The paper "Cluster Analysis - SPSS" is a perfect example of a marketing case study. The first thing that should be checked is the appropriateness of conducting a factor analysis. This can be done by making an observation if there is the existence of bivariate correlations. It is appropriate to use factor analysis if there are a lot of medium or large correlations. The test for correlations between the factors being tested can be done by utilizing the SPSS standard command for bivariate correlations or through requesting the output as part of the factor analysis in SPSS. There are formal statistics which can be put into use in order to test if using factor analysis is appropriate that are offered under SPSS.
Bartlett’ s test of sphericity test which involves using a chi-square test to determine whether the variables are uncorrelated is one of the methods. In this method when there is a large value of the statistics, it acts as an indication that the null hypothesis should be rejected. Kaiser Meyer-Olkin measure of sampling is another method that can be used to measure the appropriateness of using factor analysis.
In this method, the correlations between the variables are tested whether they are small, where values are close to one serves as an indication that the use of factor analysis is appropriate. A number of ways can be used when conducting factor analysis, where the choice depends on the intended purpose. There are two common forms of factor analysis: principal components analysis (PCA) and common factor analysis. When one wants to establish the minimum number of factors or/and would like to utilize the factors in performing subsequent analysis, then PCA is the right choice.
On the other hand, common factor analysis is applicable where one’ s concern is identifying the underlying factors. There are four main methods which can be engaged in determining the number of factors. The four are the theoretical determination of factors, use of eigenvalues larger than one, by use of scree plot and by looking at percentage variance explained. Factor matrix is one of the outputs from factor analysis. The factor matrix displays coefficient that gives the relationship that exists between a factor and variables.
Large coefficients are an implication of close relationship while small coefficient implies a weak relationship or no relationship at all. Due to the difficulty that can be encountered in interpreting factors, as factors are related to many variables, the rotation method is used in transforming the factor matrix in order to make it easier to interpret. The most common form of rotation is the varimax procedure. In order to test for the fitness of a factor model through estimation 0f “ residual” . These are done by comparison of the original correlation matrix with correlation matrix estimated using the result of factor analysis.
The difference that exists between observed and estimated correlations is what is referred to as residual. In case there is very many large and significant residual it will be an indication that the model is of poor quality. Cluster analysis When performing cluster analysis using SPSS, the first step is to choose the variables that one would wish to perform the cluster analysis on. One of the possibilities is just to use variables that were collected in the survey, though this presents a difficulty if there are large numbers of variables.
Another alternative is to use the factor scores obtained through factor analysis or making use of variables that had the highest loadings on each of the factors.