Journaltalk - Statistical Significance in the New Tom and the Old Tom: A Reply to Thomas Mayer

Statistical Significance in the New Tom and the Old Tom: A Reply to Thomas Mayer

About this article

  • Deirdre N. McCloskey and Stephen T. Ziliak
Keywords Significance tests, t’s, p’s, confidence intervals, oomph
Volume Number 9
Issue Number 3
Pages 298-308
File URL Statistical Significance in the New Tom and the Old Tom: A Reply to Thomas Mayer
Publication year 2012

Flag this article

Flag this article for moderation.

Close this.

About Econ Journal Watch

Grouping social sciences
Categories economic, economics

Flag this journal

Flag this journal for moderation.

Close this.

Add a comment to this discussion.


  1. The misuse of significance tests in econometrics is part of the larger problem of formal criteria that have been imposed and must be respected if one wishes to publish, at least in a mainstream journal. For example, in macroeconomics one is expected to base ones model on the assumption of a rational, utility maximizing representative agent, even though there exists a vast literature that demonstrates that the representative agent cannot be justified by any plausible set of assumptions. The requirements change as research programs change, but they never have a scientific justification. In the post-WWII era for example, it was thought that a macro-econometric model, to be of any use, must have hundreds, perhaps more than one thousand equations.

    I believe that the slave like adherence to absurd formal criteria is a consequence of the recruitment practices in economics over essentially the past half century. There was a vast expansion in the number of professional economist, while the principal criterion of admission was some degree of mathematical competence. A specific interest in economics was not required. Many who became professors of economics must have chosen this career in the same way as one chooses to become a dentist or a lawyer. Their education consisted essentially of learning to solve the problems that they were presented with, without much thought as to their relevance. This is how they continued to work as professional economists. If they had learned how to formulate and estimate, say a rational expectations model, including significance tests, then that is what they did and were expected to do in order to be able to publish in mainstream journals.

    I think the problem viewed this way is rather simple, but unfortunately unsolvable, at least over the foreseeable future.

    Claude Hillinger

    posted 24 Sep 2012 by Claude Hillinger

Log in to Journaltalk to discuss this article!

Don’t have a Journaltools account? Sign up now.


Log in to Your Account

Member login

feed Jt Article Discussions

31 Mar

Christianity Changes the Conditions of Government
Trygve Hoff's Appeal to Ragnar Frisch: Four Letters from 1941
Classical Liberal Think Tanks in Greece, 1974–2024
Classical Liberalism in Russia
Revisiting Hypothesis Testing with the Sharpe Ratio
What Is the False Discovery Rate in Empirical Research?
Journaltalk: Opening the journals to civil voices everywhere!

All contents © 2024 by Daniel Klein unless otherwise attributed. All rights reserved.