Lying With Statistics

2011

Lying with statistics is as old as statistics themselves. Unfortunately it has perhaps become more prevalent as we have become more mathematically illiterate, or at least less inclined to "do the math" ourselves. As a result, 63.7% of statistics are just plain wrong. I invented that, but only to make the point that one of the easiest ways to lie with statistics is to simply invent them. Making them appear more "precise" also makes them more believable too. That's why I didn't round that off to 64%.

The first two lessons here then, are to watch for unrealistically precise figures when you read or hear statistics, and see if there is a valid source. Who can actually say to the nearest hundredth of a percent how overweight a population is, for example, or exactly how many homeless people there are in the country?

What else should you be on the lookout for? Consider an example pulled from the May 2008 issue of Sierra Magazine. It had a small piece on the "hyper-consumptive, carbon spewing ways" of Western countries," which included the following quote: "Stephen Pacala, director of the Princeton Environmental Institute, claims that the planets richest 700 million people - a mere 7% of the world's population - are responsible for half of the global greenhouse-gas emissions produced by fossil fuels." Did you catch that? If 700 million is 7% of the world's population, we suddenly have 10 billion people in the world, right? Actually there were about 6.7 billion when the article was written.

This is a common mistake in reporting found in magazines and newspapers. It's a simple error most likely, but it does seem that the mistaken figures are often convenient for making the desired point. After all, it sounds bad enough if 10.3% of the world's people produce over half of carbon emissions, but it's a politically a stronger point if a mere 7% are guilty of this. So look for this kind of common mistake and/or manipulation of the statistics in news stories.

For other examples of lying with statistics, just watch election reporting on television news broadcasts. I've seen the vote percentages given for the Republican and Democratic candidates add up to 100% in major elections - even presidential races. For that to be true there would have to be no votes for any other candidates in other parties. But these other parties often get several percent of the votes, so what the news organizations are apparently reporting is the votes that they think are important. Manipulation like this gives the impression that there are no other political parties, but there were a dozen that ran presidential candidates last time I voted (2008).

You also need to watch for the more subtle forms of lying with statistics. Consider this question: If a company's profits go from 3% to 6%, did they rise 100% or 3%? It's true that profits doubled, which is a rise of 100%. But as a return on equity (or a portion of sales; depending on which measure is used) they moved only from 3% to 6% - still a poor rate of return.

Cases like these will be reported according to the political slant of the news organization or reporter. To attack the company for making too much money, they can write a headline that says; "XYZ Company sees Profits Increase 100%!" It's an accurate statement, but it hides a lot of truth.

Probably lying with statistics is most evident in opinion polls. The manipulation of the truth is accomplished with the phrasing of the questions. For example, suppose we ask a thousand people, "Should the government help people who face losing their homes to foreclosure?" You can bet that the number of people who say yes would be higher than if a thousand were asked, "Should you be forced to pay more taxes to help people make the payments on their houses when they have trouble?" Both might accurately describe what a proposed program would aim to do, but the nature of the question would certainly change the resulting answers to such a poll.

While watching the evening news or reading newspapers and news magazines, stop at each statistic given, and ask a few questions. Is it likely to be accurate? How it was figured? Does it tell the whole truth? What other ways could the relevant things be tallied? Look for other information online or in alternative news sources. It is certain that lying with statistics will continue, but you don't have to be mislead by these lies and half-truths.

Was this page interesting or useful? Please let others know with one of these...