he random variation you get whenever you
deal with people's responses?
You never know for certain which is which, but a statistical test is an objective way of being
reasonably sure, one way or the other. Statistical tests use the known mathematical properties of
numbers to let you decide when a difference is probably due to chance, and when a difference is so large
that chance seems unlikely. The exact mathematics aren't of interest to the general reader, but only the
outcome, the probability figure.
If the outcome of a particular test could have happened by chance only
five or fewer times in a hundred trials (conventionally expressed in this book as p<.05, probability equal
to or less than 5/100),* we begin to doubt that this is chance variation. It probably represents a real
difference between the groups. If the probability is even smaller that the outcome is due to chance, say
less than one in a hundred (p < .01) or less than one in a thousand (p < .001), we can feel quite certain
that we are dealing with real, important differences.**
(1 of 2)4/15/2004 7:01:42 AM
On Being Stoned - A Note
Thus in this book the lower the probability figure in parentheses, the greater the difference between
the groups being compared.
Footnotes
*More exactly, the sign should be [less than or equal to