Laserfiche WebLink
<br />The second section, which begins on page 12 with the section titled <br />"Reclamation Aoe", serves as the primary basis for most of the summary <br />recommendations. This section involves regression analysis of "108 <br />parameters" and certain independent variables such as reclamation age of an <br />area, soil depth, woody plant diversity, and species diversity. These <br />regression analyses are flawed for several reasons as follows: <br />A <br />B <br />Inclusion of Narrative Describing Analysis Methods <br />Generally, when a study includes the use of regression analysis, <br />narrative must be included to validate that it was oroperly used and to <br />clarify to the reader how and why it was used. Appendix A includes only <br />minimal references to methodology and intent. A description of the <br />analyses should be included when the operator wants to use data gathered <br />from it as proof for recommendations. Important issues that should have <br />been addressed include: 1) why regression analysis was chosen versus <br />correlation; 2) whether the assumptions associated with correlation or <br />regression were met for the data set that was considered (see Johnston <br />(1980), p. 37-45 or Sokal and Rolff (1981) p. 459-461, or Zar (1974) p. <br />205.); 3) whether the sample set was la roe enough to have reduced <br />sampling error to an acceptable level; 4) what methods, references, and <br />computer programs were used in carrying out simple and stepwise <br />regression and the significance testing associated with these; and 5) <br />exactly what the "108 parameters" used in the regression analyses were. <br />Though not required, it would have been valuable for the report to have <br />included the raw transect data itself. <br />Representative Data <br />The authors appear to have used only 1988 vegetation data (describing <br />1974, 1981, 1483, and 1986 reclamation) in the rea ression analyses. It <br />had been Kerr Coal's practice in 1986 sampling to calculate sample <br />adequacy at the 90/10 level independently for cover, production, and <br />woody plant density. Thus, in 1986, a different number of samples were <br />collected for each category, with woody plant density usually requiring <br />the greatest number of samples to meet adequacy. The following chart <br />shows how many samples were taken on each reclaimed area per year of <br />sampling. <br />Chart 1: Comparison of The Number of iloody Plant Density Transects <br />Evaluated During 1984, 1.986 and 1988 Vegetation Monitoring <br />Reclaimed Area: 1979 1981 1983 1986 <br />Sample Year <br />1964 20 20 - - <br />1986 34 56 22 - <br />1988 5* 6* 6* le* <br />* Double Sample - Belt transects were run on both sides of the 50 meter <br />transect line rather than on lust one side. However, the two belt <br />transects were averaged to produce one data point per cover transect and <br />cannot be considered to be two independent samples. <br />-2- <br />