Online College Courses for Credit

+
Bias

Bias

Author: Ryan Backman
Description:

Determine whether bias exists in a study.

(more)
See More

Try Our College Algebra Course. For FREE.

Sophia’s self-paced online courses are a great way to save time and money as you earn credits eligible for transfer to many different colleges and universities.*

Begin Free Trial
No credit card required

37 Sophia partners guarantee credit transfer.

299 Institutions have accepted or given pre-approval for credit transfer.

* The American Council on Education's College Credit Recommendation Service (ACE Credit®) has evaluated and recommended college credit for 32 of Sophia’s online courses. Many different colleges and universities consider ACE CREDIT recommendations in determining the applicability to their course and degree programs.

Tutorial

Video Transcription

Download PDF

Hi. This tutorial is all about bias. Let's start with the definition. Bias is the tendency for collected data to differ from what is expected in a systematic way. Biased data can often favor a specific group of those studied. Any time you are conducting a statistical study, either an experiment or an observational study, you want to avoid introducing any bias into your study. So whenever you can eliminate bias, that's good because bias is bad.

While most studies are conducted with integrity and diligence, sometimes bias can be introduced into a study in unintentional ways. Studies can be carried out and evaluated better if different types of bias can be recognized. So we're going to go through a couple different types of bias. I'll give you the definition, and then I'll give you an example of each.

So the first one we're going to look at is what's called selection bias. Selection bias occurs when some groups in the population are left out of the sampling of the process of choosing the sample. Selection bias yields an unrepresentative sample.

So let's take a look at an example where selection bias has occurred. A telephone-only survey is used to determine the views of the city's population on public policy. So what we want to think about is, is there any way in the way that this sample or that this study was conducted where a group was systematically left out of the possible sample space? And then yes. People without phones will be unable to be selected from the survey since it was a telephone-only survey.

And then specifically, many homeless individuals may not own phones. Since these members of the population would provide a unique view towards public policy and are unable to be selected, selection bias would occur.

So since people without phones and specifically the homeless people, since they do have-- their responses would affect the response, we do have some bias that is introduced here. So if you were running this study, you may try to think of some other way other than using the telephone to conduct your survey.

The next type of bias is what we call participation bias. Participation bias is a type of selection bias where one group of the population is unwilling or uninterested in participating in the study. When participation bias has occurred, there is an unrepresentative sample. So when participation bias occurs, people are-- have the opportunity to be in the study. They're just unwilling or uninterested.

So an example of that is that older adults would probably be less inclined to take a survey about sexuality than younger adults due to generational differences regarding the topic. Important information would be lost due to participation issues.

So since older adults aren't interested in taking the survey, the survey is going to only be able to represent younger adults. So this would yield an unrepresentative sample since we would have a far lower number of participating older adults than younger adults.

Another type of bias is what we call publication bias. Publication bias is a type of bias that occurs when researchers only publish positive results. So publication bias tends to overstate the effects of explanatory variables.

So an example of publication bias is suppose a researcher is interested in the effects of a new antidepressant drug by doing a meta analysis. Meta analysis is a collection of other relevant studies. It's where you take a bunch of studies that are studying the same thing and combine them together.

If a researcher includes only positive studies but does not include inconclusive or negative studies of the drug, publication bias has occurred. So that's going to overstate the effect of this antidepressant drug if the researcher didn't include the studies that were either inconclusive or painted that drug in a negative light.

A fourth type of bias is response bias. Response bias is the type of bias that occurs when the behavior of the respondent or the interviewer can cause response bias. The wording of a survey can influence responses, also causing response bias. So due to the behavior of the respondent or how the intervener responds or interacts with the participant and the wording of the survey can also influence responses.

So if you take a look at this, we have two examples here. So under Mayor Johnson's term, violent crime has risen 25%. Unemployment has risen 10%. Do you support Mayor Johnson? So we're going to get some response bias in the answers to this question because of the way the question is worded.

So there's two pretty negative-- violent crime and unemployment are pretty negative societal issues. So that may influence the responder into kind of more nos to support Mayor Johnson than yeses. So here's an example of when wording of the question can cause response bias.

The second question-- have you committed a felony in the past three months? This question is worded appropriately. But it's not going to be the behavior of the respondent that is going to introduce response bias here. The participant may not be interested in owning up to this felony or admitting to a felony on paper.

So even though they may have committed a felony, they're not going to admit to it. So in this case, you're going to have a disproportionate amount of people that say no when really they probably should say yes. So in this case, the behavior of the respondent would yield an unrepresentative number of nos.

Another type of bias is what's called the Hawthorne effect, a type of response bias where participants in the study behave differently just because they know they're in the study. So an example of this. Suppose that a sample of people is told they're testing an appetite suppressant. So as you're running the study, many in the sample may just begin to eat less because they know that they're in the study.

So something in their subconscious says, well, I'm taking this anti-appetite-- excuse me-- this appetite suppressant. I'm just going to eat less. So in this case, maybe the response in the amount that they eat is actually due to them just unconsciously eating less, rather than the appetite suppressant itself. So there are a couple ways, blinding and placebo treatments, so those are some methods that are used to help avoid this Hawthorne effect.

So the main types of bias we have looked at in this tutorial are selection bias, participation bias, publication bias, response bias, and then the Hawthorne effect. So this has been your tutorial on bias. Thanks for watching.

Terms to Know
Bias

The tendency for collected data to differ from what is expected in a systematic way. Biased data can often favor a specific group of those studied.

Hawthorne Effect

People have the tendency to change their behavior when they know they are being monitored.

Participation (Voluntary Response) Bias

Bias that occurs when a sample consists entirely of volunteers. People with strong opinions may be the only ones who volunteer.

Publication Bias

The desire of researchers (and research publications) to only print the most sensational or interesting articles.

Response Bias

Bias that occurs when a respondent tells the interviewer "what they want to hear" or lies due to the sensitive nature of the question.

Selection Bias

Selecting a sample in such a way that certain subsets of the population are systematically excluded.