Saturation in qualitative research

And we're back to talking about qualitative research! Sorry not sorry.

What is saturation?

Saturation is commonly used as a determinant of sample size in qualitative research. So what is saturation? It is usually thought of as the point where there is no new codes, or no new data for development of themes, or where the complete range of theoretical constructs is fully represented by the data. There are four main models of saturation, and sometimes more than one model may be combined:

1. Theoretical saturation- Uses the development of categories and theory as criteria for additional data collection. Sampling is guided by the necessary similarities and contrasts required by the emerging theory, in a process known as theoretical sampling.
2. Inductive thematic saturation- Focuses on the identification of new codes or themes and on the number of these themes, rather than their completeness.
3. A priori thematic saturation- Collects data in order to exemplify a theory, rather than to refine or develop theory. Points to the idea of pre-determined theoretical categories.
4. Data saturation- Stops collecting when nothing new is found. Relies on simply identifying redundancy, without referencing theory at all.

Where and why should we seek saturation?

Saturation may have different relevance and meaning in different types of research. In a more deductive study (using theory to analyse data), the concept of saturation may help you to determine how well the pre-determined codes are represented. In a more inductive study (using data to generate a theory), the concept of saturation might involve how well the new codes and themes are identified. On the other hand, saturation might not be a very relevant concept in some other types of research, such as narrative research.

When and how can we achieve saturation?

The question of "when is saturation achieved?" generally depends on the model of saturation chosen. It is important not to decide on saturation too early on, as this may produce superficial results.

Alternatively, instead of asking, "When is saturation achieved?" or "Has saturation occurred?" (i.e. viewing saturation as a process), it may be more appropriate to ask, "How much is enough?" (i.e. viewing saturation as an endpoint). The problem with the "endpoint" model is that looking for a saturation "point" often requires making predictive claims on data that you don't have yet, as you're basically assuming that all of the other potential data out there is just going to be a rehash of what you already have. In fact, some studies consciously sample beyond the saturation "endpoint," just in case. Using a "process" model requires using ongoing judgement to figure out the point of diminishing returns with regard to sampling and discovery of new insights.

References

Saunders, B, Sim, J, Kingstone, T, Baker, S, Waterfield, J, Bartlam, B, Burroughs, H, Jinks, C 2018, 'Saturation in qualitative research: exploring its conceptualization and operationalization', Quality & Quantity, vol. 52, no. 4, pp. 1893-1907.