Saturday, 12 November 2016

Losing vector directions

 

A few hours ago I posted up a commentary on a paper:

Brad Verhulst, Lindon J. Eaves, Peter K. Hatemi.  Correlation not Causation: The Relationship between Personality Traits and Political Ideologies. Am J Pol Sci. 2012; 56(1): 34–51.

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3809096/

Somewhat warily I added: I have scrabbled around for some guidance on this (personality and political attitudes) and came across a single paper to start the ball rolling. You find the meta-analysis on political attitudes and we are well on our way.

Sure enough, a single paper is an insufficient basis for a comment, particularly when the authors in an Erratum later admit that:

The interpretation of the coding of the political attitude items in the descriptive and preliminary analyses portion of the manuscript was exactly reversed. Thus, where we indicated that higher scores in Table 1 (page 40) reflect a more conservative response, they actually reflect a more liberal response.

http://onlinelibrary.wiley.com/doi/10.1111/ajps.12216/pdf

My thanks to reader LemmusLemmus.

I have written to the author of the Erratum to ask if the original miscoding was intended as a projective test for readers. I will await the response, but the paper, in that polite phrase, is “not suitable for hypothesis testing”, and in my view should be set aside.

A.E. Maxwell would have said “get to know your data before worrying about statistical tests”.

 

 

1 comment:

  1. Rutherford said that your experimental data needed statistical treatment, you should have performed a better experiment.

    ReplyDelete