How to Mislead with Statistics
The number taken out of context is one of the staples of statistical deceit (↓). George Will gives us a fine example.
The nonpartisan Center for Responsive Politics reports that in 2004, of the top five organizations in terms of employee per capita contributions to presidential candidates, the third, fourth and fifth were Time Warner, Goldman Sachs and Microsoft. The top two were the University of California system and Harvard, both of which gave about 19 times more money to John Kerry than to George W. Bush (New York Post, Online Edition, 28 Nov 2004).
The Center for Responsive Politics, unlike some of the “nonpartisan” outfits cited by columnists like Will, really is nonpartisan. It tracks contributions to candidates and parties, and is essential for anyone who wants to understand to whom and from whom the mother’s milk of politics flows. A look at the tables of contributing institutions shows that among the top overall donors, the University of California (which refers, I take it, to the entire system, and not just to Berkeley) is no. 35 in total donations, with $1,567,964, all of which were individual (and not through PACs). Harvard is no. 85, with $960,002, all of them individual. By comparison the top ten were:
|2||National Assn of Realtors||$3,821,689||98%||2%|
|5||JP Morgan Chase & Co||$2,644,714||31%||69%|
|8||National Auto Dealers Assn||$2,471,050||100%||0%|
Will was clearly looking for some statistic or other that would make academic liberals look powerful. But all those deep-pocketed profs, secretaries, and custodians at Harvard and Cal managed to come up with only a little more than the National Auto Dealers, three-quarters of whose money went to the Republicans. They were distinctly outclassed by the bankers and brokers, who the last time I looked were not manning the barricades (though Goldman Sachs did contribute 60% of their money to the Democrats) (↓).
Anyone who doesn’t know what “per capita” means, and probably some who do, might well gather from Will’s quotation that the Democrats got more money from Harvard and Cal than the Republicans did from Time Warner, Goldman Sachs, and Microsoft. They didn’t (↓).
On the other hand, a look at the breakdown by party shows that Will is, if anything, too sanguine about the liberal threat. All the organizations mentioned by him contributed more to the Democrats than to the Republicans. So I suppose we need diversity among the brokers & geeks too. Such fertile ground here for fantasies of persecution! But thank goodness for Wal-Mart, UPS, SBC, Deloitte Touche Tohmatsu, Merrill Lynch, and the AMA!
The misuse of numbers is not the only meretricious turn in Will’s column. But this little bit of stable-cleaning is all I have energy for tonight. See Juan Cole’s response (and earlier pieces at History News Network); the passage from Will is quoted favorably by Thomas Galvin, a fan of David Horowitz’s attempts to promote an “academic bill of rights”. Check out the complaint form. You are being watched.
(↑) Darrell Huff’s How to lie with statistics (New York: Norton, 1954) is still a classic in the don’t-get-hoodwinked genre (BB, ABE, Pow). Newer works in the same vein include Joel Best’s Damned lies and statistics (BB, ABE, Pow) and John Allen Paulos’ A mathematician reads the newspaper (BB, ABE, Pow).
(↑) The point of Will’s column is that the academy is overwhelmingly liberal and that, since political “diversity” is a Good Thing, something Ought To Be Done to bring Balance to the Halls of Academe (somehow Will’s prose brings out the George Ade in me). That outfits like Goldman Sachs contribute heavily to the Democrats shows, however, that the mere fact of contribution to the Democrats implies very little about the specific politics motivating the act.
(↑) The Republicans received $3.8 million in contributions from Goldman Sachs, Time Warner, and Microsoft. The entire contribution of Cal and Harvard to both parties was about $2.6 million.