A Cluster Analysis of Activity, Frequency, and Environment Variables to Identify Water-based Recreation Types

Abstract

Utilizing a sample of 250 individuals from a data base that describes water-based recreation patterns of 2174 heads of household in northeastern Wisconsin, this paper presents the findings of two cluster analyses. The first is based on participation frequencies for eight activities; the second is based on 23 variables combining type of environment with activity frequencies. For both analyses, a replication with a second subsample of 250 individuals was carried out indicating stability of the clusters derived. A suite of computer programs developed by Wishart (1970) was used for analyzing the data. The first analysis identified eight mutually exclusive clusters of individuals; clusters distinguished from each other by the kind and frequency of their water-based recreation activity. Each cluster was named and characteristics that distinguished the cluster from the total sample were described. The second analysis based on kind, frequency, and type of environment yielded nine clusters. These clusters were named and cluster characteristics described. Inclusion of the location variable added an important dimension to cluster analysis and enabled more useful descriptions of participant groups than has previously been done. After the remaining 1924 individuals were assigned to one of the nine clusters based on their standing on multiple participation variables, each cluster was cross tabulated with 13 predictor variables. With cross tabulations, the exact makeup of the clusters could be established and used for predictive purposes. Some theoretical, methodological, and practical implications of cluster analysis and the clusters derived are discussed.