INT6067 Chap.6 Qualitative Research and Methods
Qualitative Research and Methods
Five lines that hide every serious decision
The course teaches qualitative method across two sessions and hands out a short procedure: take videos of the classroom, have the video transcribed into text covering both movement and discourse, build a coding scheme, code against that scheme, and analyse the codes into categories and themes.
Interpretation enters at exactly two points in that sequence, when the scheme is written and when codes are gathered into themes, and those are the two steps a reader will challenge. Everything else is record keeping.
The most expensive misunderstanding here is that qualitative work is the option with fewer rules; it has different rules and at least as many, because a coding study can only be judged by whether somebody else, given your scheme, would produce the same codes from the same recording.
Five ways in, and what each one can see
The handout is written for a classroom study and for a usability test of an application, system or website at the same time, and it names the same collection modes for both.
In person observation is fast and cheap and cannot be revisited. Remote observation is less disruptive and loses the room. Video capture supports a second coder and costs consent and transcription time. An open question or interview is the only mode that reaches reasons. A structured survey of experience and attitude towards a design scales quickly and reports what participants say rather than what they did.
Observation tells you what happened and cannot say why; an account tells you what a participant thinks happened and cannot say whether it did, which is why pairing a behaviour mode with an account mode produces findings that survive questions.
Labels have to be defined by what a camera could see
Two labels are a starting point rather than a scheme, because a construct such as engagement is an inference and a coder can only see behaviour.
A usable scheme lists, under each label, the observable events that count as instances of it, states the boundary rule for the hard cases, declares the unit that carries a code, and has been piloted with a second coder on one recording before the rest is touched.
The unit matters more than students expect: a fixed interval, a conversational turn and an episode bounded by a task change give different counts from the same video, so a scheme that omits it cannot be replicated even by its own author.
Categories and themes are two operations
A category groups codes that belong together by kind.
A theme is a claim about what those categories mean for the research question, and it is a sentence rather than a heading. Collaboration is a category; collaboration talk rises when the task has no single right answer is a theme. A findings section that lists categories and stops has done four of the five steps.
What this chapter covers
- 01
The five step route from recording to themes
- 02
The two points at which interpretation enters
- 03
Five collection modes and what each can and cannot see
- 04
Pairing a behaviour mode with an account mode
- 05
Defining labels by observable events rather than by constructs
- 06
Boundary rules for the cases that ruin coder agreement
- 07
Declaring the unit of analysis before coding begins
- 08
Categories against themes, and why the distinction costs marks
- 09
Consent and privacy as timeline items rather than paperwork
Two labels turned into a scheme a second coder could use
- 3Write observable definitions under each label.
- 2Write the boundary rule for the hard case.
- 3Declare the unit and describe the pilot.
Key terms
- Coding Scheme
- The document listing the labels applied to qualitative data, their definitions in observable terms, the unit that carries a label, and the rules for boundary cases.
- Observable Definition
- A statement of what a coder would have to see for a label to apply, written in the vocabulary of the recording rather than of the theory behind the label.
- Coder Agreement
- The extent to which two people independently applying the same scheme to the same data produce the same codes. Where they disagree is more informative than how often.
- Category
- A grouping of codes that belong together by kind. It is a heading, and on its own it is not yet a claim about the research question.
- Theme
- A sentence asserting what a set of categories means for the research question. A findings section that lists categories without themes has stopped one step early.
- Usability Test
- An evaluation of whether an application, system or website functions for its users, run with observation, video capture or an experience survey. The course handout treats it with the same apparatus as a classroom study.
- Positionality
- The researcher's own place in the setting being studied. Naming it is a strength in a proposal, because a teacher studying her own class sees things an outsider cannot and misses things an outsider would notice.
Qualitative Research and Methods FAQ
Can I report counts from coded data?
Yes, and describe them as what they are. A statement that one label appeared eighty-four times and another nineteen times in one class is a legitimate summary of your coding and carries every judgement the scheme built into it. What it is not is a statistical result, because there is no sampling behind it.
A proposal that reports frequencies plainly reads as stronger than one that dresses them up, and a percentage with no sample behind it costs a reader's trust in the rest of the section.
What does the proposal need if the data does not exist yet?
The apparatus rather than the findings: the setting and participants, the collection mode and why, the unit of analysis, the starting labels and where they came from, the plan for checking agreement with a second coder, and the consent and privacy arrangements.
For classroom recording those arrangements are substantial and slow, and one of the course readings on educational data mining closes on data privacy as a serious issue for the field rather than as paperwork.
Where should my starting labels come from?
Both from published work and from the data, in that order. Labels adopted from studies in the same area arrive with definitions other people have already argued about, and they let you compare your findings with theirs, which is one of the things a reader will ask of your work. Labels invented from one recording drift towards whatever that lesson happened to contain.
Any label the data forces on you is legitimate provided you mark it as added during coding and justify it.
How do I convince a reader that my interpretation is not the only possible one?
Three commitments are cheap to make and convincing to a marker. Put a second coder on a portion of the data and resolve disagreements by revising definitions rather than by averaging. Report quotations or observed episodes alongside each theme, so a reader can inspect the evidence and disagree with you specifically. And state your own position in the setting, since that shapes both what you notice and what you miss.
Assessment move
Take ten minutes of any recording you have lawful access to, even a public lecture, and code it twice a week apart without looking at the first pass. The disagreements with your earlier self are the fastest possible lesson in why definitions have to be observable, and they cost nothing in consent. Then write your scheme in the form a second coder would need before you write a word of the method section.
Working through Qualitative Research and Methods in INT6067? Sia is AskSia’s AI Education tutor — ask any INT6067 Qualitative Research and Methods question and get a clear, step-by-step explanation grounded in how INT6067 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.
