REP47979 - Laserfiche WebLink

Home Browse Search

The samples taken in 1988 were double the size of those taken in 1986 and 1984. However, the number of 1988 samples, even when doubled, is no more than half the amount taken in 198ti. Furthermore, no sample adequacy calculations were carried out for 1988 woody plant data as was done in 1986. Thus it has not been shown that an adequate sample has been taken to characterize woody plant density. The authors themselves on pages 5, 7 and 9 note that the data may not have been a true representation of present density levels. Johnston (1980) states, "the smaller the sample, the greater the chance of sampling error, of the sample being unrepresentative." Woody p aT nt andTiq sagebrush density are used as independent variables in a regression that serves as the basis of the authors' recommendations regarding vroody plant establishment. Because the representativeness of the sample taken is in doubt, so are the regression and recommendations based on this data, as the sampling error may be quite large. C. Correlation Coefficient Values Tables 9, 10, 11, 12, 13, 14, and 15 of Apoendix A represent the results of regression analysis with dependent variables of reclamation age (including 1979 site data), reclamation age excluding 1979 site data, respread topsoil thickness, big sagebrush density, total shrub density, species richness, and species diversity respectively. In these tables, the authors present the value of the coefficient of correlation, usually labeled "r" in most statistics manuals. The highest "r" listed in each table and thus the largest r value for each individual regression, is listed below. Also listed is the corresponding value for r2, which is the souare of the r value and indicates the amount of variance in the independent variable that is explained by the dependent variable. Chart 2: Highest Coefficients of Correlation and Variances from Tables 9-15 Table ~ r-Value r2 Value 9 0.5741 D.33 10 0.5467 0.30 11 0.2855 0.08 12 0.3654 0.13 13 -0.3824 0.15 14 0.3467 0.1 2 15 0.3174 O.1D Note: In Tables 14 and 15, the r value for the regression of species richness or species diversity has been excluded as these values are expected to be highly correlated. Thus, for all the regressions, the highest percentage of variance explained (as measured by r2) is 330. This is a very low amount of variance even for vegetation data. The bulk of regressions on which many of the recommendations to the reclamation plan are based are well below even 33a. While significance testing was included, without an explanation, it is unknown what hypothesis (slope of the lines is greater than 0, etc.) was tested. Please clarify this point. However, conclusions can not be made with certainty when only 33a or less of the variance in the aependent variable can be explained by the independent variable. -3-