Sampling

Amanda Williams; Emma Duke

6 Sampling

Learning Objectives for Chapter

Differentiate between populations and samples.
Define non-probability sampling.
Identify instances in which a researcher might choose a non-probability sampling technique.
Evaluate different types of non-probability samples.
Differentiate between probability sampling and non-probability sampling.
Define generalisability and describe how it is achieved in probability samples.
Describe the various types of probability samples available.
Identify questions to ask about samples when reading the results of research.

Introduction

Sampling is a fundamental aspect of research that involves selecting a subset of individuals or elements from a larger population to study. It provides researchers with valuable insights into the characteristics and behaviours of a population without having to examine every single member. In sampling, two main approaches emerge: non-probability sampling and probability sampling. Each approach has its strengths and limitations, offering researchers a range of method options based on their research goals, resources, and the level of representativeness they seek.

This chapter delves into the world of sampling techniques, exploring the distinctions between non-probability and probability methods. We will uncover the underlying principles of each technique, discuss scenarios in which they might be employed within the realm of communication studies and highlight the considerations researchers must keep in mind when selecting an appropriate sampling strategy.

Moreover, we will provide a set of key questions to ask when evaluating a sample within research studies. We hope that by understanding the nuances of sampling techniques and learning how to analyse the representativeness and potential biases of a sample, you will be better equipped to navigate the complex landscape of research literature and draw well-informed conclusions.

The Difference Between Population and Sample

A population describes the cluster of events, things, people, or other phenomena a researcher is interested in. Populations look at the ‘who’ or ‘what’ in question. They can be as broad as ‘Americans’ but will likely be a little more specific –perhaps the population of interest is Americans over 18, for example. Because it would be impossible to interview every Canadian over at the age of 18, a researcher would gather a sample. A sample is the cluster of people or events from or about which you will gather data. In this case, a sample might be 300 individuals who live in Canada and are 18 and older.

Sampling is the process of selecting observations that will be analysed for research purposes. A researcher will choose a method of sampling to either: a) make sweeping conclusions about the population of interest, with a fair amount of confidence, or b) make theoretical contributions about the larger population.

Because the goals of qualitative and quantitative researchers differ (due to their epistemological and ontological commitments as we explored in Chapter 2), so too do their sampling methods.

Sampling in Qualitative Research

Qualitative researchers typically make sampling choices that enable them to deepen understanding of whatever phenomenon it is that they are studying. In this section we’ll examine the strategies as well as the various types of samples that qualitative researchers are most likely to use in their work.

Non-Probability Sampling

Non-probability sampling refers to sampling techniques for which a person’s (or event’s or researcher’s focus) likelihood of being selected for membership in the sample is unknown.

Because we do not know the likelihood of selection, we do not know with non-probability samples whether a sample represents a larger population or not. However, the goal of non-probability samples is not to represent the larger population.

So, when are non-probability samples ideal? A non-probability sample might be used when designing a research project. If we are conducting survey research, we may want to administer our survey to a few people who seem to resemble the folks we are interested in studying in order to help work out kinks in the survey. This can be a quick way to gather some initial data before diving into a more extensive study.

Researchers also use non-probability samples in full-blown research projects. These projects are usually qualitative in nature, where the researcher is trying to achieve in-depth, rather than general understanding. Evaluation researchers whose aim is to describe some particular small group might use non-probability sampling techniques, for example. Researchers interested in contributing to our theoretical understanding of some phenomenon by contributing to social theories – expanding on them, modifying them, or poking holes in their propositions, might also collect data from non-probability samples.

Types of Non-Probability Samples

There are several types of non-probability samples that researchers use. These include purposive samples, snowball samples, quota samples, and convenience samples. While some of these strategies may be used by quantitative researchers from time to time, they are more typically employed in qualitative research.

A purposive sample is when a researcher begins with very specific perspectives he wishes to examine, and then seeks out research participants who cover that range of perspectives.

In a communication studies research project, a researcher might use purposive sampling to study the social media behaviours of influential celebrities. They could specifically select participants who have a significant following on platforms like Instagram and Twitter, focusing on individuals known for promoting social or political causes. By targeting this specific group, the researcher aims to gain insights into how these celebrities use their online presence to communicate messages and engage with their followers, contributing to a deeper understanding of the role of social media in influencing public opinion and social change.

Snowball sampling is a research method used in qualitative studies where initial participants, often known to the researcher, are chosen purposively. These participants then assist in identifying and recruiting additional participants for the study. As the research progresses, the sample “snowballs” as each newly recruited participant suggests or introduces the researcher to others who fit the study’s criteria. This approach is particularly useful when studying hard-to-reach, stigmatised, or tightly knit groups, as it leverages existing connections to access and expand the participant pool.

In a communication research study, let’s say a researcher is interested in exploring the communication patterns within a tight-knit online gaming community. Due to the specialised nature of this group and the challenge of accessing its members, the researcher starts by interviewing a couple of active participants they know are involved. These initial interviewees then introduce the researcher to other members of the community who might be willing to participate in the study.

For instance, the researcher interviews Player A, who is well-known in the gaming community. During the interview, Player A mentions Player B and Player C as active and influential members of the same community. With Player A’s endorsement, the researcher also contacts Player B and Player C for interviews. As the study progresses, Player B and Player C suggest additional participants, and the process continues, gradually expanding the sample through referrals from existing participants.

This snowball sampling approach allows the researcher to gain access to a group that might be challenging to reach through traditional methods. It also builds a sense of trust and rapport among participants, as they are connected through shared affiliations and can vouch for the researcher’s intentions and credibility within the community.

This method is also known as chain referral sampling, one research participant refers another, and that person refers another, and that person refers another—thus, a chain of potential participants is identified.

Quota sampling is another non-probability method. Unlike the other methods, quota sampling is employed by quantitative researchers as well as qualitative. In a quota sample, a researcher identifies subgroups within a population of interest and then selects some predetermined number of elements from within each subgroup.

In a communication research study focusing on consumer preferences for television content, a researcher might use a quota sampling method to ensure a diverse range of participants based on specific demographic characteristics. The goal is to capture a representative sample that reflects the population’s diversity while maintaining control over specific demographic proportions. For example, the researcher aims to interview 100 participants from different age groups (18-24, 25-40, 41-60, 61+), gender identities, and income levels. Within each age group, the researcher sets a quota for gender and income distribution. They start by selecting participants who fit the criteria of the first age group, ensuring a mix of gender identities and income levels. Once the quota for that group is met, the researcher moves on to the next age group and repeat the process.

By using a quota sample, the researcher ensures a balanced representation of participants across age, gender, and income categories, while still maintaining a manageable sample size. This approach helps gather insights into how different demographic groups perceive and engage with television content, enhancing the study’s validity and applicability.

Convenience sampling, also known as haphazard sampling, is a method used by both quantitative and qualitative researchers. It involves collecting data from individuals or elements that are readily accessible to the researcher. This approach is particularly useful in exploratory research and is often employed by media professionals who need quick access to individuals from their population of interest. For instance, brief street interviews featured on the news are a common example of haphazard sampling.

In a communication study investigating young adults’ experiences with social media and mental health, a researcher might employ convenience sampling by recruiting participants from their university campus. They would post flyers in the student union building, send emails to student organisations, and utilise personal social media accounts to reach out to friends and acquaintances within the 18-25 age range. The researcher would then conduct in-depth interviews with 20 responding students, exploring their perceptions and experiences with social media and its impact on their mental wellbeing.

Although the convenience sample may not be representative of the larger population, it can provide valuable insights into how social media is integrated into young adults’ daily lives and its potential effects on their mental health. This approach allows for an in-depth exploration of the research question, but the findings may not be generalizable beyond this specific context.

The Value of Non-Probability Samples

All non-probability samples are non-generalisable. Non- generalisable samples can be valuable in the following ways:

Non-generalisable samples can offer detailed and nuanced insights into specific cases or contexts. Researchers can deeply explore unique circumstances and gain a comprehensive understanding of complex issues.
Samples focusing on specific variables or concepts allow researchers to test and refine theoretical frameworks. These samples may provide evidence that supports or challenges existing theories, leading to the development of new ideas.
In exploratory or preliminary research stages, non- generalisable samples can help researchers identify trends, patterns, and potential areas of interest for further investigation.
Non-generalisable samples can provide context and depth to research findings, making them more relevant and applicable to specific situations.
Qualitative research often relies on non-generalisable samples to uncover rich qualitative data, such as personal experiences, motivations, and perceptions.
Non-generalisable samples can be valuable in comparative studies that aim to contrast different cases, contexts, or groups to identify similarities and differences.

While generalisability enhances the external validity of research findings, there are scenarios where a non-generalizable sample limitations do not necessarily diminish its usefulness. Researchers should carefully consider their research goals, methodology, and the specific insights they aim to gain when deciding on the type of sample to use.

Table 6:1

Summary of Non-Probability Samples

Sample type	Description
Purposive	Researcher seeks out elements that meet specific criteria.
Snowball	Researcher relies on participant referrals to recruit new participants.
Quota	Researcher selects cases from within several different subgroups.
Convenience	Researcher gathers data from whatever cases happen to be convenient.

Sampling in Quantitative Research

While there are instances when quantitative researchers rely on non-probability samples (like when doing exploratory research), they tend to rely on probability sampling techniques. The goals and techniques associated with probability samples differ from those of non-probability samples. We’ll explore those unique goals and techniques in this section.

Probability Sampling

Probability sampling refers to sampling techniques for which a person’s (or event’s) likelihood of being selected for membership in the sample is both known and generalisable.

In most cases, researchers who use probability sampling methods are aiming to identify a representative sample from which to collect data. A representative sample is one that resembles the population from which it was drawn in all the ways that are important for the research being conducted.

If your population varies in some way that is important to your study, your sample should contain the same sorts of variation.

Obtaining a representative sample is important in probability sampling because a key goal of studies that rely on probability samples are generalisability. Generalisability is the idea that a study’s results will tell us something about a group larger than the sample from which the findings were generated. A core principle of generalisability is that all elements in a researcher’s population have an equal chance of being selected for the study. This is called random selection.

If a researcher uses random selection techniques, they will be able to estimate how closely the sample represents the larger population from which it was drawn by estimating the sampling error. Sampling error is a statistical calculation of the difference between results from a sample and the actual parameters of a population. Parameters are the actual characteristics of a population on any given variable; determined by measuring all elements in a population (as opposed to measuring elements from a sample).

Types of Probability Samples

Researchers may use a variety of probability samples. These include simple random, systematic, stratified and cluster samples.

Simple random samples are the most basic type of probability sample. To draw a simple random sample, a researcher begins with a list of every member of their population of interest, numbers each element sequentially, and then randomly selects elements from in which they gather data. The list with all the elements in the population is called a sampling frame.

In a communication research study exploring public attitudes toward political advertising, a researcher might employ a simple random sampling method to ensure that every eligible member of the population has an equal chance of being included in the study.

To achieve this, the researcher would obtain a comprehensive list of registered voters in a specific city. From this list, they would use a random number generator to select a sample size of 300 individuals. These selected individuals would then be contacted and invited to participate in a survey about their perceptions of political advertisements.

By using a simple random sample, the researcher ensures that each registered voter has an equal probability of being chosen for the study, reducing bias, and increasing the likelihood that the sample accurately represents the broader population’s views on political advertising. This approach allows the researcher to make valid inferences about the attitudes of registered voters in the city without disproportionately favouring any particular subgroup.

Drawing a simple random sample can be quite tedious. Systematic sampling techniques are somewhat less tedious but offer the benefits of a random sample. As with simple random samples, you must be able to produce a list of every one of your population elements. Once you’ve done that, to draw a systematic sample you’d simply select every kth element on your list. k is your selection interval or the distance between the elements you select for inclusion in your study. To find your selection interval, divide the total number of population elements by your desired sample size. Suppose you want to interview 25 fraternity members on your campus, and there are 100 men on campus who are members of fraternities. In this case, your selection interval, or k, is 4. To determine where on your list of population elements to begin selecting the names of the 25 men you will interview, select a random number between 1 and k and begin there. If we randomly select 3 as our starting point, we will begin by selecting the third fraternity member on the list and then select every fourth member from there.

In a communication research study investigating media consumption habits, a researcher might opt for a systematic sampling method to gather data from a diverse range of television viewers.

The researcher selects a list of television channels that cover various genres, such as news, entertainment, sports, and documentaries. To create a systematic sample, the researcher surveys every tenth viewer on each channel who watches a particular program during a specified time slot. For instance, if the researcher chooses the news segment at 8:00 PM, they will survey the tenth viewer tuning into that specific program on each selected channel. This pattern continues across different channels and time slots. The researcher records the responses from these systematically selected viewers regarding their preferences, reasons for watching, and perceptions of media content. By using a systematic sample, the researcher ensures a structured and evenly distributed approach to selecting participants, avoiding potential biases that might arise from convenience or judgmental sampling methods. This method allows the researcher to collect data from a wide range of viewers, offering insights into media consumption patterns across different genres and channels within the population of interest.

If your sampling frame has any pattern to it, systematic sampling should not be employed, as this could bring bias into your sample.

Let’s consider an example to illustrate this point. Imagine you are conducting a study on the preferences of customers at a local coffee shop. Your sampling frame is a list of customers who visit the coffee shop daily. However, upon closer examination, you notice that the list is organised based on the days of the week, with Monday’s customers listed first, followed by Tuesday’s customers, and so on. If you were to employ systematic sampling in this scenario, selecting every fifth customer from the list, you might inadvertently introduce bias into your sample. Since the list is organised by days of the week, systematic sampling could lead to a disproportionate representation of customers who visit on specific days. For instance, if your systematic selection falls on a Friday, you might end up including more weekend visitors in your sample, potentially skewing the results, and failing to capture the full diversity of customer preferences across all days of the week.

In cases where the sampling frame is unbalanced, it would be better to use a stratified sampling technique. This is when a researcher divides a study population into relevant subgroups and then draws a sample from each subgroup at a select interval. For example, in a communication research study that analyses media preferences among different age groups, a researcher might employ a stratified sampling technique to ensure representative data from each age category. First, the researcher divides the target population into distinct strata based on age groups, such as 18-24, 25-40, and 41-60. The researcher randomly selects participants within each stratum using a simple random sampling method. For instance, if the target population consists of 500 individuals, with 100 in each age group, the researcher randomly selects 25 participants from each age group. Once the participants are selected, they are invited to participate in a survey or interview regarding their media consumption habits. The researcher gathers insights about how different age groups engage with various forms of media, such as television, social media, and online news platforms. By using a stratified sampling technique, the researcher ensures that each age group is adequately represented in the study, allowing for more accurate comparisons and analyses of media preferences across different generations. This approach helps avoid potential biases and provides a more comprehensive understanding of how communication habits vary among individuals of different age ranges.

Stratified random samples are also valuable when a subgroup makes up a smaller proportion of the study population you are interested in. For example, if you wanted to include both men and women’s perspectives in a study, but men make up 75% of the population, there is a chance that a simple random or systematic sampling strategy might not yield any female participants. By using stratified sampling, we could ensure that our sample contained the proportion of women that are reflective of the larger population.

A final choice is cluster sampling occurs when a researcher begins by sampling groups of population elements and then selects elements from within those groups. You would use a cluster sample when getting a master sampling frame which would be almost impossible.

For example, if you wanted to study the behaviours of communication professionals in Canada, generating a list of everyone working in the field across the country might be impractical. However, here is how you might conduct a cluster sample. You would begin by categorising media professionals into distinct clusters based on specific criteria. Clusters could be formed based on factors such as geographical location, type of media organisation (e.g., newspapers, television stations, online platforms), or specialisation (e.g., journalism, public relations, broadcasting). Next you would randomly select a representative number of clusters from your identified categories. For instance, you might choose clusters from different cities or regions where media professionals are located. Within each selected cluster, you would further narrow your focus by randomly selecting specific media organisations or outlets. You might select several newspapers, radio stations, and online news platforms from each chosen city. Within each selected media outlet, randomly sample media professionals to participate in your research. You could choose a specific number of journalists, editors, producers, broadcasters, or other relevant roles depending on your goals. You would then contact the selected media professionals within each outlet and invite them to participate in your study. Depending on your research approach, you can conduct surveys, interviews, or other data collection methods to gather insights about their experiences, perspectives, and challenges in the media industry. Using the cluster sampling technique in this context allows you to capture various viewpoints from media professionals across various locations and types of media organisations. It enables you to explore variations in practices, perceptions, and challenges within different clusters of media professionals, contributing to a more comprehensive understanding of the media landscape.

Table 6.2

Summary of Probability Samples

Sample type	Description
Simple random	Researcher randomly selects elements from the sampling frame.
Systematic	Researcher selects every kth element from the sampling frame.
Stratified	Researcher creates subgroups then randomly selects elements from each subgroup.
Cluster	Researcher randomly selects clusters then randomly selects elements from selected clusters.

A Word of Caution: Questions to Ask About Samples

We often come across research results in our reading and discussions, but we might forget to ask important questions about where the people in the research come from and how they were chosen to be part of the study. Sometimes, we get caught up in the exciting findings and overlook the steps taken to do the research.

Now that you’re aware of the various methods used to select participants for research, you can begin to pose crucial questions about the findings you encounter. This will help you be more responsible when you read and use research information.

Who Was Included in the Sample? Understanding who the participants are and their characteristics is crucial. This helps you determine if the sample is representative of the target population and whether the findings can be generalised.
How Was the Sample Selected? Inquire about the sampling method used (e.g., random sampling, convenience sampling) and whether it was appropriate for the research question.
Sample Size: Ask about the size of the sample. A larger sample size generally increases the statistical power and generalizability of the results.
Demographic Information: Gather information about demographic characteristics such as age, gender, ethnicity, socioeconomic status, etc. This helps assess the diversity and representativeness of the sample.
Inclusion and Exclusion Criteria: Understand the criteria used to include or exclude participants from the study. This affects the study’s applicability to specific groups or conditions.
Sampling Frame: Ask about the source of the sample and how it was obtained. A well-defined sampling frame ensures the sample accurately reflects the target population.
Response Rate: Inquire about the proportion of invited participants who actually participated in the study. A low response rate could introduce nonresponse bias.
Attrition Rate: Find out how many participants dropped out of the study over time. High attrition rates can affect the internal and external validity of the results.
Comparison with Desired Population: Compare the sample characteristics with those of the target population. Significant differences may affect the generalizability of the findings.
Validity of Inferences: Assess whether the study’s findings can be appropriately generalised beyond the sample to a larger population.

By asking these key questions, you gain a comprehensive understanding of the research study’s sample, ensuring that the findings are accurate, reliable, and relevant.

Reflection Question

After reading about different sampling techniques, consider a research scenario where you’re investigating people’s preferences for online streaming platforms. Which sampling method would you choose, and why? Discuss the advantages and potential limitations of your chosen sampling technique in gathering data for this particular study. Document your thoughts in a 200–300-word post.

Key Chapter Takeaways

• A population is the group that is the main focus of a researcher’s interest; a sample is the group from whom the researcher collects data.

• Non-probability samples might be used when researchers are conducting exploratory research, by evaluation researchers, or by researchers whose aim is to make some theoretical contribution.

• There are several types of non-probability samples, including purposive samples, snowball samples, quota samples, and convenience samples.

• In probability sampling, the aim is to identify a sample that resembles the population from which it was drawn.

• There are several types of probability samples including simple random samples, systematic samples, stratified samples, and cluster samples.

• The value of a researcher’s findings isn’t solely determined by their generalizability. Samples that facilitate comparisons of theoretically significant concepts or variables can produce insights that enrich our social theories and deepen our understanding of social processes.

• Sometimes researchers may make claims about populations other than those from whom their samples were drawn; other times they may make claims about a population based on a sample that is not representative. As consumers of research, we should be attentive to both possibilities.

Key Terms

Population: The collection of individuals, occurrences, objects, or other phenomena that hold your primary interest— the “who” or “what.”

Sample: The assemblage of people or events from which you’ll gather factual information.

Sampling: The act of choosing observations that warrant examination for research purposes.

Non-probability sampling: Sampling methods wherein the likelihood of an individual’s inclusion in the sample is uncertain.

Purposive sample: A researcher identifies specific viewpoints to investigate and then enlists participants representing this entire spectrum. Also utilised when seeking individuals who meet specific, stringent criteria.

Snowball sampling: Also termed chain referral. A researcher identifies a couple of participants for a study and relies on their help in identifying additional participants. Commonly used in qualitative research.

Quota sampling: A type of non-probability sampling where a researcher identifies subgroups within a population and selects a predetermined number of elements from each subgroup. Employed by both qualitative and quantitative researchers.

Convenience sampling: A researcher collects data from readily accessible individuals or relevant entities. Valuable in exploratory research and applied by both qualitative and quantitative researchers.

Probability sampling: Sampling techniques where the likelihood of an individual’s (or event’s) inclusion in the sample is known and random.

Representative sample: Mirrors the relevant attributes of the population for the conducted research.

Generalisability: The concept that a study’s findings extend to a larger group than the sampled population.

Random selection: A fundamental tenet of probability sampling. “Random” in sampling refers to a selection process where every individual or element in the population has an equal and independent chance of being chosen for inclusion in the sample. In other words, randomness ensures that each member of the population has a fair opportunity to be selected without any predictable pattern or bias.

Sampling error: A statistical calculation of the variance between sample outcomes and the actual parameters of a population.

Parameters: The authentic traits of a population concerning any specific variable; deduced from assessing all elements in the population rather than just the sample.

Simple random samples: The most elementary form of probability sampling. Researchers start with a comprehensive list of all individuals in their population of interest, sequentially assign numbers, and then randomly choose elements for data collection.

Sampling frame: A list of all elements within a population.

Systematic sampling: Researchers list all population members, assign sequential numbers, and then choose every kth element on the list.

Selection interval: Derived by dividing the total population elements by the desired sample size.

Stratified sampling: Researchers divide the study population into relevant subgroups and then draw samples from each subgroup.

Cluster sampling: Researchers initiate sampling by selecting groups (or clusters) of population elements, followed by selecting elements within these clusters.

License

Icon for the Creative Commons Attribution-NonCommercial 4.0 International License

Navigating Communication Research: A Practical Guide for Media Practitioners Copyright © 2025 by Amanda Williams and Emma Duke is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License, except where otherwise noted.