ESSAY FOR SEMINAR ON QUANTITATIVE METHODS
Michaela Bátorová, PhD student Department of Regional Studies / Major: Local Governance
I decided to have a look at the method of hierarchical cluster analysis. The reason why I have chosen this particular method is very simple. Even if this method is not considered as the most common one in the social sciences (Cramer, 2003), the field of comparative politics and intercultural management (my research fields) use this method very often.
1. WHAT IS CLUSTER ANALYSIS?
As Romesburg noted: “Cluster analysis is a name for a variety of mathematical methods, numbering in the hundreds that can be used to find out which objects in a set are similar”(2004). It is a set of mathematical methods for creating clusters – types, classes or groups out of the set of studied cases (objects, or units) and their variables (attributes, or characteristics). These clusters are the result of comparing and grouping the cases’ similarities (dissimilarities) based on their variables. The variables come from the observations about the cases. At the same time, cluster analysis allows us to compare and group the similarities of the variables (Lorr, 1983; Johnson & Wichern, 1992; Romesburg, 2004,). However, that cluster analysis is more often used for grouping cases. Cluster analysis is also called as segmentation analysis or taxonomy analysis and it is used when the researcher does not know the number of groups in advance but wishes to establish groups and then analyze group membership (Garson, 2009). In the methodological books we can find several methods for cluster analysis. Two main groups are hierarchical and non-hierarchical clustering methods, within which we can find several techniques for treating our data. For example single, complete of average linkage methods, k-means method, multidimensional scaling, etc. In this paper I will focus on hierarchical cluster analysis with average linkage method – one of...