December 10, 2004

THE LATEST ISSUE of Econ Journal Watch is out -- download it here (pdf). Some highlights:
-- Letters on statistical significance by Thomas Schelling and Robert Gelfond with replies by Stephen Ziliak and Deirdre McCloskey. (pdf)

-- In a study of the indices of leading textbooks Dan Johansson shows that core Ph.D. theory is devoid of entrepreneurship, institutions, property rights, and economic freedom. Johansson sees this as rather like Hamlet without the Danish prince. (pdf)

-- Randall Holcombe dissects the National Research Council�s rankings of economics departments, and reflects on the impact on economic research. (pdf)

If you don't want to download the whole thing, you can download the Table of Contents containing individual links to the articles.

Quotable from the McCloskey & Ziliak letter (pdf):

We were pleased to read the letter by Tom Schelling, showing decisively with a new example that statistical significance is neither necessary nor sufficient for proving economic significance. Tom notes that the beast�The Standard Error�is still breathing. We can add a little precision: now it dominates in 80% or more of the published papers. We�ve got to kill it if economics is to progress.

We asked our old friend William Kruskal, a distinguished statistician and past president of the American Statistical Association, �How could the confusion of statistical significance for economic significance proliferate?� He replied, �Well, I guess it�s a cheap way to get marketable results.� Indeed. An economist couldn't put it better .. We propose an Angie's List of economists who understand the insignificance of statistical significance. We will ask a carefully constructed set of major economists and econometricians (every editor of every major journal, for example) to state publicly and for publication .. their support for the following propositions:

1. Economists should prefer confidence intervals to other methods of reporting sampling variance.

2. Sampling variance is sometimes interesting, but a low value of it is not the same thing as scientific importance.

3. Economic significance is the chief scientific issue in economics; an arbitrary level of sampling significance is no substitute for it.

4. Fit is not a good all-purpose measure of scientific validity, and should be deemphasized in favor of inquiry into other measures of importance.

We will publish the names and responses, or lack of response, of everyone asked. On reflection any economist and econometrician who understands basic statistics will of course agree with the implied standards. In the decades since 1919 that we and scores of others have been making this point, no one has ever been able to defend the practice of significance testing. True, many people have gotten angry at the challenge. But no one has actually met it.
Posted by Greg Ransom | TrackBack