Cluster analysis for the respondent firms
In order to classify the firms according to their common characteristics, a cluster analysis was conducted. A cluster analysis is a multivariate data analysis technique used in grouping objects according to the characteristics they possess. If the classification is successful, the objects within clusters will have similar characteristics and there will be important differences between clusters. In this study, Hierarchical Agglomerative Clustering was used. In this method, each object starts out as its own cluster. In subsequent steps, the two closest clusters are combined to form a new cluster, thus reducing the number of clusters in each step. Wards Method is selected as the agglomeration procedure because this clustering procedure minimizes the within-cluster sum of squares at each stage. It attempts to combine clusters with a small number of observations and form clusters which tend to have the same number of objects. The distance between objects is measured via a squared Euclidean distance measure. (For more information about cluster analysis, one may consult Hair, 1995). In this study, the programs are clustered according to the similarity of importance that they attach to the seven factors revealed in the previous step. The analysis of the dendrogram and ANOVA used to test the significance of the differences between the group means finally resulted in four significant clusters. The categories common in the programs assigned to each cluster were specified and a title was given according to these common categories. The firms that are grouped in the same cluster, together with the mean and standard deviation of each cluster, are given in Figure