How many people do we have to interview?

You've already asked this question, right?        

One of the techniques I like to use in qualitative methodology is theoretical saturation. In this approach, we start from pre-established categories - the main ones you want to discover - and those that emerge from the research and are considered relevant.

As interviews (or other techniques, such as focus groups, for example) are conducted, the data is added to their respective categories. Therefore, in this technique, it is essential that data analysis is carried out simultaneously with the data collection process and not at a later stage.

Here, I give an example of hypothetical research that seeks to understand the motivations of people who seek health services as their first resource and the motivations of people who self-medicate as their first resource.

Some initial categories are hypothetically identified that will serve as a basis for some questions:

• Difficulty scheduling an appointment
• The habit of drinking tea to cure pain
• Trust in the family doctor
• Reliance on traditional methods

It is important to emphasize that these hypothetical categories must be identified a priori based on a relevant theoretical framework (deductive process). Other categories that emerge from the data (inductive process) can later be associated with different theories or serve to develop new theories.

➡️ The process is based on "filling in" the categories as the research is carried out. Saturation occurs when interviews no longer provide new information beyond that already identified. New interviews would add a few elements to discuss what has already been obtained.

 A step by step would be:

1. Identify themes for each pre-category or new categories;
2. Identify in which interview the themes of each category appear;
3. Verify theoretical saturation for each category.

Ex:

💡It is important to note that one category may be saturated in the third interview, while another needs 6 or 7 interviews, for example. Or, even if interviews are continued, a category that appeared once may no longer be mentioned. This does not mean that it should not be considered but analyzed carefully to understand the reasons behind low representation.

In the example above, the first topic (difficulty in scheduling an appointment) is saturated based on the analysis of the data already obtained if the answers are repetitive. If there are any doubts left or the researcher considers that they should explore more answers to be able to analyze more details, for example, they do not necessarily need to consider them saturated just because they appeared in all the interviews.

💡What will dictate the rules is the data quality of the research question rather than the quantity.

Making it clear which technique is used to define a research sample attributes scientific validity and demonstrates methodological rigour. In a final presentation, it is highly recommended to at least go through the chosen technique to define the sample. This will bring confidence to those who are interested in the search results.

🤓 Here is another of the contributions of social scientists and their training with scientific solidity for application in user research.

Tell me if I helped you with anything?

Zaíra Bosco

References