Page 5 - Statistics and Data Analysis in Geology

In addition to the downloadable files from the 3rd edition of Statistics and Data Anal-
ysis in Geology, you may also find additional data sets and exercises at this site as
they are made available from time to time.
The basic arrangement of topics covered in the book is retained from earlier
editions, progressing from background information to the analysis of geological
sequences, then maps, and finally to multivariate observations. The discussion of
elementary probability theory in Chapter 2 has been revised in recognition of the
unfortunate fact that fundamentals of probability often are passed over inintroduc-
tory courses in favor of a cookbook recitation of elementary statistical tests. These
tests are also included here, but because probability forms the basis for almost all
data analysis procedures and a thorough grounding in the concepts of probability is
essential to understanding statistics, this introductory section has been expanded.
The discussion of nonparametric methods introduced in the 2nd edition has been
expanded because geologic data, particularly data collected in the field, seldom sat-
isfy the distribution assumptions of classical parametric statistics. The effects of
closure, which results in unwarranted relationships between variables when they
are forced to sum to a constant value, are examined in detail. Geological measure-
ments such as geochemical, petrographic, and petrophysical analyses, grain-size
distributions-in fact, any set of values expressed as percentages-constitute com-
positional data and are subject to closure effects. The statistical transformations
proposed by John Aitchison to overcome these problems are discussed at length.
In the 2nd edition, I revised the discussion of eigenvalues and eigenvectors
because these topics had proved to be difficult for students. They are still dif-
ficult, so their treatment in the chapter on matrix algebra has been rewritten and
a new section on singular value decomposition and the relationship between R- and
Q-mode factor methods has been added to the final chapter on multivariate
analysis.
The central role of geostatistics and regionalized variable theory in the study
of the spatial behavior of geological and other properties is now firmly established.
With the help of Ricardo Olea, I have completely revised the discussion of the many
varieties of kriging and provide a series of simple demonstrations to illustrate how
geostatistical methodologies work. I also have revised the section on contour map-
ping to reflect modern practices.
A discussion of fractals has been added, not because fractals have demon-
strated any particular utility in geological investigations, but because they seem to
hold a promise for the future. On a more prosaic topic, the section on regression
has been expanded to include several variants that have special significance in the
Earth sciences. To make room for these and other discussions, some subjects that
proved to be of limited utility in geologic research have been deleted. Moving most
tables to the WWW sites has made additional room in the text.
Because this is not a reference book, references are not emphasized. Citations
are made to more specialized or advanced texts that I have found to contain espe-
cially lucid discussions of the points in question rather than to the most definitive
or original sources. Those who wish to pursue a topic in depth will find ample
references to the literature in the books I have included; those that simply want
an elaboration on some point will probably find the books in Suggested Readings
adequate for their needs.
I am fortunate to have enjoyed the help and encouragement of many people
in the creation and evolution of this book throughout its several editions. The

vi