Evaluation of some statistical methods of interpreting multi-element geochemical drainage data from New Brunswick

Log transformations in geochemistry

“…Miesch has perhaps misunderstood the purpose and content of my original Letter to the Editor (Chapman, 1976a). For example, in Chapman (1973), the positive skewness of stream sediment data collected from southwest New Brunswick was such that geologically meaningful results could not be obtained without log-transforming data and, in Chapman (1976b), regression analysis and other statistics did not give an efficient parametric statistical analysis with an untransformed data set and the data were log-transformed. I state (Chapman, 1976a, p. 212) that "the log-transformation may produce misleading results."…”

“…Frequently, parametric statistics will be far more efficient computed from log-transformed data compared to the untransformed data (as illustrated by the results in Chapman, 1973Chapman, , 1976b. Dr. Miesch notes that the paper of refers only to the estimation of means.…”

“…If the parametric statistic is used with the untransformed data, it may be sufficiently robust with respect to the form of the distribution to give geochemically meaningful results but this is certainly not always so. For example, the largest known deposit in the area, The Nigadoo Deposit, leaves an almost indeterminable dispersion pattern compared to background in the streams at the surface (Chapman, 1973). However, I have found some of these statistics to be uneconomical for use with large data sets such as those collected from most exploration geochemical surveys.…”

“…An alternative approach is to use the less powerful distribution-free statistics (i.e., nonparametric statistics). This is an interesting approach; however, the writer favors an approach of attempting to isolate anomalous populations and sort or identify samples with these populations (e.g., Lepeltier, 1969;Bolviken, 1971 ;Govett and Pantazis, 1971;Govett, 1972;Govett, 1974;Chapman, 1973;Whitehead and Govett, 1974). In summary, there is a problem in reconciling theoretical statistics with the empirical distribution of minor element concentrations and much work has been done by computer simulation to investigate this by the Exploration Geochemical Group at the University of New Brunswick (Govett and others, 1975).…”

Διεργασιες Συγχρονης Ιζηματογενεσης Στον Κορινθιακο Κολπο

“…(Merriam and Harbough 1964, Harboughand and Merriam 1968, Makris and Todt 1978, Sutterlin and Hastrings 1986. Η μέθοδος όμως έχει εφαρμοσθεί κυρίως σε πετρογραφικά, ιζηματολονικά και γεωχημικά δεδομένα (Read at al 1971, Κίΐίβεη and Heier 1975, Trapp and Rockaway 1977, Chapman 1978, Doyle and Feldhansen 1981, Watner 1985. (Howarth, 1967).…”