How to Read Lies, Damned Lies, and Statistics
September 11, 2012
Notes from the Research Desk highlights the best practices, salient data and emergent perspectives uncovered by the Government Business Council’s (GBC) team. Each week, Research Manager Dana Grinshpan will share the discoveries most important to federal managers.
The 19th century British Prime Minister, Benjamin Disraeli, was the first quoted as saying, “There are lies, damned lies, and statistics”. His point? Anyone can use statistics to prove just about anything. Because statistics have a bad reputation, due to their penchant for misuse, it’s important federal managers be savvy consumers of numbers.
As the GBC’s Research Manager, I am responsible for designing survey instruments that yield statistics and my analysts are responsible for discussing these statistics in our reports. Therefore, one of my most important responsibilities is ensuring the numbers we publish are factual and easily understood. With inspiration from Joel Best, author of Best Damned Lies and Statistics, here are a few tips to help you sort out fact, fiction and spin:
Always ask questions. When you see statistics, ask questions! Take, for example, the ConAgra Foods Foundation: Have you heard that more than 16 million, or almost one in five, American children are at risk of hunger? Quite an alarming figure. But what does it really mean? What is the definition of at risk? Does it mean skipping lunch or skipping several meals? Is it based on income or food consumption? Be mindful of the assumptions implicit in some research findings.
Be wary of the dark figure. The dark figure is an expression employed by criminologists to refer to the proportion of crimes that go unreported, or that don’t appear in crime statistics. The general rule, however, can be used for reading all statistics. Is the statistic you are reading undercounting someone or some groups?
Suspect too many round numbers. Have you ever noticed that the numbers in political ads are always in denominations of 10, such as 50 percent, 60 percent, so on and so forth? Statistics can be messy, and often people like to round up or down to make numbers look pretty or more extreme. In practice, if you are reading something and every number is in denominations of 10, look deeper into what you’re reading and who it is coming from. Find the source.
Spot Mutant Statistics. Not all statistics start out bad. But some are misinterpreted or misrepresented, mutating them into something they don’t mean. When you read a startling statistic taken from secondary research, check the source and make sure that the scholarly paper or report phrased the stat the same way.
Find context. A lone statistic without context is also a suspicious thing. For example, what if I told you that in 2010, 33.9 percent of people between the ages 19 to 25 lacked health insurance? Is that a lot or a little? Maybe you are of the opinion that even one person is too many. But add this piece of information to the fray: the proportion of people from of the ages 19 to 25 who do not have health insurance is now 27.9 percent, down from 33.9 percent in 2010. All of a sudden, this doesn’t look so bad. Context matters—reserve judgment until you’ve seen more than one number.
What tips help you avoid getting bamboozled by slick statistics?
(Image via Prosotphoto/Shutterstock.com)
September 11, 2012