October 14, 2004

Statistical Methods

Posted by Kieran

Maria’s post about required statistics courses reminds me of a possibly apocryphal story. I think it concerns one of the very early British social surveys of urban poverty by Charles Booth, or Mackintosh or one of those guys. The results were resisted by many for political reasons, and one strategy was to discredit the new-fangled methods they relied on. Thus, one critic in (I believe) the House of Commons asserted that he could not find the results credible because the report “only relied on a sample of the population — and a mere random sample, at that.”

If anyone knows the source of this (doubtless mangled) story, let me know in the comments.

Posted on October 14, 2004 03:58 AM UTC
Comments

Kieran,

For a more recent example,:
http://www.census.gov/Press-Release/cb97-c25.html

- Of course I’m disgusted - haven’t you been paying attention?

Nathaniel

Posted by Nathaniel · October 14, 2004 05:56 AM

My favorite example in this vein is Supreme Court Justice Lewis Powell’s ridiculing any argument based on statistics as “numerology.” In this view, its far better to rely on a judge’s “gut” than data.

Posted by pj · October 14, 2004 02:00 PM

George Bush’s “fuzzy math” allegation against Al Gore in the 2000 debates?

Posted by praktike · October 14, 2004 06:27 PM

Kieran,

I’ve not heard the quote, but from the context you provide, it sounds like it could have related to G. Udny Yule’s 1899 study of pauperism in England.

See here for more details on the study.

Posted by Ross · October 14, 2004 06:37 PM

This is not a response to Kieran’s specific question, but a reference to an extremely interesting and useful book for anyone interested in statistics. (I’d have posted this earlier, but have been working from home with a bad back and the book is in my office.) The Lady Tasting Tea: How Statistics Revolutionized Science in the Twentieth Century W. H. reeman, 2001), by David Salsburg, is about the uses (and some of the abuses) of statistics in helping us understand the world around us.

Posted by Donald A. Coffin · October 15, 2004 06:37 PM

Nathaniel,

The debate about whether the Census should use statistical sampling was not merely about statistical ignorance, but rather about the very phrase in the Constitution which established the Census Bureau, which requires an ‘actual enumeration’. Not a statistically sampled ‘count’, but an actual physical count.

It may be gratifying to put the argument against sampling down to ignorance, but not entirely accurate I think….

Posted by Don · October 16, 2004 03:16 PM
Followups

This discussion has been closed. Thanks to everyone who contributed.