's Chi Sig. Function Correlation Lambda Sq
.473 862.2 .000 1 .242 .442 .587 612.7 .000 2 .137 .347 .667 465.2 .000
Centroids (group means) Function 1 Function 2
Riviera .466 -.169 Sahara .053 .122 Stardust .251 -.114 The Sands .147 -.272 The Dunes .335 -.386 Frontier .232 .088 Imperial Palace .298 -.213 Maxim -.303 -.292 Hacienda -.308 -.270 Four Queens -.128 .441 Holiday Inn -.449 .079 Barbary Coast .066 .233 Union Plaza -.661 .754
Significant Variables
Friendly, courteous service .160 -.106 Good food specials and buffets .165 -.035 Safe place to go at night .184 -.146 Offers the kind of games or gambling I like to do .178 .232 Good place for the small-stakes player .060 .554 Good place to take out-of-town guests .254 -.016 Serve plenty of free drinks .189 .273 Good atmosphere and decor .308 -.155 Good parking, always available .178 .035 It's always clean and well-kept .342 .040 A good place for "high rollers" .818 .039 Popular with young people .157 .076 Fun and exciting atmosphere .141 .173 Popular with locals -.052 .400 Better odds there .156 .294 Popular with older people .160 .194 Casual, relaxed atmosphere .028 .132 Good check cashing policy .145 .106 Realistically, a hotel where I would stay if I was visiting Las Vegas .034 -.106 Popular with tourists .275 .359 Always have good entertainment in the bars or good lounge acts .126 .068 Elegant and sophisticated atmosphere .260 -.071 A good place for foreign travelers .169 .056 Good place overall .133 .103 Always have better luck there .108 .177
Classification matrix revealed that 30% of the cases were classified correctly. Proportional error reduction (PER) = 24% compared to 8% by chance alone. Tatsuoka's Explained Variance = 44%
This map suggests that the variable with the most discriminatory power is "good place for high rollers" (l). This is followed by the variable "good place for the small- stakes player" (f), popular with tourists (u), and popular with locals (o). It would appear that the Union Plaza and the Four Queens are more likely to be thought of as good places for the small stakes player and popular with locals than other hotels in this cluster.
In contrast, the Stardust, Riviera, Imperial Palace, Frontier, are more likely to be thought of as a good place for "high rollers" (l). In addition, Figure 6 suggests that they are also thought to be "safe places to go at night" (c), have "good atmosphere and decor" (i), and "friendly, courteous service" (a).
Again, the lack of vectors pointing in the direction of Holiday Inn/Strip, Hacienda, and Maxim suggest that these hotels have relatively low scores on the 27 variables -- especially the attribute good place for high rollers. This suggests that these three hotels will not do as well with local residents if they concentrate on competing directly with the other hotels in their class.
A possible strategy for these three hotels would be to redefine their competition. Rather than competing with other hotels in the "Oldsmobile Group," they might redirect toward the "Chevrolet" hotels (those located in the upper right quadrant of the first map -- the Landmark, Lady Luck, etc.). Such a move would make them the high end of one group rather than the low end of another. Since they are already positioned to the right of the origin on the market-wide map, this reposition should not be difficult to accomplish.
SUMMARY This paper has shown -- via an example -- the methodology employed to determine the relative position(s) of the major hotels/casinos in Las Vegas. It has then used this information to suggest possible new positions for the various hotels to occupy.
It is important to remember that the sample consisted entirely of Las Vegas residents. No attempt was made to consider the opinions/perceptions of the out-of-state tourists. Keep this in mind when reviewing the findings, as there is a chance that non-locals would position the same hotels quite differently. Therefore, a strategy geared to reposition the hotels for local residents may be inappropriate for the out-of-town guest.
If one wishes to derive a positioning for another target segment, the same methodology could be used to measure tourists perceptions. However, the following caveats are worth noting. First, not only would the sample need to match your population understudy (e.g. list of heavy gamblers, residents of Los Angeles metro, etc. ), but it may be necessary to reduce the number of hotels under study. Since it is unlikely that non-residents would have the same depth of knowledge about the hotels, care must be taken to ensure that respondents are qualified to answer the questions.
Once the current position has been defined and possible new positions have been decided upon, the next steps -- as suggested by Lewis and Chambers (1989) -- are to initiate the repositioning campaign and then repeat the steps discussed to determine if the repositioning campaign worked.
The methods described throughout this paper are useful because they are an easy and relatively inexpensive way to accurately measure consumers' perceptions. Given the magnitude of money involved in repositioning a property (whether it be spent for advertising or physical improvements), it makes sense to use these techniques to see where your property is before deciding where you want it to be.
Stowe Shoemaker, formerly a Vice-President at I/H/R Research Group, a full-service marketing research firm headquartered in Tustin, CA., is currently a PhD candidate at Cornell University. He wishes to thank Robert Lewis, Ph.D., (University of Guelph) and Ronald E. Clark (Senior Partner, I/H/R) for their help with the article.
REFERENCES Cochran, William and Gertrude Cox (1957), Experimental Designs. New York: John Wiley & Sons, Inc. Second Addition.
Davison, Mark L. (1983), Multidimensional Scaling. New York: John Wiley & Sons, Inc.
Dillon, William R. and Matthew Goldstein (1984), Multivariate Analysis. New York: John Wiley & Sons, Inc.
Lewis, Robert C. and Chambers, R. (1989), Marketing Leadership in Hospitality. New York: Van Nostrand Reinhold.
Lewis, Robert C., (1984) "The Market Position: Mapping Guests' Perceptions of Hotel Operations," Cornell HRA Quarterly (August), 86-99.
Rink, David R. (1987), "An Improved Preference Data Collection Method: Balanced Incomplete Block Designs," Journal of the Academy of Marketing Science, 15, No. 1 (Spring), 54-61.
Smith, Scott M. (1986), PC-MDS (Statistical Software), Brigham Young University, Provo, Utah.
Tatsuka, Maurice M. (1970), Discriminant Analysis: The Study of Group Difference. Champaign, Illinois: Institute of Personality and Ability Testing.
THE JOURNAL OF THE INTERNATIONAL ACADEMY OF HOSPITALITY RESEARCH
John Bareham
Brighton Polytechnic, U.K.
Horace A. Divine
University of Denver
Donald E. Hawkins
The George Washington University
Chuck Gee
University of Hawaii-Manoa
Michael Haywood
University of Guelph
Ontario, Canada
Mahmood A. Khan
Virginia Polytechnic Institute
and State UniversityWilliam Kent
Auburn University
Robert C. Lewis
University of Guelph
Ken McCleary
Virginia Polytechnic Institue
and State University
Robert C. Mill
University of Denver
Michael Olsen
Virginia Polytechnic Institute
and State University
Abraham Pizam
University of Central Florida
Brian Wise
Queen Victoria University
Turgut Var
Texas A&M University
SUBSCRIPTIONS For the years 1992 and 1993, electronic subscriptions to JIAHR are offered free of charge. To subscribe, simply send your name and e-mail address (either Bitnet or Internet) to SCHOLAR@VTVM1 (Bitnet) or SCHOLAR@VTVM1.CC.VT.EDU (Internet); you can send it by regular mail to the Publisher, JIAHR, Scholarly Communications Project, University Libraries, Virginia Tech, P.O. Box 90001, Blacksburg, VA 24062-9001. In the past, and probably again in the future, the price of an annual subscription: Institutional $30 (U.S.); Individual $20; Student (with professor's attestation) $10.
PAPERS FROM PREVIOUS ISSUES Issue No. 1: HOTEL YIELD MANAGEMENT USING OPTIMAL DECISION RULES by Ralph D. Badinelli and Michael D. Olsen. November 26, 1990.
Issue No. 2: CONJOINT ANALYSIS AND ITS APPLICATION IN THE HOSPITALITY INDUSTRY by Sophie Ding, Ursula Geschke and Robert Lewis. February 20, 1991.
Issue No. 3: FUNCTIONAL AND SYMBOLIC CONGRUITY APPROACHES TO CONSUMER SATISFACTION/DISSATISFACTION IN TOURISM by Kye-Sung Chon and Michael D. Olsen. May 28, 1991.
Issue No. 4: SOLID WASTE AND DEGRADABILITY: SAVING GRACE OR FALSE PROMISE? by Leslie E. Cummings. December 9, 1991.
Issue No. 5: THE AMERICANS WITH DISABILITIES ACT: PERSPECTIVES ON REASONABLE ACCOMMODATION by Suzanne K. Murrmann, September 8, 1992.
ACCESS TO BACK ISSUES To obtain titles, abstracts or full texts of articles previously published in JIAHR, subscribers should send the appropriate message via e-mail to LISTSERV@VTVM1 (Bitnet) or to LISTSERV@VTVM1.CC.VT.EDU (Internet). To obtain titles, the appropriate message is: GET JIAHR CONTENTS. For abstracts, the appropriate message is: GET JIAHRyy ABSTRACT where "yy" represents the last two digits of the year of publication of the abstracted article. For full text of an issue, the appropriate message is: GET JIAHRn TEXT where "n" represents the number of the desired issue. For more detailed instructions, with examples, on how to obtain this information electronically, send the message: GET JIAHR HELP.
This journal is registered with the Copyright Clearance Center, Inc., 27 Congress Street, Salem, MA 01970, U.S.A. Limited duplication is permitted for academic or research purposes only. Subscribing libraries are permitted to distribute this issue of the journal electronically to institutional faculty, students and employees via local area networks or institutional mainframe computers.
Published for the International Academy of Hospitality Research by the Scholarly Communications Project of Virginia Polytechnic Institute and State University.
INSTRUCTIONS TO AUTHORS "The Journal of the International Academy of Hospitality Research" (JIAHR) welcomes for review original papers on all aspects of hospitality and tourism research. Each paper should be submitted electronically in ASCII format, on disk or as an electronic file, along with the source document on disk and a hard (paper) copy. The editors welcome tables with articles but, because of technical problems in transmission, cannot accept graphs or pictorial illustrations. Papers generally should be written in conformity with the Publication Manual of the American Psychological Association, Third Edition. Because of electronic transmission of articles among editors and referees, a prompt response to submissions can be expected. For further information and/or guidelines about submitting a paper, contact Associate Editor, JIAHR, Room 362, Wallace Hall, Virginia Tech, Blacksburg, VA 24061-0429, USA; telephone 703/231-5515 or 703/231-4922, USA; or send e-mail to JIAHRED@VTVM1 (Bitnet), or JIAHRED@VTVM1.CC.VT.EDU (Internet).
