GENERAL32515 - Laserfiche WebLink

Home Browse Search

-3- The above statements contain my major objection to the proposals. Trapper has endeavored to derive predictive regression equations from functionally independent variables. Specifically, Trapper incorrectly asserts (by employing regression) that reclaimed area herbaceous cover and production are dependent upon herbaceous cover and production of respective reference areas. This is not logically so, for one can not say that manipulation of the reference areas will subsequently affect herbaceous cover and production on the reclaimed areas. A correlation between reference areas and reclaimed areas exists, and that is the basis for the use of reference areas. But correlations can not be used as the predictive tool Trapper desires to set a numerical standard in advance. Should Trapper derive a predictive multiple regression equation based on one or more independent variables all affecting a dependent variable (e. g., cover being influenced by topsoil depth, season length, annual precipitation, and others) which was then evaluated by regressing against known data contained in reference and reclaimed areas, it could be employed to predict revegetation success standards. A further concern with the proposal is contained in Trapper's proposed standard value (Y-25).9). If one is to employ the multiple regression for a specific set of independent factors, a derived value of the dependent variable (y) is produced. Trapper has not explained why this predicted value needs to be further modified by subtraction of the value of two standard deviation units and then by a 90 percent devaluation. I have not been able to identify the source of this proposal in standard references or in Trapper's proposal. ExtraDOlation A concern which has been expressed by reviewers of the proposal (Redente, 1982) and in texts ( Zar, 1974, Steel and Torrie, 1980) is the extrapolation of the regression equation outside of limits imposed by actual values of independent variables. The regression equation explains variation and the relationship of the variables within the constraints of the data originally employed and cannot be safely extrapolated beyond those values. Practically speaking, for the predictive regression to be successful under all circumstances Trapper would have to include values for independent variables for the worst years (drought) and the most favorable years in order to be able to use the model. Source of Data for Regressions In Tables 4.4-10, 4.4-11, and 4.4-12 Trapper presents data on annual growing season, monthly average temperatures, and monthly precipitation; all abiotic independent variables which were considered in the preparation of the stepwise multiple regressions as affecting cover and production on reclaimed surfaces. No source location for this information was presented. It is presumed since Trapper serves as an official meteorological reporting station for the Craig area these measurements came from mine site. Trapper should clarify the source of this data, since data from other than the specified area of interest might adversely affect results for Trapper's location.