ASSESSMENT OF DATA ADEQUACY 



As is usual with field investigations, the sampling methods employed in this 

 study may or may not reflect biological reality. However, an effort was made to 

 employ standard, widely-used techniques in this study. These sampling methods 

 are probably the best presently available and are commonly used by contemporary 

 investigators. The methods employed are rep^^atable and the data are useful for 

 comparisons within this study, with future studies, and with most studies reported 

 in current literature. An effort was also made to minimize mathematical manipu- 

 lation of data to both minimize magnification or error, and to present data in 

 their most basic and comparable form. Several methodological attributes which 

 may affect data adequacy deserve mention here. 



Observer Bias 



A problem with nearly all biological sampling methods based on auditory or 

 visual observations is observer bias. Six observers recorded data at some time 

 during this study, and the visual acuity, experience, training, and background 

 of each probably adds substantial bias to the data. Especially important in this 

 respect are such subjective estimates as habitat use, topography, and on roadside 

 surveys, numbers of individuals. In the latter case, observer bias was minimized-- 

 or at least held constant— by using a single observer for all runs from February 

 through October. Observer bias also enters into interpretation of breeding bird 

 census data. Several authors (Best 1975, Dickson 1978, Eagles and Tobias 1978) 

 have provided evidence that this may substantially affect results. 



Observabi lity Bias 



Also important is the differential detectability or trappability of different 

 animals or of a single animal species in different habitats. Except for small 

 mammals, nocturnal animals were, seldom encountered in this study, and nocturnal 

 habitat use was not determined. Use of densely vegetated habitats was biased 

 against for such species as the white-tailed deer, as discussed earlier. Seasonal 

 changes in detectability were especially important for such species as the ring- 

 necked pheasant (figure 8). Observability bias inherent in the roadside wildlife 

 survey has been noted by Peterson (1975), Robbins and Van Velzen (1969) and Van 

 Velzen and Robbins (1971). Weather conditions, especially wind, can affect runs, 

 as was the case with March and April runs of the Missouri River route (figure 40), 

 although Weber and Theberge (1977) reported weather effects upon their results 

 to be minor. 



Non-randomness of Samples *■ 



As discussed earlier, sampling effort was not always uniform for all habitats. 

 This was especially true in the case of the general ground surveys, where consid- 

 erable bias toward roadside and open habitats is probable. 



194 



