Extracting factors to investigate Consumer attitudes Using SPSS
                               Dr. Kalyan Sengupta*, Prof Atish Chattopadhyay*



Introduction


A consumer survey was carried out to measure relative importanceof attributes the
consumers attach to various attitudinal variables, while purchasing branded bakery
products from bakery chains of the city of Kolkata. A sample of 287 consumers were
contacted at purchase points and asked about nine different attributes influencing the
choice of specific brand. The factors listed on the questionnaire were: convenience,
décor and ambience of shop, service, food quality and taste, availability of a wide variety,
brand reputation, value for money, freshness and portion size.


It was interesting to assess how much weight the consumers put onto each of these nine
factors in a five point Likert Scale measuring 1 for absolutely no value and 5 for
extremely high value. The respondents constituted a good representation of buyers with
different age, income, and gender groups.


Initially a number of Chi-Square tests were performed to examine dependency of each
attitude factors with respect to different demographic variables, namely age, income and
gender using SPSS program. It was found out from the test attitudes data that the factors
were independent of the said demographic variables. This may be interpreted as the
measured attributes influence consumers of different groups more or less in the same
manner.


Factor Analysis for further understanding of attitudes


Factor Analysis is a Statistical Model which uncovers the latent structure of a set of
variables, by reducing the number of variables and clubbing them into a smaller number
* Faculty Members, ICFAI Business School, Kolkata
of factors. SPSS has several methods or models for extracting factors from a set of
variables. Principal Component Analysis (PCA) method was chosen for the purpose.
This method seeks a liner combination of variables in a way to explain maximum
variance. Once one factor is extracted, the model searches for another linear combination
to explain maximum amount of remaining variance and the iterative process continues
until some tolerance limit is come across.


Use of SPSS


In our case of attitude data, SPSS extracted four factor components, explaining a total of
60.5 percent variance in the variables, after a factor rotation (varimax) treatment. The
rotated component matrix is illustrated in the figure 1. From the correlation matrix, it
was possible to interpret those four dimensions in the following way:


•   Factor Component 1: Reflects onto some trendy attitude of the consumers, including
    the attitude variables like shop décor, brand name, etc.
•   Factor Component 2: Reflects onto some economic value related attitude related to
    the variables indicating good value for money, portion size, freshness of the product.
•   Factors Component 3: Reflects onto service and quality related attitudes.
•   Factor Component 4 reflects onto convenience (low access time) related attitude
    variable.
Figure.1: Rotated Component Matrix for Factor Analysis

                                                        Component
                                      1                2           3                4
Factors-convenient location         0.295            0.103       0.136            0.748

Factors-shop    décor        &      0.801         -6.286E-02     8.847E-02      7.352E-02
ambience
Factors-service                     0.170            0.147         0.748        8.180E-02

Factors-food quality/taste       -6.252E-02       -3.095E-02       0.836        -4.723E-02

Factors-availability   of    a      0.375            0.122         0.101          -0.697
wide variety
Factors-brand                                                0.712                          0.166   -5.643E-03   -5.091E-02
name/reputation
Factors-value for money                                   4.099E-02                         0.691     -0.137       0.223

Factors-freshness                                         -9.536E-02                        0.574     0.144        -0.199

Factors-portion size                                         0.238                          0.758     0.129      -1.696E-02

Extraction Method: Principal Component Analysis.
Rotation Method: Varimax with Kaiser Normalization.


The components may be viewed from the factor component plots to locate their positions
in the multidimensional space. Such an example plot is illustrated in the figure 2.




                                   Component Plot in Rotated Space




                      1.0

                                                   faporsiz favalmon
                                                        fafresh
                     0.5
       Component 2




                                                        faavail fabrname
                                                      faserv faconloc
                     0.0
                                                  fafoodqu            fadecamb

                     -0.5



                     -1.0
                                                                                        -1.0
                            -1.0                                                 -0.5
                                     -0.5                                  0.0
                                            0.0                      0.5
                                                    0.5    1.0 1.0           nt         3
                                     Compone
                                             nt 1                       pone
                                                                     Com




                                        Figure 2
The component score coefficient matrix, which could be used to compute factor scores of
the individual respondent from the original variables, is displayed in figure 3. These
individual scores may be automatically computed by SPSS and saved in the spread sheet.


Figure 3: Component Score Coefficient Matrix

                                                      Component
                                      1               2           3                  4
Factors-convenient location         0.176           0.014       0.070              0.646

Factors-shop    décor        &      0.581           -0.172         -0.009          0.055
ambience
Factors-service                     0.023           0.024           0.546          0.071

Factors-food quality/taste          -0.127          -0.077          0.651          -0.036

Factors-availability of      a      0.252           0.032           0.026          -0.610
wide variety
Factors-brand                       0.494           0.016          -0.091          -0.054
name/reputation
Factors-value for money             -0.061          0.505          -0.162          0.186

Factors-freshness                   -0.167          0.423           0.076          -0.177

Factors-portion size                0.048           0.507           0.016          -0.024

Extraction Method: Principal Component Analysis . Rotation Method: Varimax with
Kaiser Normalization. Component Scores.


The individual factor scores were used for further causal analyses of the data. Thus
factoranalysis is mainly a data reduction technique with which the analyst can reduce the
variable space, which can ease up the computations forcomplex analytical models.


           Key to the variable names in the Factor Analysis
Attribute Name                                   Variable
Factors-convenient location                      faconloc
Factors-shop décor & ambience                    fadecamb
Factors-service                                  faserv
Factors-food quality/taste                       fafoodqu
Factors-availability of a wide variety           favail
Factors-brand name/reputation   fabrname
Factors-value for money         favalmon
Factors-freshness               fafresh
Factors-portion size            faporsiz

Extracting factors to investigate Consumer attitudes Using SPSS

  • 1.
    Extracting factors toinvestigate Consumer attitudes Using SPSS Dr. Kalyan Sengupta*, Prof Atish Chattopadhyay* Introduction A consumer survey was carried out to measure relative importanceof attributes the consumers attach to various attitudinal variables, while purchasing branded bakery products from bakery chains of the city of Kolkata. A sample of 287 consumers were contacted at purchase points and asked about nine different attributes influencing the choice of specific brand. The factors listed on the questionnaire were: convenience, décor and ambience of shop, service, food quality and taste, availability of a wide variety, brand reputation, value for money, freshness and portion size. It was interesting to assess how much weight the consumers put onto each of these nine factors in a five point Likert Scale measuring 1 for absolutely no value and 5 for extremely high value. The respondents constituted a good representation of buyers with different age, income, and gender groups. Initially a number of Chi-Square tests were performed to examine dependency of each attitude factors with respect to different demographic variables, namely age, income and gender using SPSS program. It was found out from the test attitudes data that the factors were independent of the said demographic variables. This may be interpreted as the measured attributes influence consumers of different groups more or less in the same manner. Factor Analysis for further understanding of attitudes Factor Analysis is a Statistical Model which uncovers the latent structure of a set of variables, by reducing the number of variables and clubbing them into a smaller number * Faculty Members, ICFAI Business School, Kolkata
  • 2.
    of factors. SPSShas several methods or models for extracting factors from a set of variables. Principal Component Analysis (PCA) method was chosen for the purpose. This method seeks a liner combination of variables in a way to explain maximum variance. Once one factor is extracted, the model searches for another linear combination to explain maximum amount of remaining variance and the iterative process continues until some tolerance limit is come across. Use of SPSS In our case of attitude data, SPSS extracted four factor components, explaining a total of 60.5 percent variance in the variables, after a factor rotation (varimax) treatment. The rotated component matrix is illustrated in the figure 1. From the correlation matrix, it was possible to interpret those four dimensions in the following way: • Factor Component 1: Reflects onto some trendy attitude of the consumers, including the attitude variables like shop décor, brand name, etc. • Factor Component 2: Reflects onto some economic value related attitude related to the variables indicating good value for money, portion size, freshness of the product. • Factors Component 3: Reflects onto service and quality related attitudes. • Factor Component 4 reflects onto convenience (low access time) related attitude variable. Figure.1: Rotated Component Matrix for Factor Analysis Component 1 2 3 4 Factors-convenient location 0.295 0.103 0.136 0.748 Factors-shop décor & 0.801 -6.286E-02 8.847E-02 7.352E-02 ambience Factors-service 0.170 0.147 0.748 8.180E-02 Factors-food quality/taste -6.252E-02 -3.095E-02 0.836 -4.723E-02 Factors-availability of a 0.375 0.122 0.101 -0.697 wide variety
  • 3.
    Factors-brand 0.712 0.166 -5.643E-03 -5.091E-02 name/reputation Factors-value for money 4.099E-02 0.691 -0.137 0.223 Factors-freshness -9.536E-02 0.574 0.144 -0.199 Factors-portion size 0.238 0.758 0.129 -1.696E-02 Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization. The components may be viewed from the factor component plots to locate their positions in the multidimensional space. Such an example plot is illustrated in the figure 2. Component Plot in Rotated Space 1.0 faporsiz favalmon fafresh 0.5 Component 2 faavail fabrname faserv faconloc 0.0 fafoodqu fadecamb -0.5 -1.0 -1.0 -1.0 -0.5 -0.5 0.0 0.0 0.5 0.5 1.0 1.0 nt 3 Compone nt 1 pone Com Figure 2
  • 4.
    The component scorecoefficient matrix, which could be used to compute factor scores of the individual respondent from the original variables, is displayed in figure 3. These individual scores may be automatically computed by SPSS and saved in the spread sheet. Figure 3: Component Score Coefficient Matrix Component 1 2 3 4 Factors-convenient location 0.176 0.014 0.070 0.646 Factors-shop décor & 0.581 -0.172 -0.009 0.055 ambience Factors-service 0.023 0.024 0.546 0.071 Factors-food quality/taste -0.127 -0.077 0.651 -0.036 Factors-availability of a 0.252 0.032 0.026 -0.610 wide variety Factors-brand 0.494 0.016 -0.091 -0.054 name/reputation Factors-value for money -0.061 0.505 -0.162 0.186 Factors-freshness -0.167 0.423 0.076 -0.177 Factors-portion size 0.048 0.507 0.016 -0.024 Extraction Method: Principal Component Analysis . Rotation Method: Varimax with Kaiser Normalization. Component Scores. The individual factor scores were used for further causal analyses of the data. Thus factoranalysis is mainly a data reduction technique with which the analyst can reduce the variable space, which can ease up the computations forcomplex analytical models. Key to the variable names in the Factor Analysis Attribute Name Variable Factors-convenient location faconloc Factors-shop décor & ambience fadecamb Factors-service faserv Factors-food quality/taste fafoodqu Factors-availability of a wide variety favail
  • 5.
    Factors-brand name/reputation fabrname Factors-value for money favalmon Factors-freshness fafresh Factors-portion size faporsiz