Laserfiche WebLink
II <br />Subtleties Of Statistical Comparisons <br />Because the current regulatory procedures for evaluating revegetation suc— <br />cess rely heavily on statistics, it is also important to understand the real <br />effects of statistical analyses. For example, although a single—tailed t test <br />will minimize the sample size required and. reduce sampling costs, it will <br />result in increasing the stringency of the test of success. In other words; it <br />will result in disproving success sooner than will a two—tailed analysis. <br />Another point to be realized is that although the 80: confidence level speci— <br />fied for testing for shrublands intuitively seems less stringent than a 90% <br />test, it is in fact more stringent. For example, one may be 90% certain the <br />true mean occurs between 8 and 12, but only 80% certain it falls between 9 and <br />11. In this case, an estimate of the mean equal to 8.5 could not be rejected <br />as falling outside the expected bounds of estimating error with 90% confidence, <br />.but could be rejected with 80% confidence. <br />Certainly some objective and quantifiable standard for testing success is <br />necessary to protect operators and regulatory authorities alike. Current regu— <br />latory approaches, however, require detailed and intensive ecological studies <br />of reclaimed sites without recognition of the problems associated with collect - <br />—10- <br />4 -' <br />Simplifying the Solution <br />•Probably <br />our most difficult <br />problem to overcame in developing <br />an efficient <br />and effective system for proving <br />success of revegetation on mined <br />lands is the <br />professional biases we hold from <br />traditional ways of doing things. <br />Traditional <br />ways of evaluating vegetation were never intended to demonstrate <br />success or <br />failure of reclaimed lands. They <br />were developed instead to be management <br />tools <br />to indicate animal carrying capacities <br />and the influence of animal <br />use on range <br />condition and trend largely by a <br />trial and error process. <br />Certainly some objective and quantifiable standard for testing success is <br />necessary to protect operators and regulatory authorities alike. Current regu— <br />latory approaches, however, require detailed and intensive ecological studies <br />of reclaimed sites without recognition of the problems associated with collect - <br />—10- <br />4 -' <br />