# Lesson 7 — Comparing Variables and Grounded Theory (v3 expanded)

*Companion-podcast transcript • Sarah & Kiffer*  
*~4994 words • ~27.7 min audio*

---

**Sarah:** Welcome back to Office Hours. I'm Sarah.

**Kiffer:** And I'm Kiffer. Today's episode is Lesson 7 of our qualitative methods course, Comparing Variables and Grounded Theory. And honestly, this might be the lesson where students start to feel like the apparatus they've been building over the last six weeks finally turns into actual analysis.

**Sarah:** Right. By this point students have a research question, a sampling plan, transcripts they trust, a codebook with maybe thirty codes, and an analysis framework with memos. And now they're sitting in front of twenty transcripts asking, okay, what do I do with all of this?

**Kiffer:** Exactly. And the textbook, Bernard, Wutich, and Ryan, treats Chapter nine and Chapter ten as a single arc, because both chapters are doing the same intellectual work in different vocabularies. Chapter nine is matrix displays. Chapter ten is grounded theory. And the engine underneath both of them is the same engine. Constant comparison.

**Sarah:** So that's where we'll spend our time today. Matrix displays first, then the grounded theory tradition, the pipeline of open, axial, and selective coding, and then how all of this lands in the capstone milestone for the week.

**Kiffer:** That's the arc. Let's start with the matrix tradition, because I think it's the more underappreciated of the two.

**Sarah:** Okay. Walk me into it. Where does matrix analysis come from?

**Kiffer:** The matrix tradition is associated above all with Matthew Miles and Michael Huberman, whose nineteen eighty-four sourcebook was the first widely adopted operational manual for big qualitative datasets in education and policy research. They were program evaluators. Multi-site studies. Thirty schools, twenty clinics, fifteen reform programs. Narrative case-by-case reporting did not scale. Matrices did. The third edition, revised by Johnny Saldaña in twenty fourteen after Huberman's death, is the canonical reference today.

**Sarah:** And the headline claim of that tradition is what they call display drives analysis. Unpack that for me, because at first it sounds almost trivial.

**Kiffer:** It sounds trivial and then you actually try to do it, and it stops being trivial. The claim is that the form in which you arrange your data on the page determines what patterns you can see. Running text hides comparisons. A two-dimensional grid makes them visible. So producing a good display is itself analytic work. It's not a presentation step after the analysis. It's a step in which the analysis happens. The cells of the matrix are the unit of analytic compression. Deciding what counts as the content of cell code-X, case-Y is an interpretation.

**Sarah:** Right. And that compression is where the analytic work lives. You're not pasting raw quotes into the cells. You're digesting the case into a phrase or two that captures what that participant said about that code.

**Kiffer:** Right. And the textbook walks through dozens of matrix variants, but most working researchers in health use just four. Conceptually ordered, time-ordered, role-ordered, and the two-dimensional code-by-case matrix. Each one does a different job.

**Sarah:** Let's go through those quickly. Conceptually ordered first.

**Kiffer:** A conceptually ordered matrix arranges its rows and columns by theoretical construct. So for the loneliness study we use as our running example, rows might be dimensions of the phenomenon, like triggers, embodied features, coping moves, structural critique. Columns might be theoretical contrasts, like situational versus existential loneliness. It's most useful when you already have an analytic framework and you want to populate it with evidence.

**Sarah:** Time-ordered?

**Kiffer:** Columns are chronological. Before the loss, immediate aftermath, one year out, current. Each cell is a compressed account of the phenomenon in that period for that case. Time-ordered matrices are how you recover a trajectory from a corpus of cross-sectional interviews that nevertheless contain retrospective accounts.

**Sarah:** Role-ordered?

**Kiffer:** Columns are participant roles. Patients, providers, family caregivers, system administrators. The point is to expose how the same phenomenon looks different depending on who's reporting it. These are workhorses in implementation science because they make visible the gap between the program-as-designed and the program-as-experienced.

**Sarah:** And then the workhorse of workhorses. Code-by-case.

**Kiffer:** Rows are codes, columns are cases. Each cell is a compressed analytic summary of what that participant said about that code. This is the matrix that sits directly on top of your coded transcripts and is the one most students will build for their capstone milestone.

**Sarah:** Let's get concrete. The lesson walks through a worked code-by-case matrix from a loneliness study. Six codes, eight participants. Codes are trigger, embodied feature, coping move, what helped that surprised them, structural critique, identity stake. And the participants span ages from twenty-two to eighty-two, with different life-stages, immigration statuses, and circumstances.

**Kiffer:** Right. And the part that's easy to miss is that producing the matrix is only half the work. Reading it is the other half. The textbook lays out three reading moves a disciplined analyst makes systematically, and a fourth that separates excellent matrix work from competent matrix work.

**Sarah:** Walk me through them.

**Kiffer:** First, read across rows. A row tells you how a single code varies across the corpus. Take the trigger row. Read across it and you start to see that triggers cluster around specific times and around specific contrasts. Maya's trigger is environmental, the SkyTrain at nine p.m. on a Sunday. Linda's is an object, her late husband Bill's empty chair. Aarav's is bicultural, the hour after a phone call home that highlights the displacement. Frank's is interpersonal, visitors leaving. Reading down a row generates analytic candidates. Here, you start to think loneliness triggers might sort into three sub-types. Environmental cues, relational contrasts, temporal markers. That sub-typing is a finding the row produces.

**Sarah:** And the second move is reading down columns.

**Kiffer:** A column tells you how the codes hang together for one person. Read down Amira's column and you see that her triggers, her embodied features, her coping moves, her structural critique, and her identity stake form a coherent picture. The loneliness she's describing is refugee loneliness as a specific configuration of trauma, displacement, language, and unrepeatable memory. Read down Chen's column and you see a different configuration. Achievement-bound loneliness, where the trap is partly self-built and partly structural. Column reading is how you preserve the integrity of the case while still doing comparison.

**Sarah:** So the matrix is doing two jobs at once. The rows force cross-case comparison. The columns force you to keep each person whole.

**Kiffer:** Exactly. And the third reading move is empty cells. Empty cells are information. The rule of thumb is to ask first whether the data are missing because of an under-elicitation issue or whether the case really has nothing in that slot. Missing data goes in the limitations section. Substantive absence goes in the findings. The relative thinness of the structural-critique row for some of the younger participants in the loneliness corpus reflects a real pattern. Younger participants are less likely to articulate their loneliness as a product of policy or institutions. That's a substantive finding, not a hole in the data.

**Sarah:** And the fourth move, the one that separates excellent from competent.

**Kiffer:** Reading for anomalies. An anomaly is a cell whose content doesn't fit what its row and column would predict. The grounded theorists call this looking for negative cases. Frank's identity-stake cell, his calm acceptance of companionship with absent persons, is anomalous against the broader late-life pattern of acceptance-with-sadness. Anomalies generate the most productive analytic memos because they force you to either refine the dominant pattern or identify an alternative type.

**Sarah:** Before we leave matrix work, I want to ask about something that gets people from interpretivist traditions a little nervous. Magnitude coding. Counting in qualitative work.

**Kiffer:** Yeah, this is the one. Magnitude coding is attaching a numerical or ordinal value to a coded extract. Low, medium, high. Or simply counting how many transcripts contain a code. Saldaña treats it as a recognized coding family. Miles, Huberman, and Saldaña treat it as routine matrix-cell content. Bernard, Wutich, and Ryan defend it whenever the question being answered is about commonness, dominance, or change.

**Sarah:** And the defense is basically about intellectual honesty. If you've decided that a pattern is real because it shows up often, you owe the reader a defense of often.

**Kiffer:** Right. The least defensible version of often is in my impression. The most defensible version is in N of M transcripts. Counting transforms an impressionistic claim into a checkable one. It doesn't turn the qualitative study into a quantitative study. The units being counted are qualitative codes, not pre-specified variables.

**Sarah:** And the textbook is also honest about the argument against counting.

**Kiffer:** Yeah. Frequencies in qualitative work are artefactual. They reflect what the interview guide asked, who was sampled, and how long each interview ran. A code that appears in eighteen of twenty transcripts may show up that often because the interview guide asked about it, not because the phenomenon is widespread. So the defensible practice is to report counts and be explicit about their limits. Count when commonness matters to the claim. Don't count when the claim is about meaning or mechanism. Report the denominator. Acknowledge the artefactual character of the count when the code was directly elicited.

**Sarah:** And there's a lovely phrase from Howard Becker that frames this whole thing. Quasi-statistics.

**Kiffer:** Becker coined it in nineteen fifty-eight. Quasi-statistics is the informal counting that good qualitative researchers do as part of their work. The running estimate of how many cases fit, how many don't, how strong the trend is. There are no confidence intervals. There's no inferential machinery. But quasi-statistics discipline the analyst against over-claiming. Magnitude coding in a matrix is one operationalization of that. The point is intellectual honesty, not statistical inference.

**Sarah:** Okay. So matrix work systematizes comparison visually. Now grounded theory. Which the textbook places next to matrix analysis because it systematizes the same intellectual move procedurally.

**Kiffer:** Right. And here we need to slow down, because grounded theory is the most-cited qualitative methodology in health research and probably the most misused. Most papers that claim to have used a grounded theory approach have used some version of thematic coding and called it grounded theory because the name carries methodological prestige.

**Sarah:** So let's establish what it actually is. Start with the origins.

**Kiffer:** Barney Glaser and Anselm Strauss published The Discovery of Grounded Theory in nineteen sixty-seven, and it was a polemic. Mid-century American sociology was dominated by Talcott Parsons's grand theory, which proposed elaborate conceptual schemes that empirical work was expected to verify, and by Paul Lazarsfeld's quantitative survey tradition. Qualitative work, what survived of it, was treated as preliminary to real research.

**Sarah:** So Glaser and Strauss said, actually, qualitative research can generate theory directly from data. Theory that's grounded in evidence rather than imposed from above.

**Kiffer:** Exactly. And they laid out the moves. Theoretical sampling, where you sample to develop categories, not to represent populations. Constant comparison. Coding in successive levels of abstraction. Memo-writing as the engine of conceptual development. Theoretical saturation as a stopping rule. The book is combative in tone. Glaser and Strauss were arguing for the legitimacy of a kind of research that mainstream sociology was actively dismissing.

**Sarah:** And the nineteen sixty-seven book had a tension in it that ended up splitting the methodology in the nineteen nineties.

**Kiffer:** Yeah. The book emphasized that the analyst should approach data without preconceived theoretical commitments and let categories emerge. But the book also emphasized systematic procedures, coding levels, axial relationships, that imposed a particular structure on what was emerging. Was grounded theory radically inductive or systematically procedural? The book held both stances in tension. The next generation pulled the two apart.

**Sarah:** The Glaser-Strauss split.

**Kiffer:** Strauss, with Juliet Corbin, published Basics of Qualitative Research in nineteen ninety. The book emphasized procedural rigor. Specific coding stages, open, axial, selective. A formal paradigm model for axial coding. Causal conditions, phenomenon, context, intervening conditions, action strategies, consequences. A structured route from open codes to a core category. Glaser objected. In Basics of Grounded Theory Analysis in nineteen ninety-two, he argued that Strauss and Corbin had betrayed the original method by imposing a coding apparatus on data before allowing categories to emerge. The paradigm model, in Glaser's view, was forcing data into a pre-given structure.

**Sarah:** And the technical disagreement, whether to use the paradigm model in axial coding, tracked a deeper philosophical disagreement.

**Kiffer:** Yeah. Glaser's stance is closer to mid-twentieth-century positivist sociology, despite his methodological radicalism. The analyst is a neutral discoverer of theory already in the data. Strauss's stance is closer to symbolic interactionism. The analyst is a procedural constructor of theory through specific coding moves. Health researchers don't need to take a side in this dispute, but they do need to know it exists, because the phrase grounded theory in a published paper can mean very different things depending on which lineage the author is in.

**Sarah:** And then there's the third path. Kathy Charmaz.

**Kiffer:** Charmaz was a student of Strauss. Her substantive work was on chronic illness identity. Her methodological text, Constructing Grounded Theory, two thousand and six and a second edition in twenty fourteen, reframed the methodology in explicitly constructivist terms. The theory is not discovered in the data; it's co-constructed by analyst and participants through interpretive engagement.

**Sarah:** And that variant is now the most cited in health research.

**Kiffer:** By a wide margin. What Charmaz changes. First, the epistemology. There isn't a single theory in the data waiting to be found. Multiple defensible readings are possible. The analyst's positionality, theoretical commitments, and interpretive choices shape what's seen. Second, the literature. Where Glaser insisted on literature review only after analysis, Charmaz treats the literature as an interlocutor the analyst is always in conversation with. Third, attention to power, voice, and the social context of meaning-making. Fourth, the coding procedures. She preserves the open-axial-selective sequence but loosens the paradigm-model requirement. She emphasizes focused coding, selecting the most analytically productive codes, over rigid axial procedures. And fifth, memos take on heightened importance. The line between memo and finding is intentionally blurred.

**Sarah:** And the practical takeaway for students is that all three variants share the procedural backbone. Open coding, then axial coding, then selective coding around a core category. The differences are philosophical and procedural at the edges.

**Kiffer:** Right. And the textbook's stance is that all three are legitimate, but students should be able to say which one they're working in and describe their actual practice honestly rather than invoking grounded theory as if it were a single thing.

**Sarah:** One more clarifying question before we walk through the pipeline. How is grounded theory different from theme identification, which we covered earlier in the course?

**Kiffer:** Good question. There's significant overlap. Both are interpretive. Both involve coding. Both require comparison. Both produce categories of meaning. But there are real differences. Theme identification ends with a descriptive account of what the major themes in the dataset are. It's a finding about what is in the data. Grounded theory ends with a substantive theory. An explanatory account of how the elements of a phenomenon relate to one another, organized around a core category. It's a finding about why things in the data are the way they are.

**Sarah:** And the deeper difference is the procedural commitment to theoretical sampling and theoretical saturation.

**Kiffer:** Right. Theme identification on a fixed dataset, the way our capstone is structured, can't literally do theoretical sampling, because the data are already collected. Most contemporary health-research grounded theory studies acknowledge this and report a grounded-theory-informed or modified grounded theory analysis rather than claiming to have done full grounded theory. The Bernard-Wutich-Ryan stance, and our stance in this course, is to claim what you actually did and be transparent about what was unavailable to you.

**Sarah:** Okay. The pipeline. Walk us through open coding first.

**Kiffer:** Open coding is the first analytic engagement with the data, and its purpose is to break the data open. To fragment the seamless flow of an interview into small analytic chunks that can be compared with chunks from other interviews. The fragmentation is deliberate. A transcript read straight through induces what Charmaz calls a narrative trance. The analyst gets caught up in the participant's story and stops noticing comparisons. Open coding interrupts that trance.

**Sarah:** Two main techniques. Line-by-line and segment-by-segment.

**Kiffer:** Line-by-line was Glaser's original move, developed by Charmaz. The analyst attaches a code to every line, asking continuously, what is happening in this line? what is this an instance of? The codes are typically gerunds. Action-words ending in -ing. Naming the trigger. Evading the shame. Rationing the call home. Gerunds because gerunds make process visible. Line-by-line is laborious and tends to over-generate codes. It's most useful at the very start of a grounded-theory project. Segment-by-segment is less fragmenting, generates fewer codes, and is what most working researchers actually do once they have a feel for the data.

**Sarah:** And the codes at this stage are intentionally over-generated. You're being generous, not selective.

**Kiffer:** Right. A fifteen-line excerpt might produce eight to twelve codes. A forty-five-minute interview might produce sixty to a hundred. Many will turn out to be near-duplicates and get merged. Many will be too narrow and get subsumed under broader categories. The selection happens later, in axial coding. The point at this stage is to over-generate.

**Sarah:** Let's make this concrete. The module walks through Maya's definitional passage about loneliness. Sunday at nine p.m. on the SkyTrain. Everyone on their phones. Nobody acknowledges anybody else. The passage is short but it produces something like nine open codes on a line-by-line pass. Naming a specific trigger. Locating loneliness temporally. Marking the surrounded-yet-alone paradox. Articulating non-recognition. Imagining one's own invisibility. Defining loneliness as being around people who don't see you.

**Kiffer:** Right. And notice three features of that code list. First, every code is a gerund or a noun phrase tied to an action. Second, the codes are pitched at the level of what the participant is doing in the talk, not what their inner life is. And third, several codes look near-duplicate. Non-recognition, invisibility, non-noticing. The duplication is intentional. The codes will be compared and consolidated later.

**Sarah:** And then constant comparison enters immediately. You take Maya's passage about being lonely in a crowd on the SkyTrain. You compare it to Aarav's passage where he uses the Telugu word ekakitatvam, alone in a crowd. Same phenomenon, different vocabulary. You merge those into a provisional category. Loneliness-inside-crowdedness.

**Kiffer:** Then you read Linda's passage about Bill's empty chair. And constant comparison forces you to recognize that this is not the same phenomenon. This is loneliness anchored in the absence of a specific other. Now you have at least two sub-types. Presence-paradox loneliness and absence-anchored loneliness. That sub-typing is the kind of conceptual product open coding plus constant comparison generates.

**Sarah:** Move us to axial coding.

**Kiffer:** Where open coding fragmented the data, axial coding reassembles it. Not in the original narrative order, but along analytic axes that link codes to one another. The name comes from the image of an axis. A category that other codes rotate around. The work of axial coding is identifying which categories are central and how the other codes relate to them.

**Sarah:** And Strauss and Corbin's paradigm model is the apparatus that's both the most operationalizable and the most controversial.

**Kiffer:** Right. The paradigm model asks you to organize codes related to each category along six dimensions. Causal conditions. What gives rise to the phenomenon? Phenomenon. The central category itself. Context. The specific properties of the setting. Intervening conditions. Broader background conditions. Action and interactional strategies. What people do in response. Consequences. What results from those actions. Glaser's objection was that this template forces data into a pre-given mold. Charmaz preserves the spirit, identifying central categories and the codes that orbit them, but doesn't require the strict six-cell template. Her version is focused coding.

**Sarah:** And the module walks through an axial pass on the loneliness data around a category called loneliness-inside-companionship. Take us through what that looks like.

**Kiffer:** Sure. The phenomenon is loneliness inside companionship. The felt aloneness that arises specifically in the presence of others. The causal conditions cluster as a mismatch between what the surrounding others can offer and what the participant needs. Phones-and-strangers for Maya. Housemates-who-don't-share-mother-tongue for Aarav. Husband-who-couldn't-name-it for Elena. The context is specific spaces, like the SkyTrain or a shared apartment, at specific times. Intervening conditions are migration, life-stage transitions, the precarity of attention economies. Action strategies sort into sub-types. Portable rituals, like cooking from memory or daily phone calls. Seeking specific matched others. Numbing. Building a new frame, like joining a group with a shared experience. And consequences sort by whether the coping moves match the underlying deficit or mismatch with it.

**Sarah:** And that axial display does analytic work the open codes alone couldn't. It locates the central phenomenon, identifies its causal structure, and gives you a way to sort coping strategies into sub-types based on whether they match or mismatch the deficit.

**Kiffer:** Right. And then selective coding is the third stage. By this point you have many axial categories. Selective coding integrates them around a single core category that captures the central analytic insight of the study. The core category is the one that, when you tell it, the whole pattern of categories falls into place.

**Sarah:** Four criteria for a core category.

**Kiffer:** It must be central. It appears in most of the cases. It must have explanatory reach. The other categories relate to it in specifiable ways. It must be abstract enough to apply beyond the immediate data but specific enough to retain its content. And it must hold up under negative-case examination. Cases that seem not to fit either are explained, or the core category is refined to accommodate them.

**Sarah:** And for the loneliness study, the candidate core category that the worked example arrives at is, the work of converting presence into recognition.

**Kiffer:** Right. This is more abstract than loneliness-inside-companionship, which is just the phenomenon. It proposes an explanatory process. Loneliness, in these data, is what happens when participants cannot convert physical co-presence into felt recognition. Their coping moves can then be sorted by whether they go after presence or after recognition. The shape of late-life loneliness, like Linda and Frank and Helen, becomes a particular case of this work. The recognition-anchors are now memories of the dead, and the work is converting present-day life into something the deceased other would have recognized.

**Sarah:** And the negative case sharpens the core category rather than refuting it. Frank's calm acceptance of companionship with absent persons could initially read as not fitting the category. But the closer read is that Frank has actually completed the work. He's built a stable relationship with the absent through ritual and routine, and his calm reflects the resolution of the work, not its absence.

**Kiffer:** Exactly. And that's the move that adds a temporal-trajectory dimension to the category. Recognition work has a trajectory from acute failure to chronic struggle to settled adaptation. Frank is the developmental endpoint, not the negation of the category. This is what good negative-case analysis looks like.

**Sarah:** Let's zoom out for a minute. Constant comparison is the engine of all of this. The module names four levels where it operates.

**Kiffer:** Yeah. Within-extract, comparing parts of a single extract to one another. Within-case, comparing extracts within a single transcript. Across-case, comparing extracts and codes across transcripts. And category-to-data, comparing the developing analytic category back to specific instances. The discipline is not optional. Without continuous comparison at all four levels, what looks like grounded theory devolves into thematic description with a fancier name.

**Sarah:** And theoretical sampling. On a fixed dataset, theoretical sampling becomes a reading-order question.

**Kiffer:** Right. With live recruitment, theoretical sampling means deciding who to interview next. With a fixed corpus, it means deciding which transcript or extract to read next. The principle is the same. The choice is driven by analytic need, not by completeness. If a category looks productive, you sample the corpus for transcripts that should test its boundaries. Confirming cases, candidate negative cases, boundary cases. Your analytic memo should explain why those were chosen and what they were chosen to test. Reading transcripts in numerical order without analytic justification is not theoretical sampling.

**Sarah:** And then the contested concept. Theoretical saturation.

**Kiffer:** Yeah. The classical formulation is that you stop sampling when additional cases stop producing new conceptual content. Strauss and Corbin extended this to three sub-criteria. No new properties of categories emerge. Relationships between categories are well-developed. The categories cohere into a theoretical model.

**Sarah:** And the contemporary critique.

**Kiffer:** Three lines of critique. The operational critique is that no new information is in the eye of the analyst. Saturation has been used to justify ending data collection at sample sizes ranging from six to sixty in the published literature, with very little consistency. Hennink and Kaiser in twenty twenty-two reviewed published studies and found that most saturation claims couldn't be reproduced from the available methods information. The conceptual critique is that saturation assumes data are samples from a finite conceptual space, which is incompatible with constructivist epistemologies. If meaning is co-constructed, there's always more to be constructed. And the practical critique is that in fixed-dataset studies, saturation in the original sense is unreachable, because there are no further cases to sample.

**Sarah:** So the defensible contemporary practice is to use saturation as a guide rather than claim it as a binary state.

**Kiffer:** Right. Most thoughtful published studies now report saturation alongside an explicit account of what would have shown saturation had not been reached. Something like, by the eighteenth interview, no new properties of the four core categories emerged; analysis was stopped after the twentieth interview confirmed this. That's reportable. That's checkable. Just claiming saturation as if it were a fact is not.

**Sarah:** And memos. Memos are the engine of theoretical development. The lesson names three kinds.

**Kiffer:** Code memos define a particular code, give exemplars, note its boundaries, flag instances that nearly fit but don't. Category memos define an analytic category at a higher level, name its properties and dimensions, describe its relations to other categories, explore its negative cases. Theoretical memos work out the overall structure of the developing theory, name the candidate core category, show how the other categories relate to it. The three kinds form a hierarchy of abstraction, and the final write-up of the substantive theory is, in effect, the integration of accumulated theoretical memos into a single coherent account.

**Sarah:** Let's land the practical piece. The week seven milestone has students choose between Option A, a matrix display, and Option B, a grounded-theory open-coding pass. How should they think about the choice?

**Kiffer:** It should be driven by the analytic stance of their capstone. Matrix work is the cleaner choice when the research question is comparative. How does loneliness vary across subgroups, contexts, or life-stages? If you can name your comparison dimensions in advance, build a matrix. Grounded theory is the cleaner choice when the research question is explanatory. What is loneliness as a process or mechanism in this corpus? If you have a sense of a candidate core category but want the open-coding pass to test it, that's grounded theory.

**Sarah:** Both options require two to three grounded memos plus a one-page reflection. And both are graded on the same criteria. Procedural fidelity. Analytic depth. Evidence of constant comparison. Reflection quality. Transparency about analytic choices.

**Kiffer:** Right. And the line about transparency is the one I want students to take seriously. Document which transcripts you chose and why. Document your reading order. Document the codes that emerged and how you consolidated them. The methods section of your eventual paper should let a reader retrace your steps.

**Sarah:** Okay. Let's pull this together. Six or seven takeaways.

**Kiffer:** Sure. First, matrix analysis and grounded theory are doing the same intellectual work with different vocabularies. Both are engines for moving from a stack of coded transcripts to a defensible explanatory account. Both run on constant comparison. The matrix tradition systematizes comparison visually. Grounded theory systematizes it procedurally.

**Sarah:** Second, display drives analysis. Producing a matrix is itself analytic work. Reading a matrix is half the work after that. Rows for cross-case patterns. Columns for within-case coherence. Empty cells for informative absence. Anomalies for memo-writing.

**Kiffer:** Third, magnitude coding, counting in qualitative work, is defensible whenever the claim is about commonness, with the denominator reported and the artefactual character of the count acknowledged. It's quasi-statistics in Becker's sense, not statistical inference.

**Sarah:** Fourth, grounded theory exists in three substantively different variants. Glaserian, where categories emerge and the literature is reviewed after analysis. Straussian, with the paradigm model in axial coding. Charmazian, which is constructivist, co-constructed, and now dominant in health research. The procedural backbone is shared. Open coding, axial coding, selective coding around a core category.

**Kiffer:** Fifth, the pipeline is held together by constant comparison at four levels, theoretical sampling driven by analytic need, and memo-writing as the substantive analytic vehicle. Take any of those out and you've got thematic analysis under a more prestigious name.

**Sarah:** Sixth, theoretical saturation is a useful but genuinely contested concept. Treat it as a guide to analytic effort, not as a binary state. Report what saturation would have looked like and what evidence supports the claim.

**Kiffer:** And seventh, the most important grading criterion is transparency. Claim what you actually did. Don't claim full grounded theory if your dataset was fixed. Don't claim saturation without evidence. Don't claim theoretical sampling if you read transcripts in numerical order. The discipline of qualitative work depends on naming the difference between what you did and what you claimed to do.

**Sarah:** That last point is the through-line of this whole course, honestly. Procedural honesty is the rigor.

**Kiffer:** It really is. Okay, that's Lesson seven. The week seven milestone is your first sustained piece of comparison-based analysis on the loneliness corpus. Pick matrix or grounded theory based on your capstone stance. Build the artefact. Write the memos. Write the reflection. Be transparent.

**Sarah:** Next time we step from grounded theory's depth-with-explanation toward content analysis, which turns the comparison engine outward to text corpora at scale. And we'll see how the quantitative-leaning content-analytic tradition sits beside the more interpretive work we've been doing.

**Kiffer:** Looking forward to it. Thanks for spending the time with us today.

**Sarah:** Thanks everyone. We'll see you in Lesson eight.