Saturation in Qualitative Research: What It Means and What It Does Not Mean

Because qualitative research does not use the term in one universal way.

Malterud and colleagues note that saturation is closely tied to methodology and often applied inconsistently.

Grounded theory has a specific tradition of theoretical saturation linked to category development and theoretical sampling.

Other studies use saturation more broadly to mean:

additional interviews are no longer adding new information.

Those are not always the same standard.

Hennink, Kaiser and Marconi provide one of the most useful empirical distinctions.

You have identified the range of relevant issues.

In their study, this occurred around nine interviews.

Think:

“We have heard the main topics.”

You have developed a richer, more nuanced understanding of those issues.

In their dataset, this required approximately 16-24 interviews.

Think:

“We understand the dimensions, variation and meaning of the topics.”

This distinction explains why:

“No new codes emerged”

does not necessarily mean:

“Further interviews could teach us nothing important.”

In grounded theory, theoretical saturation has a more specific meaning.

It relates to whether additional data are contributing new properties, dimensions or relationships to the developing theoretical categories.

This is connected to:

• theoretical sampling;
• ongoing category development;
• the emerging theory.

Do not write:

“Theoretical saturation was achieved”

in a generic thematic analysis simply because you finished interviews.

If you did not conduct theoretical sampling or build theoretical categories, the term may not fit.

Sometimes saturation is used as a stopping criterion.

But it is not the only way to think about sample adequacy.

Malterud's information-power model offers an alternative framework based on:

• aim;
• sample specificity;
• theory;
• dialogue quality;
• analysis strategy.

Some qualitative approaches do not make saturation central at all.

Before writing:

“Interviews continued until saturation”

ask whether the methodology you chose actually uses saturation in that way.

One practical approach is to maintain a saturation log during data collection and analysis.

Example:

This is not a universal saturation formula.

It is documentation.

The important point is to make the decision process visible.

If qualitative data collection is iterative, emerging findings may suggest:

• another participant category is missing;
• a subgroup is underrepresented;
• one process needs deeper probing;
• contradictory cases should be sought;
• enough depth has been reached for the study aim.

A saturation claim that has no relationship to sampling decisions is less informative.

Saturation in focus-group research is not simply:

number of focus groups × eight participants.

The unit of data generation includes group interaction.

You may need variation across:

• age;
• sex;
• setting;
• participant type;
• group composition.

Whether additional groups add meaning can differ from individual interview saturation.

Do not take an interview sample-size paper and mechanically convert it into a focus-group rule.

No.

That standard would be impossible.

Human experience is open-ended.

A new participant could almost always say something unique.

The practical judgement is whether new data are adding information sufficiently important to the study's analytic purpose.

That makes saturation dependent on the research question.

You can, but define it.

Better than:

Data saturation was reached.

Write:

After interview 16, subsequent interviews did not identify new major codes relevant to the primary research question, and interviews 17-19 added limited new dimensions to the existing themes. The team therefore judged the dataset adequate for the planned thematic analysis.

Or, if your concern is meaning:

Recruitment continued after the main thematic categories had been identified because additional interviews were still adding important variation and depth to the interpretation of [theme]. By interviews 20-22, further data contributed little additional nuance to the principal themes.

The wording should reflect what actually happened.

Suppose 17 participants describe services as judgemental.

Participant 18 describes the clinic as unusually supportive.

That interview may not produce a new high-level theme.

It may still be analytically important because it reveals:

• conditions under which judgement is reduced;
• variation across providers;
• a different mechanism;
• limits of the dominant interpretation.

A mature saturation process does not treat contradictory data as noise.

Before claiming saturation, answer:

1. What definition of saturation fits my methodology?
2. What exactly was monitored?
3. Did analysis occur during data collection?
4. What did the final interviews add?
5. Were key participant groups adequately represented?
6. Did contradictory cases alter the interpretation?
7. What evidence supported the decision to stop?

If the answer is:

“We had reached the number approved in the protocol”

that is a recruitment endpoint.

Not necessarily saturation.