This blog has moved to:

« Spinning "Convergence" | Main | Which SEC Roadmap Comments Matter, and What They are Saying »

July 20, 2009



Regarding your comments on the letters sent to the SEC about IFRS adoption:

Your analysis of the comments goes to show that a multi-disciplinary framework for thinking about problems is a major need, even for the smartest specialists. What I mean is that a cursory knowledge of psychology will show the flaw in your argument that because most of the comment letters were negative, then most of the public companies in the US are against IFRS adoption.

The flaw is the existence of a self-selection bias. Most companies may simply not care much about IFRS, or see it as an inevitability, or only favor it slightly. However, those who feel strongly against or for, will make their views known. So, all that we can reasonably see from the results is that more companies are fanatical in their resistance to IFRS than are fanatical in their agreement with IFRS.


Excellent post, Tom. Keeping the fingures crossed here. Hat tip to Mary Schapiro -)


I'd go one step further than commenter Andy above, since we know that Tom Selling knows perfectly what self-selection biais is.

Compare the relevant paragraph from the current post to the following comment: 'Let's agree that the population of interest, audit committee members of public companies, number very approximately 10,000. Only 253 of them volunteered in response to invitations to participate. Given the low participation level, it is highly likely for non-respondents to have significantly different opinions than the self-selected volunteers. If anything was done to test that proposition (yes, academics routinely apply techniques for doing so), CAQ did not report it.'

The latter was written by Tom Selling on 6 April 2008 in a post entitled "Low Quality Stats from Center for Audit Quality" (

I estimate that there are very approximately 10,000 public companies in the US (the order of magnitude is the only thing I'm interested in). There were 240 comment letters about the SEC's roadmap, of which about half were from public companies. Yet, in contrast to Tom's criticism of the CAQ's suvery quoted above, in this case we are supposed to believe that self-selection is just not an issue and that the respondents to the SEC's public consultation are necessarily (mathematically) representative of the overall population.

Remember, last time self-selection was "highly likely" due to the small sample size but this time "any errors due to small sample size would have to be tsunami-sized in order to reverse the results" - even though the sample size and overall population size are essentially the same in both cases.

One of these claims must be wrong. Which one is it, Mr. Selling?

The comments to this entry are closed.

This blog has moved to: