清华MBA系列课件消费行为学Perceptual Map of Beers Example.docx

1、清华MBA系列课件消费行为学Perceptual Map of Beers ExamplePerceptual Map of Foreign Beers in the PRC: An ExampleBackgroundBecause of its potential to surpass the US as the largest beer market in the world (in terms of volume) by the year 2000, the Peoples Republic of China (PRC) is a very attractive market for e

2、very multinational beer company. Several multinational beer companies are currently operation in China. Some companies have been quite successful while some are still struggling to gain foothold in this market. The example demonstrated in this paper is an actual research that was conducted for a mul

3、tinational beer company. The objective of this study is to investigate the beer consumption behavior of Chinese consumers as well as to investigate the relative positions of several leading foreign beers in these consumers mind. Only the procedure, findings, and managerial implications of the percep

4、tual mapping of the research will be demonstrated and discussed here.Research procedureStep1: Exploratory ResearchThe objective of the exploratory research was to identify product attributes that are important to Chinese beer drinkers. The product attributes identified were then used to design the q

5、uestionnaire for the main study. In the exploratory research, a focus group interview of eight Chinese consumers was conducted to identify attributes or characteristics of beer that are important to them. Altogether, ten product attributes were identified: package design, taste, color, level of alco

6、hol, variety of sizes, price, country of origin, availability, advertising and reputation.Step2: Questionnaire DesignBased on the product attributes identified in the focus group interview, the questionnaire for the main study was designed. (See the questionnaire distributed earlier.) Although five

7、brands of foreign beers were included in the study: Carlsberg, Foster, Guinness, Heineken, and San Miguel. Note that the first ten questions for each brand is the question for measuring the overall attitude toward the brand and the intention to choose the brand. The attitude rating question is neede

8、d for the identification of the “ideal line” on which the ideal position is located by using the attitude toward the brand as the dependent variable and he factor scores as the independent variables. The beta coefficients of the independent variables (which represent the relative important of each f

9、actor score in influencing the attitude) will be used to calculate the angle for the ideal line. The intention rating is needed to measure the relationship between the attitude toward the brand and the intention to buy by using the intention as the dependent variable and the attitude as the independ

10、ent variable. This regression analysis is needed to see whether or not attitude toward the brand is a good predictor of intention to buy the brand.Step3: The Main StudyResearch Design. The research design used in the main study was a sample survey by face-to-face interview.Sample and Sampling Proced

11、ure. The data were collected from Chinese beer drinkers in Guangzhou, Beijing, and Shanghai by convenient sampling at restaurants and bars/pubs. Altogether, 67 complete and useable questionnaires were obtained. 79% of the samples were male. For martial status, 56.7% were married. In terms of educati

12、on, 88.1% had college education. For occupation, 44.8% were professional, 43.3% were white collars.Data Collection. The data were collected by personal interview at restaurant and bars/pubs. The interviewers read the questions in the questionnaire to the respondent and recorded the answers obtained

13、in the questionnaire.Step 4: Data AnalysisData Coding and Data Entry. After the data collection was completed, the questionnaires were coded and entered into SPSS for Windows Version 12.0.1. Since each respondent evaluate five brands of beers, the data from each respondent are represented by five “c

14、ards” or rows of data input. Exhibit 1 shows the sample of the data input file. Statistical Analysis. To create perceptual maps for the foreign beer in China, the coordinates for each brand of beers are needed. For perceptual mapping by factor analysis, the coordinates for the brands are obtained by

15、 averaging the factor score of each brand across the respondents. The factor scores can be obtained by the following procedure:a. From the main menu, select “Analyze” then select “Data Reduction” then select “Factor Analysis” (see Exhibit 2). The factor analysis window will be open at this point. Th

16、e factor analysis window consists of a box showing all variables in the data set and the “Variables” box. There are also five bottoms for options including “Descriptive”, “Extraction”, “Rotation”, “Scores”, and “Options”. b. Select variables that represent product attributes (i.e., X1 to X10) from t

17、he box showing the list of variables into the “Variables” box (see Exhibit 3). This step identifies variables to be analyzed by factor analysis.c. Open “Rotation” box and select “Varimax” rotation (see Exhibit 4), and then click “Continue”. This step specifies the type rotation to be used.d. Open “S

18、cores” box and select “Save as variables” and for “Method” select “regression” (see Exhibits 5), and then click “Continue”. This step will create factor scores by regression method and add them to the data set as new variables.e. For “Extraction” use defaults for “Method” (i.e., Principal component)

19、, “Analyze” (i.e., Correlation matrix), “Display” (i.e., Unrotated factor solution) and “Extract” (i.e., Eigenvalues over 1), and then click “Continue”. This step will command the program to compute correlation matrix for the variables from the raw data. This correlation matrix will be, in turn, ana

20、lyzed to obtain a factor matrix. Principal component analysis is specified as the method of choice in extracting the factors. For the number of factors to be extracted, the latent root or eigenvalue criterion (greater than 1) will be used. The (unrotated ) component matrix will also be given as an o

21、utput.f. For “Descriptive,” use default for “Statistics” (i.e., Initial solution) and then click “Continue”.g. For “Options” use default for “Missing Values” (i.e., Excluded cases listwise). For “Coefficient Display Format”, select “Sort by size” (see Exhibit 6), and then click “Continue”. This step

22、 will create factor matrices of which variables are sorted by size in descending order.h. At this point, we are back at the factor analysis window. Click “OK” to run the program. The output will be shown in the output window.Results and InterpretationBased on the above factor analysis procedure, the

23、 following output are obtained: Communalities, Total Variance Explained, Component Matrix (or Factor Loadings Table), Rotated Component Matrix (or Rotated Factor Loading Table), and Component Transformation Matrix. The relevant outputs for perceptual mapping are the Total Variance Explained and the

24、Rotated Component (or Factor) Matrix. See the output distributed earlier for detail.Step1: Determination of the Number of Factors to RetainThe first step in the analysis of results is to determine the number of factors to be retained for further analysis. The Total Variance Explained in the output c

25、ontains the information regarding the ten possible factors and their relative explanatory power as expressed by their eigenvalues. In addition to assessing the importance of each factor, eigenvalues are also used to determine the number of factors to be retained. According to the latent root or eige

26、nvalue criterion, two factors will be retained because only component 1 (factor 1) and component 2 (factor 2) have latent roots or eigenvalues greater than 1 (see Exhibit 7).Step2: Naming the FactorsThe naming process involves substantive interpretation of the pattern of factor loadings for the vari

27、ables, including their signs, in an effort to name each of the factors. Before interpretation, a minimum acceptable level of significance for factor loading must be selected. All significant factor loadings typically are used in the interpretation process. But variables with higher loadings influenc

28、e to a greater extent the name or label selected to represent a factor.The Varimax rotated component analysis factor matrix which contains factor loadings is shown in the Rotated Factor Matrix (or factor loading table) in the output (Exhibit 8). For the sample size of 67, factor loadings have to be

29、0.65 or above to be considered significant. From the matrix, its clear that taste, color, package design, reputation, and country of origin load significantly on Factor 1, and price, availability, level of alcohol, level of advertising, and variety of sizes load significantly on Factor 2. None of th

30、e variables loads significantly on more than one factor.For Factor 1, all the five variables (taste, color, package design, reputation, and country of origin) represent different dimensions of the product. It is reasonable to name Factor 1 as product quality. For Factor 2, the five variables (price,

31、 availability, level of alcohol, level of advertising, and variety of sizes) represent the other three components of the marketing mix. It is reasonable to name the factor marketing activities.Note that the naming process is based on the subjective opinion of the researcher. For this reason, the pro

32、cess of naming factors is subject to considerable criticism. But if a logical name can be assigned that represents the underlying nature of the factors, it usually facilitates the presentation and understanding of the factor solution and therefore is a justifiable procedure.Step 3: Identification of

33、 Positions of the BrandsSince two factors are obtained in the factor analysis process, two factor scores are generated and added to the data set. Factor scores are composite measure of each factor computed for each subject for each brand. In this study, two factor scores can be used to replace the original set of ten variable