What is theoretical sampling in grounded theory and other qualitative research?
Today, we’re going to dive into this question by exploring the origin of this approach and distinguishing theoretical sampling from other types of sampling.
By the end of this video, you’ll fully understand the tradition of the term, why theoretical sampling is different, and, of course, how you can apply it in your own empirical work.
Grounded Theory (Background)
To grasp what we mean by theoretical sampling, we need to go back to the origin of the Grounded Theory methodology.
In the 1960s, sociologists Barney Glaser and Anselm Strauss developed the Grounded Theory approach together. Their aim was to counter the prevailing quantitative paradigm and its deductive logic with a structured method for inductive theory building based on qualitative data.
The goal of Grounded Theory is not to test predefined hypotheses and thereby review or refine existing theories. Instead, its main task is generating new theories based on empirical data.
What is Sampling?
Next, we need to understand what sampling involves. The term refers to the selection of a sample.
A sample is a “selection of people or objects that represents a larger population and provides information about it” (Statista, 2020). Samples play a crucial role in empirical social research as they provide access to the data to be analyzed for a research project.
Theoretical insights are drawn from the results based on the investigation of the sample. These insights are generally intended to be valid beyond the scope of the sample itself. That’s why choosing the right sample is so important.
When writing the methodology section of your academic work, you should always make a strong case for how your sample is composed and why this composition is advantageous for your research goal.
Sampling in Quantitative Research
The Statista definition I just mentioned is influenced by a core principle of quantitative research: the generalizability of statistical relationships from a small sample to a larger group of people or objects.
Let’s say 100 kindergarten teachers fill out a survey, and the results are analyzed. These results are often interpreted in a way that makes statements about all kindergarten teachers represented by the sample.
In quantitative research designs, we can broadly distinguish between random samples and non-probabilistic samples. An ideal random sample consists of a group randomly selected from all persons or objects belonging to the total population.
Implementing this is challenging, as you likely cannot access all kindergarten teachers in one country or the world. Therefore, systematic or arbitrary selection methods also exist, where you might include individuals or objects in the sample that you simply have access to.
Sampling in Qualitative Research
In qualitative research, we need different sampling techniques. Here, randomness is not crucial, but rather the researcher’s judgement.
In “Purposeful Sampling,” cases or individuals are selected who, in the researcher’s view, offer a particularly high degree of information richness in relation to the research subject.
In “Snowball Sampling,” an initial case or expert is identified. Based on the knowledge or contacts of this individual, the researcher then gains access to further interesting cases and experts.
This approach can be helpful because the researcher alone might never have noticed these cases or gained access without a facilitator.
What is Theoretical Sampling?
The sampling methods mentioned so far have one thing in common: the sampling occurs BEFORE data analysis.
And that brings us back to Grounded Theory and theoretical sampling. For Grounded Theory to function, data analysis and sampling must work closely together.
Round 1
Since there is no theory at the beginning of the process, you start with a typical Purposeful Sampling and collect data from an organization or individuals based on the most important criteria for you.
Then, you perform typical steps of the Grounded Theory approach. I won’t go into these steps here – please refer to other tutorials on my channel.
After performing open, selective, and theoretical coding according to Glaser or open, axial, and selective coding according to Strauss, you have identified one or more central theoretical concepts. You may already suspect connections between them or have identified subthemes.
The fact is, your theoretical idea is still in its infancy. To solidify it, you need new data.
Round 2 (Theoretical Sampling)
This is where theoretical sampling comes into play. This time, you make your selection deliberately, based on the theory you have developed at this point.
What does that mean exactly?
Let’s say you’re developing a theory that explains the factors influencing the identity formation of employees in the context of working-from-home.
After your initial interviews with employees, you might have found that the characteristics of their workplace technology are central to their identity-building.
However, you don’t know exactly what about the use of technology is so crucial for identity formation. Could it be the type of hardware, consisting of laptops and smartphones? Or the software tools? Or how they are used?
To learn more, you now select new individuals who have extensive knowledge in this particular area. This could, for example, be members of the IT department of the company. You could also interview the same individuals again, but this time ask targeted questions about the specific theoretical connection you want to better understand.
After the second round, your mini-theory may already be taking shape. But there’s still something you don’t know:
(For example) Why must employees work with technology that is outdated and, from the company’s perspective, actually has a negative impact on their identity formation?
To find the answer, there’s no way around speaking with decision-makers. To complete your theory, you finally interview employees in management positions.
Round 3?
Now your mini-theory looks quite solid. But have you overlooked something? After speaking again with two employees, they couldn’t tell you anything new. Your theory seems accurate.
This is your cue to stop data collection.
Theoretical Sampling according to Strauss and Corbin (1998)
Strauss and Corbin further specified theoretical sampling in their seminal book. They distinguish between four stages:
- Open Sampling
- Relational Sampling
- Variational Sampling
- Discriminate Sampling
These stages provide more structure and define individual steps, which can be particularly helpful at the beginning.
Note that the recommendations by Strauss and Corbin work well only with the coding methods they also propose (open, axial, and selective).
After Barney Glaser and Anselm Strauss had a bit of an argument, two interpretations of Grounded Theory developed: one by Glaser and the other by Strauss. Make sure you understand the differences and align your own work with one of these interpretations.
If you want to learn more about the dispute between the two and the differences between Glaserian and Straussian Grounded Theory, you can read it here.