David Pribyl, Head of Technical Services at the Roosevelt University Library, recently interviewed Stephen Zillak about his latest publication:
Stephen Ziliak, RU
Professor of Economics, has published a new book, copies of which are available
in the library: The Cult of
Statistical Significance: How the Standard Error Costs Us Jobs, Justice,
and Lives (University of Michigan Press, 2008). By Stephen T. Ziliak
and Deirdre N. McCloskey. David Pribyl, liaison with the economics
faculty and Head of Technical Services in the library, interviewed Ziliak,
known as "Prof Z."
RU: How, if at
all, can a book on statistics deal with issues of social justice?
Prof Z: Good
question! For years, I, like most people, did not see any
connection at all between statistics and social justice, or, to
say it more generally, between statistics and ethics. It wasn't
until 1989 - while working at the Indiana Department of
Workforce Development - that I saw for the first time how deeply human the
seemingly arcane & "objective" world of statistics
is.
I worked in the
Labor Market Information Division, where the official state & regional
labor force statistics are generated and disseminated in cooperation with the
U.S. Department of Labor. One day a citizen telephoned.
He asked me to provide unemployment rate statistics for black youth in each of Indiana's major
cities. Young men and women between the ages of 16 and 21.
I couldn't find the statistics he asked for. Now Indianapolis, Gary, Fort Wayne, Evansville had then as now tragically high amounts of minority unemployment - and
very high for young African American men in particular. No one would
be surprised if the average city rate exceeded 35 or 40% while the national
rate - for the whole economy - was only around 5%. So
where were the statistics proving it?
Finally I went to the
chief of my division. He made a phone call to the Chicago branch of the U.S. Department of
Labor, to a chief who was next in line in the chain of
command. The chief in Chicago replied "given the lack of statistical significance in the black youth
unemployment rates, the Department of Labor does not disseminate those
rates. We don't publish statistically insignificant results, only
'significant' results," he said. I was embarrassed to return
to the telephone to tell the citizen what I learned.
What the Labor
Department was saying is that if a rate is not estimated with an arbitrary
amount of precision then they won't publish or disseminate it. In
technical terms, the p-value had to be equal or less than .10. This
standard they glue to even if other evidence suggests that the rate is so
high as to constitute what Martin Luther King Jr. called a "Great
Depression in the economy of the ghetto." It was a life-changing
experience for me. I realized that important social issues get
pushed under the rug for pseudo-scientific reasons. But as I show in The Cult
very few scientists have had this kind of realization. That precision
is not the end all be all of human decision making. Like most members of
the cult of statistical significance, the DOL economists and statisticians - my
friends & former bosses - do not possess or implement a scientifically
legitimate or socially just decision rule. Zero, nil, null,
nothing. It has to change.