Know how to use numbers

Wait a minute: Nobody said there would be math! Many of us were drawn to writing—and journalism—because we’re not so good with numbers.

But numbers are a critical part of the work we do as reporters. Numbers may seem dull or intimidating, but when you know how to read and interpret them, you become a much smarter journalist. There are stories hidden in city budgets, profit and loss statements, and politicians’ tax returns. Poynter Online blogger Chip Scanlon once counted how many of the stories that appeared in his local paper on a single day involved numbers. The answer was more than a third, and he found them not just in the business and sports sections but in every part of the paper.

Gauge your math acumen with this basic test for journalists: www.ire.org/education/math_test.html. If you’re not thrilled with your score, read on. Here are three numeracy skills that will immediately help you become a better reporter. Overcoming a fear of numbers will help you as a journalist…we’re 100 percent sure!

Percentages

Not only do we use percentages to show change, we also use them to identify change and thereby identify news.  So percentages are important to understand and get right. Most individual numbers only take on real significance in comparison to other numbers. For example, if you read there have been 10 murders in Tribeca so far this year, you don’t know if that’s a lot or a little.  If you read that 10 murders represent a 50 percent increase over last year, the story becomes clear.

Percentages give us perspective on the relative share or size of something on a scale where 100 percent always represents the whole thing, or the original. It’s another way of presenting a fraction. 100 percent is the whole thing. 50 percent is half. 25 percent is one-quarter. If a number is cut in half, the result is a 50 percent decrease. If a number doubles, that’s a 100 percent increase (not, as many assume, a 200 percent increase).

Those among you who are math whizzes can skip ahead to the next section. For everyone else, here’s one commonly used way to find a percentage change. Look at your two numbers: your original number and your new number. Subtract the original number from the new number. Then divide that by the original number. Multiply by 100, and that’s your percentage change.

Let’s say the school budget was $10 million last year and is going to be $11 million next year. The difference is $1 million. Divide that by the original number, 10. That gives us 0.10. (Note the decimal point, it’s important.) Multiply by 100 and you get 10. So the school budget is going up by 10 percent.

Resources: For a helpful primer on percentages, including common mistakes people make in calculating them, see http://stats.org/faq_percentages.htm. Another of our favorite online sites is www.robertniles.com/stats, which covers percentages, as well as averages, data analysis and other mathematical concepts, especially for journalists.

Polls

A poll is an attempt to use the opinions of a small group of people to gauge what a larger group thinks. Such a quest, if not carefully conducted can draw questionable conclusions. Here are some key concepts:

• Probability-based sample: For a poll to be useful, the method of choosing participants must be such that anyone in the target population has a chance of inclusion. A “probability-based sample” is just a fancy term for random selection. Random-digit dialing is the dominant method for polling U.S. adults, but emerging factors such as phone-number portability make it difficult to include cellphones, and because a growing swath of young Americans are cellphone-only users, this surveying technique is becoming less representative. Still, it is more accurate than self-selected samples such as people who choose to answer an online survey.
• Margin of sampling error: Every poll has a margin of error. The bigger the pool of respondents, the smaller the margin of error. For example, a poll that surveys 400 people has a margin of error of plus or minus 6 percent, whereas a poll that surveys 2,500 people has a margin of error of plus or minus 2 percent.
• How to “read” a margin of error: You must add and subtract the margin of error from the results. So when the margin of error is plus or minus 6 percent, that means that if 73 percent of people surveyed answer yes, all that can be known is that between 67 percent (-6) and 79 percent (+6) of people in the larger population would be favorable.
• Why 1,000 respondents is the sweet spot: When respondents are in the hundreds, the margin of error is wide. But when a poll reaches about 1,000 people, it levels off at plus or minus 3 percent and climbs slowly from there. Any increased accuracy after 1,000 people requires a lot more polling (read time and money), so that number has become the norm.

Resources: The Poynter Institute offers a three-hour online class for journalists on polling, free of charge: http://www.newsu.org/courses/understanding-and-interpreting-polls.

Scientific Studies

Journalists cover the results of medical studies all the time, but too often their lack of perspective leads to confusing stories. “Such-and-such thing is bad for you,” one report goes, and then months later comes the news that the same thing is actually beneficial. Here’s how to interpret studies more critically.

• Causation vs. correlation: Just because two things are associated does not mean they are related. Yet so often when researchers find that, say, people who drink a lot of coffee are more likely to have some kind of health problem, journalists erroneously assert that coffee causes the health problem. In fact, the news is simply that coffee is associated with the health problem—there may be no causal relationship whatsoever. Our favorite example of this kind of fallacious reasoning is the observation that most children are taller at the end of kindergarten. But does that make it reasonable to conclude that kindergarten makes children grow?
• Relative vs. absolute risk: When the risk of something is very low, even the tiniest increase can sound dramatic. For example, let’s say that in Cerealville, USA, one person per year, on average, chokes on a Cheerio. Now imagine that all of a sudden, two people start dying that way each year. The Cheerio choking risk has doubled. That sounds awful, doesn’t it? People in Cerealville are twice as likely to choke on a Cheerio as they were in the past. The relative risk (compared to what it had been) is up 100 percent. But the absolute risk is still unimaginably small—in our hypothetical, .00002 percent instead of .00001 percent. It’s important to analyze, and provide readers with, the absolute as well as the relative risk.
• Statistically significant: This term refers to how likely it is that a finding is not due to chance.

Resources: An extremely helpful book is News & Numbers: A Writer’s Guide to Statistics, by Victor Cohn and Lewis Cope. The third edition of this classic will be released in September 2011 and can be pre-ordered at amazon.com. Another useful book is Numbers in the Newsroom: Using Math and Statistics in News, available at the online bookstore of Investigative Reporters and Editors (IRE): http://www.ire.org/store/books/.

An invaluable web resource is STATS: http://stats.org. This non-partisan, non-profit site focuses on helping journalists think quantitatively, and provides many well-explained backgrounders.