Nonresponse Bias
Nonresponse Bias is the skewing of the results of a survey because so many people refused to participate (the "nonresponse rate") that the people who did participate did not constitute a representative sample.
The problem has gotten worse over the last few decades, because nonresponse rates have been steadily increasing. Other things being equal, the higher the nonresponse rate, the greater the bias.
The average nonresponse rate is now at or near 50 percent. That is to say, if you hire a survey, the survey will probably survey only one-half of your target audience. For purposes of statistical accuracy, this is not good.
There are various reasons for the steadily increasing refusal: people are busier, more distracted, more cautious, less generous, more cynical and more paranoid – and tired of so many surveys.
Nonresponse rates are especially high when a survey involves politics. As the federal government has grown larger and more intrusive, more and more people fear that the "researcher" asking the questions might be an employee of, or an informer to, a federal police agency.
Also, the "political correctness" of some surveys scares people off. Educated people are aware that one careless remark or even an innocent word (for example, responsibility) can be falsely construed as racism – and can ruin a career.
So, when people fear that the person on the other end of the telephone connection is a cop or a snitch, they tend to refuse to participate at all, so as to avoid the minefield of forbidden words, phrases and thoughts. Or, they will participate but will refuse to answer questions they consider dangerous.
In short, audience samples for surveys are becoming a lot like juries: dominated by the bored, the lonely, the docile, the unemployed, the simple-minded, the senile and the ignorant.
What Nonresponse Bias means to you, at the very least, is that you need to engage an expert researcher. Don't go it alone. Experts usually can mitigate the skew by assessing the solidity of the sample, by drafting questions skillfully, and by other means.
Return from Nonresponse Bias to Barriers to Accuracy
Home
|