chosen a priori as the most efficient variables in identifying 

 distinct hunter types. Reasons which a majority of hunters rated 

 similarly were not used in the clustering since these variables 

 provide little help in drawing distinctions between hunter 

 groups . 



The SPSSx Quick Cluster program does not select a specific number 

 of clusters statistically. The programmer must pick the desired 

 number. Allen (1988) discussed three criteria for determining 

 the optimal number of clusters. 



(1) The number of observations in each cluster must be large 

 enough to allow economic analysis (about 100 



observations) . 



(2) The clusters should be different enough to define 

 distinct hunter subgroups, yet they must conceptually 



make sense. 



(3) The smallest number of clusters which does not mask 

 important differences between types of hunters is 



preferred. 

 Both in his study of Montana anglers (1987) and that of Montana 

 elk hunters (1988) Allen chose to use four clusters for grouping 

 recreationists. For this study, clusters of 2 , 3, 4, and 5 were 

 run and since sample size criteria were met in all cases an 

 optimal number of clusters was first selected based on maximum 

 average distance between cluster centers. This basis also 

 yielded a cluster size of four. These clusters were then 

 analyzed to determine whether the groupings made conceptual 

 sense. It was found that two of the four hunter groupings were 

 very distinct and easily labeled. These were Meat Hunters and 

 Trophy Hunters. Two questions which asked respondents to rank 

 the importance of taking a trophy deer were included in the 

 clustering process. Respondents consistently ranked these two 

 questions similarly. This suggests that respondents took the 

 clustering questions seriously. Meat Hunters and Trophy Hunters 

 responded in opposite ways to the trophy and meat questions 

 making their basic motivations easy to identify. The remaining 

 two clusters could perhaps best be termed as two Generalist types 

 of hunters. One cluster which we termed the Generalist-Enthusiast 

 ranked the importance of all reasons for hunting highly. This 

 group seemed motivated by nearly all aspects of the hunt (meat, 

 trophies, testing skills, easy access) . The second Generalist 

 class which we termed the Generalist-Meat hunter seemed most 

 motivated by good access to the hunt and the fact that they had a 

 special permit to hunt an area. Besides these factors they 

 ranked meat as a major motivation and trophies as relatively 

 unimportant. 



One source of validation for the clustering process lies in 

 examining how the different cluster groups compare in regard to 

 characteristics not used in the clustering process. If the 

 clusters were indeed distinct subgroups we would expect that 



42 



