How do you determine when you've reached thematic saturation?
This tests operationalizing saturation beyond gut feel. A strong answer cites Guest et al.'s Base Size, Run Length, and New Information Threshold, and flags consecutive interviews adding no new high-level themes.
WHAT THIS TESTS: Your ability to move beyond vague notions of knowing saturation when you see it and instead apply a reproducible, auditable framework for determining when additional qualitative data will not yield new high-level themes. Interviewers want to see that you understand the difference between sample-size planning before fieldwork and empirically assessing adequacy during or after data collection.
A GOOD ANSWER COVERS: First, an explicit operational definition of saturation framed as a measurable event rather than a feeling. Second, reference to a validated method such as the one proposed by Guest, Namey, and Chen, which centers on three elements: Base Size, meaning a minimum number of initial interviews to establish the thematic landscape; Run Length, meaning a defined streak of consecutive interviews in which no new high-level themes appear; and New Information Threshold, meaning a clear rule for what counts as a new theme versus a variant of an existing one. Third, the specific signal that indicates saturation: a run of interviews that only produce additional examples, nuances, or sub-codes under already established high-level themes, with no emergent codes that restructure the framework. Fourth, a validation step, such as bootstrapping against an existing coded dataset or inter-coder checks, to confirm that the observed run length is not a chance gap.
COMMON WRONG ANSWERS: Claiming you simply feel saturated or have developed an intuitive sense after years of practice. Using fixed rules of thumb, such as always conducting fifteen interviews, without empirically checking whether new themes continue to emerge. Treating every new quote, anecdote, or minor variation as evidence of unsaturation, which confuses high-level thematic novelty with data richness. Failing to distinguish between design-phase sample size estimation and in-process adequacy assessment.
LIKELY FOLLOW-UPS: How would you adapt this approach if you are studying multiple distinct user segments and saturation might be reached at different rates across subgroups. What would you do if a new high-level theme surfaced after you had already declared saturation and stopped recruiting. How you would report saturation metrics to stakeholders who are accustomed to quantitative statistical power.
ONE CONCRETE EXAMPLE: Imagine you are conducting inductive thematic analysis on in-depth usability interviews. You establish a Base Size of twelve interviews to capture the initial thematic landscape. You then define your New Information Threshold as a code that does not fit into any existing high-level theme and would require creating a new top-level node. After interview fifteen, you notice that interviews thirteen through fifteen added only supporting examples and sub-variants to existing themes. You treat those three interviews as your Run Length. To validate, you use a bootstrapping resampling technique on your already coded dataset and find negligible probability of discovering a new high-level theme beyond this point, giving you confidence to stop recruitment and report saturation.
Read the original → journals.plos.org
Get five bites like this every day.
Tezvyn delivers a daily feed of 60-second tech bites with quizzes to lock in what you learn.