Although this is tangential to the present discussion,
the following paper from Economics will interest many members:
McCloskey DN, Ziliak ST. The standard error of
regression. J Economic Literature 1996;34:97-114.
The paper starts with an historical note on the
beginning of use of statistical significance (honour goes to Cicero in the use
of concept and to Venn 1888 for actual use of the word significance), contains a
list of questions to ask when judging the use of statistical significance and
the results from their survey of Economic literature.
I find this quote very relevant: "Economists have not
been fooled, even by their own mistaken beliefs about statistical significance.
To put it another way, no economist has achieved scientific success as a result
of a statistically significant coefficient. Massed observations, clever common
sense, elegant theorems, new policies, sagacious economic reasoning, historical
perspective, relevant accounting: these all have led to scientific success.
Statistical significance has not."
McClosky has written more about statistical
significance.
Gopal Netuveli
Department of Primary Care and
Social Medicine
Imperial College London
