Laserfiche WebLink
not constitute independent observations. This is the reason that new Rule 4.15.11(1)(b)(i) requires <br />that quadrats associated with other quadrats along a transect must be treated as subsamples. <br />Stepping out of the realm of statistical theory, it seems fairly evident from the data that 2005 <br />rangeland reclaimed area production exceeded reference area production by a wide mazgin. The 90% <br />upper confidence limit on the reference area mean is 44.4 grams/square meter, while the 90% lower <br />confidence limit on the reclaimed area mean is 130.9 grams per squaze meter (unadjusted data). The <br />confidence intervals do not overlap, and are sepazated by a wide mazgin. A two sample reverse null <br />t-test would support the same conclusion, that the reclaimed azea production exceeds the reference <br />azea production. <br />My suggestion is that we apply some professional judgement to the 2005 data, rather than reject it out <br />of hand due to the treatment of individual quadrats as sample observations. I think we should point <br />out the concerns we have identified with the 2005 sampling, and request that the 2006 procedures be <br />properly amended, in conformance with the revised regulation. In the final bond release report, I <br />would suggest that two separate t-test demonstrations be presented for the 2005 rangeland production <br />data. The first test should be a two sample reverse null t-test, based on the sample size of 90 in the <br />reclaimed area and reference azea (each uq adrat treated as an independent observation). The second <br />test should be a two sample reverse null t-test, based on a sample size of 15 in the reclaimed azea and <br />reference area (each transect treated as an independent observation). Both tests should be based on <br />"corrected" quadrat data (undesirable component removed from the data from each quadrat prior to <br />calculation of sample statistics). If both tests conclude that reclaimed area production exceeds <br />reference azea production, I believe we could accept the data as a valid demonstration of success <br />(recognizing that the regulations specifying specific sample size requirements and treatment of <br />transect quadrats as subsamples were not in effect at the time of sampling). <br />Comments welcome. <br />