Companion to Correlation Delayed
Spring 2002

The Ukrainian Archive (UKAR, at www.ukar.org) was given seed by the provocation of Morley Safer's Ugly Face of Freedom broadcast of 23-Oct-1994, and for four years has posted mainly responses to a string of similar calumnies being heaped upon Ukraine and upon the Ukrainian people.  In recent months, however, interests have begun to broaden, and postings have drifted beyond the original focus most notably, recent postings have concerned themselves with Toxocara canis infection, keyboard dynamics and encryption, and manganese poisoning.  Long pauses between postings, atypical over the life of UKAR until recently, were being dedicated to writing a series of books on scientific method, the first one treating the subject of correlation, under the title Companion to Correlation.

Whether the topic of scientific method should occasion the building of a fresh web site, or for a while be accommodated within UKAR, is an obvious question that has arisen, and as everything can be seen to be interrelated, then at least a temporary accommodation within UKAR has been selected.  The interrelation is as follows.

One of the great battle fronts in our day, perhaps the greatest, is between two ways of thinking, of perceiving, of acting the scientific and the religious.  I am squarely on the side of the scientific, simply because it puts bread on the table where otherwise there would be crumbs, because it lights the darkness, dispels cold, prolongs life, and enriches understanding.  Scientists all over the globe, furthermore, are predisposed by their training toward cooperation, though one of the catastrophes of our day is that politicians and religious leaders are trained to no similar cooperation, and these squabbling politicians and religious leaders select for deployment from the many diverse products of science the ones that kill.  The battle between science and religion is an ancient one, but may be seen to be coming to a head in recent days, perhaps with catastrophic consequences, and perhaps ending in Armageddon.  However, were even a fraction of the energy that is today devoted to religious study devoted instead to science, then all manner of good would follow, starting with the reduction of bickering and leading to the overthrow of demagogues and eventuating in Utopia.

In any case, when I used to lecture on scientific method, the topics that I covered, each of which might soon lead to its own book, were six:

  1. Introduction
  2. Language
  3. Correlation
  4. Sampling
  5. Experiment
  6. Theory

Thus, the topic of correlation constitutes a good chunk of what I know about scientific method, and the Companion to Correlation was making great progress toward completion, with some delightful graphs being drawn along the way, as for example the four below, one each from Chapters 5, 6, 7, and 8, presented here for their aesthetic appeal alone.  I am told that these graphs will mainly give the impression that Companion to Correlation is technical and inaccessible, than which nothing could be further from the truth the book is written for the layman who lacks all background in statistics or research methodology, and it relies on no more than elementary mathematics, and the graphs below that may look difficult are approached in such gradual steps as to be within the easy grasp of an average ten-year-old.  Come to think of it, it should be taught to all ten-year-olds, as there are few tools that they will find more useful throughout their lives, no matter what paths they follow.

It is important to understand that the study of scientific method is not an ivory-tower pursuit, but rather finds application not only in the research laboratory but in everyday life; and in everyday life, not just once in a blue moon, but once every hour of the day.  Every day the information media, or our acquaintances and relations, recommend to us cures for our aches and pains, remedies for prolonging life, directions for toilet-training our children or educating them, and so on, based on some evidence or other, new or ancient, originating from a laboratory or from casual observation and any recipient of such advice who is uneducated in scientific method finds himself defenceless, and is left little option but to believe whatever is put before him, whereas in contrast someone educated in scientific method is readily able to distinguish the true conclusions from the false, the proven generalizations from the gratuitous, and the safe recommendations from the dangerous.

As a matter of fact, the topic of scientific method has already been broached twice on the pages of the Ukrainian Archive, once with respect to Morley Safer's foolish recommendation that we prolong our lives by drinking several glasses of wine each day, and again in the discussion of the manganese poisoning of Canadians (already alluded to above), and in each case what was being clarified on the pages of UKAR is a question that happens to be dealt with on the pages of Companion to Correlation the question of distinguishing correlational data from experimental, and understanding what cause-effect conclusions are justified in each case and what aren't.

Most unfortunately, at the moment my work on Companion to Correlation is suspended in the face of a fresh provocation that is impossible to ignore, which is the law suit initiated against me by Steven Rambam.  This law suit is without merit and certain to be defeated, but its mere existence distracts me and draws me back to UKAR's classic focus which is response to attacks upon Ukraine and upon Ukrainians, among whom I am one.

Nevertheless, on the assumption that the abomination of Case Number BC271433 will soon be defeated, work on Companion to Correlation may be expected to resume in the not too distant future.

Display 5-14.  Alternative portrayals of an ecological correlation between literacy and suicide over 16 Italian provinces (Durkeim, 1897/1951, p. 164).  The Goodman square in Graph K reveals that all the data fall into Triangle B, and furthermore that the data points effectively fall on the X-axis, a configuration which, though yielding a very precise estimate of the individual correlation, is unhelpful because that estimate spans zero.

Display 6-3.  Graphic presentation of a can cross region in the case of a strong manipulation on the left, and for alternative cutting points in the case of a weak manipulation on the right.

Display 7-2.  In egalitarian correlations (the scatterplot extends from one but region to another), regions of incongruity increase when comparisons are whole-group.  In hierarchical correlations (the scatterplot extends from one and region to another), regions of incongruity increase when the comparisons are sub-group.

Display 8-8.  Data from Farr (1852) demonstrate that the correlation coefficient computed when the underlying relation is non-linear (Graph A) is deflated in comparison to the correlation coefficient when one of the variables is transformed so as to make the relationship linear (Graph B).