United States 

 Department of 

 Agriculture 



Forest Service 



Intermountain 

 Forest and Range 

 Experiment Station 



Research Paper 

 INT-262 



June 1981 



Visual Impacts of 

 Forest Management 

 Activities: 



Findings on Public Preferences 



Robert E. Benson 

 James R. Ullrich 



INTRODUCTION 



A major concern of forest land managers is the visual 

 impact that activities such as timber harvest, road building, 

 and mining have on the forest landscape. Management of 

 the visual resource is a regular part of planning on National 

 Forests, and many other public and private forest land 

 managers are increasing their efforts to protect and 

 enhance this resource. 



The Forest Residue and Harvesting Research Program at 

 the Intermountain Station includes studies of the impact of 

 harvest and roading activities on visual quality. Preliminary 

 results were reported earlier (Schweitzer, Ullrich, and 

 Benson 1976). This report updates earlier findings and 

 presents highlights of more recent studies. 



OBJECTIVES AND METHODS 



The purpose of the present series of studies was to mea- 

 sure public response to various types of timber harvesting 

 activities. Frequently, managers have several alternatives 

 available which may differ in their visual impact. If the likes 

 or dislikes of the public for these alternatives can be pre- 

 dicted, the manager has a basis for comparing the costs 

 against esthetic benefits in planning. 



There are some obvious problems in attempting to mea- 

 sure and compare visual quality. Many psychometric tech- 

 niques have been used in attempting to measure viewers' 

 responses (Arthur and Boster 1976). The technique used 

 in these studies is the Scenic Beauty Estimation (SBE) 

 Technique (Daniel and Boster 1976). This procedure con- 

 sists of showing a series of randomly selected slides of an 

 area to panels of viewers who make a numerical rating be- 

 tween (dislike) and 9 (like). An SBE score and a mean 

 rating (raw arithmetic mean) for each scene are developed 

 from these ratings. 



The SBE score is a sophisticated measure of viewers' 

 response based on mathematical transformations that take 

 into account the fact that some viewers use the rating scale 

 differently than others. For homogenous groups of ob- 

 servers, the mean ratings and the SBE scores are usually 

 closely related. To simplify the presentation, mean ratings 

 are used in this report because this presents results in the 

 same units of measurement as viewers rated the scenes. 



The SBE scores were used in analyzing the results while the 

 mean ratings and critical differences presented refer to raw 

 data. In all cases these data parallel the SBE values. 



The statistical data on SBE's mean ratings, analyses of 

 variance, and tests of significance are on file atthe Forestry 

 Sciences Laboratory, Missoula, Mont. 



Scales such as the to 9 scale used in this study are cal- 

 led categorical insofar as judgments are placed in a small 

 number of categories represented by integer numbers. 

 These scales are ordinal in that they indicate order. A 

 scene given a certain numerical rating is preferred to a 

 scene given a lower numerical rating. Ordinal scale ratings 

 do not, however, indicate how much difference in prefer- 

 ence is expressed by the two ratings. That is, while a rating 

 of 4 indicates a preference over 3, and a 3 over a 2, 4 is not 

 twice as preferred as 2. Furthermore, the difference be- 

 tween 4 and 3 need not be the same as between 3 and 2. 



If these were interval scales, the difference in preference, 

 say, between a 2 and 3 rating is the same as between a 3 and 

 4. The intervals are the same. It is then possible to compare 

 scenes using conventional parametric statistical tests 

 which are more powerful than nonparametric tests used 

 with ordinal data. 



There has been considerable debate as to whether or not 

 parametric tests can be applied to categorical, ordinal data. 

 (Anderson 1972). Specifically, parametric tests would be 

 used to determine if ratings given to one scene are statis- 

 tically significantly different from another scene (i.e., 

 scenes are drawn from different populations). We have 

 adopted this view in the studies reported here using mean 

 ratings to determine if there are statistically significant dif- 

 ferences between scenes. 



Furthermore, we have elected to treat mean ratings as if 

 they were constructed from an interval scale and by so do- 

 ing compare mean ratings to show quantitative differences 

 between scenes. 



There is no "proof" that the rating scale used is an inter- 

 val scale. The hazard in using an ordinal scale as if it 

 were an interval scale is that it may erroneously overstate 

 or understate the true intensity of likes or dislikes. The 

 reader should, therefore, be aware that the mean values 

 presented show quantitative differences in preferences 

 only to the extent the assumption of an interval scale is 

 accepted. 



1 



