Methodology · Chapter 3.0

Grounded theory

Charmaz's constructivist grounded theory — from open coding to a theory that's defensible, including the Glaser–Strauss–Charmaz lineage you need to know to pick the right tradition.

10 min read · Updated May 2026

In this chapter

The promise (1967)
Three traditions, one name
Coding: initial → focused → theoretical
Memo-writing as analysis
Theoretical sampling and constant comparison
Theoretical saturation
When grounded theory fits
Further reading

The promise (1967)

Glaser and Strauss’s The Discovery of Grounded Theory (1967) made a strong claim: theories about social life can be developed systematically from qualitative data, without first deriving hypotheses from an existing framework. The argument was situated against mid-century American sociology’s tendency to test grand theory with quantitative survey work; grounded theory was a rebuttal that put close reading of qualitative data at the centre of theory construction.

The promise has aged unevenly. The methodological apparatus — iterative coding, constant comparison, memo-writing, theoretical sampling, saturation — is widely adopted, often outside the full grounded-theory paradigm. The epistemological claim — that theory emerges from data if the method is followed properly — has been substantially revised by every major contemporary proponent, including Charmaz, who argues we co-construct rather than discover.

Three traditions, one name

“Grounded theory” refers to three substantially different methods, and a methods section that doesn’t name which one is undefendable.

Classical (Glaserian) grounded theory

The original 1967 method as Glaser developed it through the 1970s and 80s. Strict separation between data collection and prior theory; the literature review is deferred until after analysis; codes “emerge” from data under a positivist epistemology that treats the analyst as a neutral observer. Glaser’s 1992 book Basics of Grounded Theory Analysis is in part a direct rejection of the direction Strauss took the method.

Strauss-Corbin grounded theory

Strauss and Corbin’s 1990 textbook introduced a more procedural version with explicit coding paradigms (open / axial / selective) and the conditional/consequences matrix. The textbook made grounded theory teachable but Glaser argued it forced data into pre-given analytical categories — the famous “emergence versus forcing” argument. Strauss-Corbin GT remains common in health and nursing research; Corbin’s 2008 revision softens some of the earlier positions.

Constructivist (Charmaz) grounded theory

Charmaz (2006, 2014) rebuilt grounded theory on constructivist foundations: the analyst is not a neutral observer; data are co-constructed with participants; the resulting theory is interpreted, not discovered. The coding procedure is similar to Strauss-Corbin’s in shape (initial → focused → theoretical) but the epistemological claims are weaker and the writing is more reflexive. Constructivist GT is the dominant tradition in qualitative methods courses in 2026.

“We are not passive receptacles into which data are poured. We are part of what we study” — and the theories we develop carry the imprint of our perspective, our questions, and our analytical choices.
— Charmaz, 2014

Coding: initial → focused → theoretical

Charmaz’s coding sequence runs through three phases. The phases are not linear — you return to initial coding when new data complicates focused codes — but the dominant activity changes.

Initial coding

Line-by-line or incident-by-incident coding of early transcripts. Codes are short, active, and as close to the data as possible. Charmaz recommends gerund-form codes (“reframing identity,” “negotiating care”) because they keep the focus on action and process rather than static categories. The output is messy: dozens of codes per transcript, many overlapping.

Focused coding

The most-applied and most analytically promising of the initial codes are elevated to focused codes, which are then applied across the remaining data. Focused codes are fewer (typically 15–40) and synthesise across initial codes. This is where the codebook begins to stabilise.

Theoretical coding

The relationships between focused codes are themselves coded. What is the conditional structure (when X appears, Y follows; X requires Y as a prerequisite; X and Y are alternative responses to the same condition)? Theoretical coding is the move from a codebook to a theory. Glaser’s Theoretical Sensitivity (1978) catalogues coding families that can prompt this move; Charmaz treats them as a non-exhaustive resource rather than a fixed taxonomy.

Memo-writing as analysis

Memo-writing is the analytical engine of grounded theory. Every code, every emerging category, every theoretical hunch gets a memo. Memos are dated, freeform, and accumulate into the analytic record from which the eventual theory is written.

Three types of memo are worth distinguishing in practice:

Code memos — what a code is for, what it includes, what it excludes, where the boundaries are still ambiguous. Code memos make the codebook defensible.
Operational memos — methodological decisions: why an interview was transcribed differently, why a code was retired, what was changed in the interview guide after the first three participants.
Theoretical memos — speculation about relationships between codes, connections to existing literature, candidate explanations for puzzling patterns. Theoretical memos are the rough drafts of the eventual theory.

A grounded theory study with thin memo-writing is, in practice, just qualitative coding labelled as grounded theory. The memos are not optional documentation; they are the analysis.

Theoretical sampling and constant comparison

Two procedures distinguish grounded theory from most other qualitative methods.

Theoretical sampling. Participant recruitment is iterative and analytically driven. The first round samples for variation on the phenomenon of interest. Each subsequent round is chosen to test, extend, or refine the emerging categories. If your initial codes suggest that experience of X varies by role, the next round of interviews targets participants in roles you haven’t yet sampled, even if convenience would point elsewhere. Theoretical sampling continues until saturation (see below).

Constant comparison. Every new datum is compared with the existing data, with the existing codes, and with the emerging categories. When a passage doesn’t fit an existing code, either the code is wrong or the passage is the start of a new code; either way, the comparison forces a decision and a memo. Constant comparison is what stops grounded-theory coding from becoming autopilot — it converts the act of coding into an analytical question every time.

Theoretical saturation

Saturation in grounded theory has a specific meaning that has not always survived the term’s migration into other methods. It is the point at which theoretical sampling stops producing new properties or dimensions of the existing categories. It is a claim about the categories, not about the interviews.

Charmaz (2014) warns that “saturation” gets invoked as a stopping rule by analysts who haven’t done the theoretical sampling that makes the concept meaningful. A study that conveniently sampled 20 interviews and declared saturation has not, technically, reached theoretical saturation — it has run out of interviews. Be specific in the methods section: state the theoretical question, the sampling decisions, and what specifically stopped producing new properties.

When grounded theory fits

Grounded theory is the right method when:

The research question is process-oriented (“how does X unfold over time?” or “what social processes constitute Y?”) rather than experience-oriented.
Theory development is the goal, not theory application or description.
Iterative data collection is feasible — you can recruit participants in subsequent rounds, adapt the interview guide between rounds, and stop when categories saturate. Grounded theory in a study where 30 transcripts are delivered up-front and recruitment is closed is grounded theory in name only.
The timeline allows for memo-writing alongside coding. Compressed-timeline studies that skip memos should pick a different method and be honest about the choice.

When grounded theory does not fit and a more appropriate method exists, picking the right one is the methodological move. Reflexive thematic analysis (chapter 2) is often the right choice for what gets described as “light-touch grounded theory.” IPA (chapter 4) is the right choice for questions about meaning and experience rather than social process.