What Is A Biased Sample? (With Definition And Examples)

By Indeed Editorial Team

Published 27 July 2022

The Indeed Editorial Team comprises a diverse and talented team of writers, researchers and subject matter experts equipped with Indeed's data and insights to deliver useful tips to help guide your career journey.

There are many ways of collecting samples when conducting research, but it is important to use a sample that represents all sections of the population equally. Sampling bias occurs when some members of the population are under-represented or left out, which can influence the study results. Understanding what a biased sample is and how to avoid it can help you conduct valid research. In this article, we answer, "What is a biased sample?", discuss its different types, explain how to avoid sampling bias and share some examples of biased samples.

Read more: What Is A Research Associate? (And How To Become One)

What Is A Biased Sample?

Knowing the answer to "What is a biased sample?" can help you conduct accurate, reliable and valid research. A biased sample does not accurately represent all elements of the population. It usually occurs when a researcher uses a sampling method that does not favour certain sections of the population. This results in more generalised outputs instead of niche findings that resonate with the sample.

For example, if a marketing team wants to understand what products customers like from their new collection, and they end up asking this question only to their in-store customers. Online customers may have a different opinion than offline shoppers. This makes the result biased towards in-store customers.

Types Of Biased Samples

Here are some common types of biased sampling:

Self-selection bias

It is strictly voluntary to participate in a research or study. An individual may or may not choose to be a part of it. It is possible that those who decide to participate in a survey may have similar views. This may lead to a bias as the views of the non-participants are missing from the data. As a result, the sample can be incomplete and one-sided.

Under-coverage bias

Under-coverage bias occurs when the population that is affected by the research is not a part of the sample. It can also happen because of limited access provided to research participants. For example, if a telephone company wants to research the quality of the Internet, but their customers require a good Internet connection to fill the survey, it will leave out participants who do not have it.

Survivorship bias

Survivorship bias occurs when the researcher chooses participants who they can describe as successful. This can lead to a biased sample, as the researcher may focus only on the winners and disregard the losers, often due to their lack of visibility. For example, if a school only allows students who score above 80% to participate in a study, the results can be impacted by only those students, while those who score under 80% can be excluded.

Exclusion bias

Exclusion bias occurs when a certain sample group is left out while conducting research. It also excludes people who have recently left or joined the sample group. For example, for a survey measuring customer loyalty, a company may leave out customers who recently joined or left their membership programme.

Non-response bias

Sometimes certain people may refuse to participate in a study or simply do not have access to the survey. This can result in a biased sample. As their inputs may be crucial for the research to be reliable, the end result is incomplete without their participation.

Pre-screening bias

Like survivorship bias, pre-screening bias occurs when researchers decide to interview candidates before allowing them to participate in the research. They may do so to influence the results in the direction of their preference. This leads to a biased sample, as the participants with specific characteristics may not pass the interview.

Read more: What Are Product Research Tools? (With Best Practices)

How To Avoid Sampling Bias

Here are a few steps you can take to avoid sampling bias:

1. Set clear perimeters

The first step towards an ideal sampling is to identify the exact perimeters for your study. This process starts with what hypothesis you want to test, followed by what resources you require to test it. List down the independent and dependent variables you are studying. These methods can help you determine the best method for choosing a sample population.

2. Identify your target population

An effective way of avoiding sampling bias is to be clear about your target audience. It is essential to prevent sampling convenience at this point, as it may affect your research. For example, if you are studying how customers react to discounts, collecting a sample of both online and offline shoppers is crucial. This way, different types of shoppers are a part of your sample.

3. Reach the target population effectively

Once you are clear about your target population, the next step is to reach them. You may opt for one of many ways to invite them to fill out your survey. For instance, you may reach out to them personally or ask your peers to refer you to them. Another way is to send out cold emails, calls or text messages.

4. Strengthen the review process

As a researcher, you may draft, edit and study the questionnaire several times to improve its clarity. It can also be helpful to have a colleague or mentor review the questions. This can help you identify any biases that you may have missed. Throughout the study, it is essential to conduct periodic external reviews to ensure that the sample does not become biased.

5. Use oversampling

Oversampling is a method where a researcher selects participants from larger groups, so they make a larger share of a sample instead of the population. Researchers use this when members of specified groups are underrepresented. They remove sampling bias by weighing the oversampled groups to their actual share of the population.

Probability Vs. Non-Probability Biased Samples

Bias can also be present in the probability or non-probability samples. Here are some examples of the same:

Sampling bias in probability-based samples

In probability sampling, all members of the population have an equal chance of being selected. The researcher randomly picks the population members for sampling, reducing the risk of sampling bias. Nonetheless, sampling bias may still occur if the members of the population do not match the test perimeters.

Example: You want to study the stress levels in school students appearing for exams using random sampling. To do so, you assign a number to every student in the school and use a random number generator to select 100 numbers. Even though you choose the members randomly, there is a chance that you leave out students with higher stress levels who would have been interested in the study.

Sampling bias in non-probability-based samples

In non-probability sampling, the researcher uses convenience sampling to pick population members. This leaves a higher chance of sampling bias, as the researcher may leave some members of the population because of unawareness or reachability issues.

Example: You want to research sustainability trends amongst students at your college. For convenience, you send out questionnaires to undergraduate students in the fashion department. They all fill out the questionnaire in exchange for grades. But this method can leave other students in your college who may be interested in sustainability.

Read more: Essential Research Skills (With Benefits And How To Improve)

Examples Of Biased Samples

Biased samples may occur in different fields of research. Here are some examples of the same:

Sampling bias in psychology

Sampling bias can occur in psychological research and clinical trials. It is more likely when the samples get collected through self-selection or convenience sampling. This can affect the validity of the results. You can reduce these errors by collecting data from diverse people. You may set your criteria and try to collect samples as close to your target population as possible.

Example: Consider a study that aims to understand the mental health of retired military personnel. To gather the data, researchers ask individuals to volunteer for the study. This method can lead to self-selection bias, as individuals with good mental health may volunteer, leaving out those with problems. So, the results may be inaccurate in representing the mental health of the community of retired military personnel.

Sampling bias in surveys

Your survey design can cause sampling bias if it favours specific individuals or groups of individuals. Even the language of your survey can exclude a considerable portion of the population. For example, if you want to survey people who cannot read, you require volunteers who would talk to the subjects and fill the responses on their behalf.

Example: Sampling bias in surveys occurs because of multiple factors. For instance, a study to understand the use of government schemes among retired people can be biased if it excludes people not enrolled in the scheme.

Explore more articles