Analysis of Dispersion Around Nonparametric Means 



As was mentioned previously, a nonparametric mean suggested by 

 Duffield and Patterson (1989) was calculated for each of the 

 estimated equations. The use of this nonparametric technique 

 allowed the calculation of a sample variance and thus the 

 construction of confidence intervals around the calculated means, 

 A comparison of the estimated logit means and the nonparametric 

 means shows them to be quite close in most cases. 



Table 15 shows the nonparametric means and 95% confidence 

 intervals for each of the four CVM questions for the entire 

 sample and the four hunter subgroups. An analysis of these 

 figures shows significant variation between hunter subgroup as 

 well as between certain improved conditions values. 



Table 16 shows the same calculated statistics for each of the 

 seven Montana DFWP administrative regions. As was mentioned 

 before, an analysis of the 95% confidence intervals surrounding 

 the nonparametric means for the seven regions shows no 

 significant difference in net economic value for deer hunting 

 experiences. 



29 



