Cluster analysis was used to classify respondents who preferred hospitals for inpatient care to identify hospital preference segments. The clustering was based on the reasons respondents gave for preferring a hospital. The demographic profiles of the grouped respondents were compared to learn whether the segments could be identified efficiently.
The k·Means clustering method (SPSS) was used for grouping the respondents based on their answers to the hospital preference items. The squared euclidean distances between all clustering variables were minimized. Because different individuals perceive scales of importance differently, each individual’s ratings were normalized before clustering. The results indicated that the respondents could best be classified into four clusters. The cross-validation procedure for cluster analysis was run twice, on halves of the total sample.
As expected, the four groups differed substantially by their distributions and average responses to the reasons for their hospital preferences. TIle names assigned to the four groups reflected the demographic characteristics and reasons for hospital preferences: Old-Fashioned, Affluent, Value Conscious, and Professional Want-It-Alls.
Applications of TwoStep Clustering
The data of Table 20.1 were also analyzed using the TwoStep procedure in SPSS. Since all of the variables were continuous, we used the euclidean distance measure. The clustering criterion was the Akaike Information Criterion (AIC). The number of clusters was determined automatically. The results are shown in Table 20.5. As can be seen, a three-cluster solution was obtained, similar to that in hierarchical and nonhierarchical clustering. Note that the AIC is at a minimum (97.594) for a three-cluster solution. A comparison of cluster centroids in Table 20.5 will! those in Table 20.3 show that Cluster 1of Table 20.5 corresponds to Cluster 2 of Table 20.3 (hierarchical clustering), Cluster 2 of Table 20.5 corresponds to Cluster 3 of Table 20.3 and Cluster 3 of TwoStep corresponds to Cluster 1. The interpretation and implications are similar to those discussed earlier. In this case all three methods (hierarchical, nonhierarchical, and TwoStep) gave similar results. In other cases, different methods may yield different results. It is a good idea to analyze a given data set using different methods to examine the stability of clustering solutions.
Sony: Attacking the Market Segment by Segment
Headquartered in Tokyo, Sony Corporation is a leading manufacturer of audio, video, communications, and information technology products for consumers and professional markets. Sony’s main U.S. businesses include Sony Electronics, Inc, Sony Pictures Entertainment, Sony BMO Music Entertainment, Inc., and Sony Computer Entertainment America, Inc. Sony recordeq consolidated annual sales of $88.7 billion for the fiscal year ended March 31, 2008. Sony’s consolldated sales in the United States for the fiscal year ended March 31, 2008, were $29 billion. According to the Harris Poll (www.themarketingresearch.com). Sony has been chosen as the number one brand for the third year in a row,and for the fifth time in the last eight years
The main focus of Sony’s marketing strategy is to get closer to the consumer. Ryoji Chubachi. SonyElectronicsCEO. says,”It is hard to different Iat products today. Differentiation is our mission.We must see the customer first, sells much products to those specific consumers as possible through the right retail channels.” In order to implement this idea, Sony has divided its target market into the following segments: AffluentAlphas (early adopters), Zoomers (55+), SmallOfIicelHomeOffice, Young Professionals (25-34), Families,andGenerationY(under30)
These six demographic segments make up the Diamond Plan. Instead of individual products being marketed by product managers, Sony plans to assign executives to these demographic segments. This new approach to marketing will affect product development and design, retail merchandising, advertising. and consumer loyalty programs. Media dollars will also be adjusted to effectively target the new segments. However.before a full-scale implementation of the DiamondPlan.RyojiChubachi would like to determine if there is a better way to segment the U.S.electronics market.suchasby psychographics and lifestyles.that will lead to increased sales and market share
The Marketing ResearchDecision
- What data should be collected and how should they be analyzed to segment the U.S. electronics market based on psychographics and lifestyles?
- Discuss the role of the type of research you recommend inenabling Ryoji Chubachi to increase sales and market share.
The Marketing Management Decision
- What new strategies should Ryoji Chubachi formulate to increase sales and market share?
- Discuss how the marketing management decision action that you recommend to Ryoji Chubachi is influenced by the data analysis that you suggested earlier and by the likely findings.
Segmenting the Market for Nordstrom
Conduct an Internet search using a search engine Ami your library’s online database to obtain information on the type of consumers who shop at high-enddepartmentstoressuchasNordstrom. Describe what data should be collected and how they should be analyzed to segment the market for a high-end department store such as Nordstrom.
As the CEO of Nordstrom, how would you segment your market?
Sometimes cluster analysis is also used for clustering variables to identify homogeneous groups. In this instance, the units used for analysis are the variables, and the distance measures are computed for all pairs of variables. For example, the correlation coefficient, either the absolute value or with the sign, can be used as a measure of similarity (the opposite of distance) between variables.
Hierarchical clustering of variables can aid in the identification of unique variables, or variables that make a unique contribution to the data. Clustering can also be used to reduce the number of variables. Associated with each cluster is a linear combination of the variables in the cluster, called the cluster component. A large set of variables can often be replaced by the set of cluster components with little loss of information. However, a given number of cluster components do not generally explain as much variance as the same number of principal components. Why, then, should the clustering of variables be used? Cluster components are usually easier to interpret than the principal components. even if the latter are rotated? We illustrate the clustering of variables with an example from advertising research.