Cluster Analysis

While the frequencies provide an indication of the acceptability or agreement of specific management issues, it does not give an overall indication to different segments of the respondents. Cluster analysis groups the respondents into groups based on similar response patterns. This provides a segmentation of respondents who hold similar views toward management.

Reliability analysis / scaling

Cluster analysis groups respondents by response patterns to questions. It is advantageous to have a diversity of questions to group respondents. However, too many questions and it may be difficult to make meaningful distinctions between groups. This survey consisted of 40 questions, however some questions measured similar concepts. To simplify the analysis, the questions that measure similar concepts can be averaged together to form one score, or scale, for the concept the questions are measuring.

The first step to constructing a scale is to ensure the responses to the questions are in fact related. One way to determine this is through reliability analysis. One such reliability analysis is Cronbach’s Alpha, which essentially computes the average of all possible correlations among items to be included in the scale. Typically an Alpha value of .6 or higher is taken as an indication the items are measuring the same concept. To be more stringent, an Alpha value of .7 was used as the cut off for this analysis. In other words, if the items had a Cronbach’s Alpha of .7 or higher, they were combined into a scale. The following items were combined into scales.

A cluster analysis was conducted on the scale items and the following items (Table 15).

Cluster 1 could be labeled as skiers / recreationists. This group, which comprised 44% of the respondents, considered it to be most acceptable to restrict research activity, least acceptable to not allow the trail widening, found it less acceptable to allow the research project along the trail, strongly agreed that walkers and dogs should not be allowed on groomed ski trails, found it unacceptable to close areas to outreach and recreation, and had the highest level of acceptability for removing trees or branches for snow and safety.

Cluster 2 might be labeled walking / dog / researchers. This group (12% of respondents) had lower acceptability toward restricting research activity than cluster 1, and found not allowing the trail widening acceptable. This group found not allowing walkers or dogs on the trails to be more unacceptable than the other groups and felt it was acceptable to expand the network of winter walking trails. This group also had the least acceptability toward removing trees and installing woodchips on trails. This group was tolerant of wheeled motorized access in winter.

Cluster 3 seems to represent researchers / skiers. This group (44% of respondents) rated restricting research activity with the highest level of unacceptability and found it acceptable to allow the research project along the trail and restrict the trail widening. Yet this group found it to be acceptable to not allow walkers and dogs on groomed ski trails and found it acceptable to remove branches for snow and safety reasons and to prohibit wheeled motorized access in winter.

Interestingly no group found it highly acceptable to close the area to outreach and recreation.

University of Alaska Fairbanks
Tanana Valley Campus
PO Box 8040-UAF

Phone: +1 907 474 7920
Fax:    +1 907 474 6725
fndlm5@uaf.edu