Online College Courses for Credit

4 Tutorials that teach Nonresponse and Response Bias
Take your pick:
Nonresponse and Response Bias

Nonresponse and Response Bias

Author: Katherine Williams

Distinguish between nonresponse and response bias in a study.

See More

Nonresponse Bias

Video Transcription

Download PDF

This is the tutorial on non-response bias. Another word for non-response bias is participation bias. This is a type of selection bias. It happens when the people who do respond to a survey differ in an important way from the people who don't. Now the people who don't respond to a survey, it might be that they can't or they won't. But what matters is that they're different from the people who do.

This can lead to an unrepresentative sample. And additionally, if we have a high rate of non-response, then we can end up with a higher rate of bias. Researchers can potentially correct for this by trying to re-contact people who are original non respondents. Additionally, this effect might be stronger in some parts of the country than in others.

Let's go through two examples. Example one says, researchers mailed a survey about life habits to elderly residents. The residents that replied were healthier and fitter than those that did not reply. Perhaps people didn't reply because they didn't have the time with all their extra doctor's appointments if they're sicker. Or maybe the idea of having to get up and walk to the post office is too much for people who are elderly and not able to move as much.

So because the people who did respond, the healthier group, differs significantly from the people who did not respond, the more sickly and infirm group, the conclusions from this research would be heavily biased and inaccurate. The sample that we obtained was not representative. So as a result, the conclusions do not represent the population and aren't going to be that good.

Example two talks about a telephone poll. If you do a telephone pole during the workday, you're only going to reach people at home that aren't working. There could be a very big significant difference between the opinions of people who do work and don't work. So because the telephone poll is only reaching the people who don't work, this poll is not going to be representative of the whole population.

So this tutorial covered non-response bias, also known as participation bias. It's a type of selection bias where you have issues with the people who do and don't respond differing in very important ways. It's important to consider this and to try to correct for it. Because otherwise, your survey will reflect that bias and reflect the unrepresentative sample. And the conclusions that you would try to draw from it would not be accurate.

Response Bias

Video Transcription

Download PDF

This tutorial covers response bias. Response bias is the type of bias that comes from when people don't answer the questions truthfully or don't report data about themselves truthfully. There's a variety of different ways this could happen.

It could come from leading words in the question, could come from a desire to please the researcher, or from releasing information all yourself. You're concerned for your morality, you don't want people to know something about you. Let's look at a few examples.

This first example says "now that you've seen the video on inhumane treatment of animals, how much meat do you eat in a week?" This question is trying to lead the participant by first showing them a video of inhumane animal treatment and then asking them the question about it and referencing that video.

It's very clear that the researcher finds the treatment inhumane and might feel differently about eating meat than someone who doesn't find the treatment inhumane. So the researchers trying to push the people to answer either that they don't eat very much meat, something along those lines.

In example 2 it says "if you found $20 on the street, would you keep it? Or would you do the honest thing and return it?" Here it's a very small leading word, the honest thing. So the researcher is implying that returning the money is honest. So if the respondent wants to please the researcher, they would say that, yes, they would return it. They would do the honest thing.

Example 3 says "do you laugh when someone gets seriously hurt?" This is a pretty trivial example of some time when you'd want to conceal information about yourself. Maybe you do laugh when someone gets hurt. But you might not want to tell the researcher that, because you think that they are going to think negatively of you.

This issue also occurs a lot on questions about illicit drug use or anything illegal or sexual activity. People might want to conceal that information about themselves so they don't get judged by the researcher. This is when your tutorial on response questions.

Terms to Know
Nonresponse Bias

Bias that occurs when the people who were unable to be reached or unwilling to participate in a sample have substantially different opinions than the people who were included in the sample, resulting in a misrepresentation of the population.

Participation Bias

Bias that occurs when participation in a study is voluntary. People who feel strongly may be the only participants.

Response Bias

Bias that occurs when either (1) the question is poorly worded so that certain responses are over-represented, or (2) the respondent is confused by the question or feel like they should lie due to the sensitive nature of the question.