Picture this: You’re reading your Sunday morning newspaper, sipping a cup of coffee when you stumble across an article claiming coffee drinkers are 70 percent more likely to contract cancer than non-coffee drinkers. You pour out your coffee and hasten to dispose of the remaining coffee in your kitchen – you will take no chances when it comes to your health.
But there is a problem with the article and the statistics the reporter used. Apparently, the scientists and researchers only observed 200 patients, 30 smoked, two lived near chemical plants and 18 were regular tanning-bed visitors. The researchers failed to mention this in their published research and a sloppy reporter used this information, failing to double check to be sure of accuracy.
This example is a little exagerated, but there are plenty of outliers and underlying factors that can skew a survey, observation, experiment or poll. As a journalist striving for fairness and accuracy in his or her reporting, it is imperative to make certain that your information is as infallible as possible. As humans we will make mistakes, but we should never fail to strive for 100 percent accuracy in our reporting.
In her helpful handbook, “Math Tools for Journalists,” Kathleen Woodruff Wickham devotes a chapter to polls and surveys. If used accurately, a poll or survey can add significantly to an article, but when used incorrectly it can hurt the reputation of the reporter and misinform readers.
Using Polls and Surveys Appropriately
You’re in a hurry. You have a deadline to meet. You need a survey for a story and http://www.crazysurVEYS.com is the first Web site you come across. You’re first instinct may be to quickly pull off some survey and the results and insert it into your story so you can submit the article on time. One of the simplest and most important things to consider is the foundation, organization or institute that conducted the poll or survey.
It is also wise to look up the person who paid for the poll, how recent is it? what was the demographic? how were questions worded? what was the sample size? what methods were used? what was the margin of error? These are all important things to consider.
If you’re striving for reliability, it is also important to check on the random selection process of the poll. Random selection means there was equal opportunity for anyone to take part in the poll.
A large sample is needed for a survey or poll when trying to represent a population. Pollsters strive for at least 400 participants for accuracy and an acceptable margin of error.
Margin of Error
500 random adults were polled on their political affiliations. 48 percent claimed to be Republican and 52 percent said they were democrat. Is it correct to say there were more democrats than republicans?
No. For a poll with 500 participants, the confidence level is 95 percent and the margin of error is 4.4 percent.