"Do loose numbers do more harm than good?"

August 18, 2002 3:08 PM

"Do loose numbers do more harm than good?" That's the question asked by Norimitsu Onishi in a thought-provoking article in today's NY Times (reg req). Inflated numbers have often had an impact on policy and people's thinking, but when the truth comes out it can make a difference, for good or ill. (More inside.)
posted by languagehat (8 comments total)

The early estimate of six million Jews killed in the Holocaust was revised downward to five million by Raul Hilberg in his classic The Destruction of the European Jews, but most people continue to use the higher figure; I think we can agree that in this case it makes no practical or moral difference. However, Onishi, after citing some hair-raising examples of guesstimation in relation to Africa ("'Extrapolating from 1.5 million to 20 million, it's shoddy, but it's the best we can do right now,' Mr. Roberts said"), gets around to the case of the alleged 15,000 child slaves in Ivory Coast cocoa plantations. It turns out:

almost all children working in cocoa fields were children of the plantation owners, not forced laborers.

As for child workers unrelated to the plantation owners, the study found that brokers had placed 2,100 foreign children, most of them ages 15 to 17, in Ivory Coast's cocoa plantations. Ninety-four percent of the children, the study says, knew the intermediary, or broker who hired them for the plantation work.

"The most frequent reason given for agreeing to leave with the intermediary was the promise of a better life," the report says. It adds: "None reported being forced against their will to leave their home abode. One hundred percent indicated that they had been informed in advance that they were going to work on cocoa farms."

Jim Gockowski, an American agricultural economist who led the study for the Institute of Tropical Agriculture, said, "By and large, the cocoa industry didn't deserve the rap it got."

I'm sure the people who first spread the 15,000 figure thought they were doing good for child laborers, but when the truth comes out it makes people skeptical of all such statistics, just as when the "Belgian atrocities" the Germans were supposed to have committed in WWI were revealed to be fabrications, it made people reluctant to believe early reports of the Holocaust.
posted by languagehat at 3:20 PM on August 18, 2002

or, more concisely: do lies matter?
posted by quonsar at 4:10 PM on August 18, 2002

Consider Nigeria. Everyone agrees it is Africa's most populous nation. But what is its population? The United Nations says 114 million; the State Department, 120 million. The World Bank says 126.9 million, while the Central Intelligence Agency puts it at 126,635,626.

I am impressed by the CIA's precision. It must be the satellites.
posted by srboisvert at 4:42 PM on August 18, 2002

Aargh, more new numbers.
posted by pekar wood at 7:15 PM on August 18, 2002

or, more concisely: do lies matter?



Whoever gets their lies/numbers out first has the upper hand. Revised figures to not, it seems, get the same attention so people latch on to the incorrect values. Plus, if you speak first, counter-figures come across as reactionary or sour grapes or spin or whatever you choose to call them.



It's better to play offence than defense.
posted by Ayn Marx at 8:34 PM on August 18, 2002

As long as you're looking only at the short term, that is.
posted by languagehat at 5:06 AM on August 19, 2002

Whoever gets their lies/numbers out first has the upper hand. Revised figures to not, it seems, get the same attention so people latch on to the incorrect values. Plus, if you speak first, counter-figures come across as reactionary or sour grapes or spin or whatever you choose to call them.
so, lies are ok if you can justify them.
posted by quonsar at 5:38 AM on August 19, 2002

"Loose figures" are a subtle kind of appeal to authority. They give an argument an appearance of truth but no real substance. The problem is that to the uncareful observer random statistics look much like the real thing, data supported by evidence, reported with statistical confidence.

This is a blind spot for the press, but they are at least aware of the problem---polls are often reported with confidence limits and error estimates ("...plus or minus 1 point, nineteen times out of twenty"). Interest groups and NGO's, however, are stupendously awful in their public statements. Does anyone take "figures" from either the NRA or the Sierra Club seriously?
posted by bonehead at 5:51 AM on August 19, 2002