The Steps in Qualitative Data Analysis
One week, five separate accounts of how qualitative analysis proceeds, attributed to four different authorities and never reconciled with each other. Here they all are, including one recovered from a slide that reads as blank to every automated extraction of the file.
The four steps, as the deck sets them out
Two slides, headed "Stages in Qualitative data analysis", carry four steps with five actions in each. The regularity - five, five, five, five - is exact and the source does not remark on it. The counts of four and twenty below are derived by enumeration here; the source states neither.
THE FOUR STEPS, VERBATIM AND IN SOURCE ORDER
| Step | The five actions, as printed |
|---|---|
| Step 1 | "Read the text as a whole and make notes at the end of your reading" · "Examine what it's about" · "Look for major themes and consistencies" · "Look for contradictions, unusual events" · "Group cases into categories, sub categories" |
| Step 2 | "Read it again thoroughly" · "Mark text with highlights, underline, make marginal notes, annotations" · "Develop labels for codes" · "Highlight key words" · "Note analytical ideas" |
| Step 3 | "Systematically mark text with codes" · "Indicate themes" · "Review codes" · "combine similar codes - condense" · "Decide on groupings" |
| Step 4 | "Relate general theoretical ideas to text" · "Coding is part of the analysis" · "Identify significance for respondents in text" · "Look at interconnections between codes" · "Check relation of codes to research" |
Source: the week's slide deck, slides 6 and 7. Capitalisation and the lower-case "combine" are as printed.
Taken whole this is a recognisable shape for a first analytical pass: read for the sense of the thing, read again while marking, mark systematically and consolidate, then look at how the parts relate and whether they answer the question you asked. If you need one working sequence out of this week, start here - provided you also carry what is wrong with it. The marking and code-labelling actions are treated in detail on Coding Qualitative Data in Practice.
What does not hold together inside the four steps
Four defects to carry, not repair
Two markings, no stated difference
Step 2 says to mark the text with highlights, underlining, marginal notes and annotations. Step 3 says to mark it systematically with codes. The only stated difference is the word "systematically" and a change of marker. Step 2 already contains "Develop labels for codes", so codes exist before step 3 begins.
Two groupings, no stated relation
Step 1 ends "Group cases into categories, sub categories". Step 3 ends "Decide on groupings". Whether the step 3 groupings are the step 1 categories revised, or a different structure built over the codes, is not stated anywhere.
An assertion in a list of actions
"Coding is part of the analysis" is not something to do. It sits between two imperatives and tells the reader nothing to perform. It is also the only place in the deck that states the relationship between coding and analysis, and it states it without argument.
Fifteen undefined nouns across twenty actions
Themes, consistencies, contradictions, unusual events, cases, categories, sub categories, codes, labels, key words, analytical ideas, groupings, general theoretical ideas, significance, interconnections. None is defined on either slide; three are defined elsewhere in the week in incompatible ways, and "theme" nowhere at all.
The three-stage model recovered from the next slide
The slide immediately after the four steps carries one line of visible text: "The diagram below illustrates the steps in qualitative data analysis described in" a 2010 methods text, at page 263. Below that line sits a diagram - a vertical chevron list of three stages, each paired with a box holding two items.
THE THREE-STAGE DIAGRAM, RECONSTRUCTED FROM THE FILE
| Stage | Chevron label, verbatim | The two items in its box, in order |
|---|---|---|
| 1 | "data" | "Raw data" · "Data organised" |
| 2 | "refined" | "Data reduced" · "Interconnections noticed" |
| 3 | "positioned" | "Themes emerge" · "Theoretical understanding" |
Attributed by the slide's own caption to a 2010 methods text, page 263; the full citation is in the references below. Three stages, two items each, six items total - counts derived by enumeration, not stated by the source. The chevron labels are lower case in the original and the items inside the boxes are capitalised.
- Raw data
- Data organised
- Data reduced
- Interconnections noticed
- Themes emerge
- Theoretical understanding
Read as a trajectory this is useful, and it is the only place in the supplied material that names an endpoint for qualitative analysis. Data arrives raw and gets ordered; it is cut down and the connections in it are noticed; out of that, themes surface and something the source calls theoretical understanding is reached. It gives you a way to say where you are in the work, which nothing else in the week does.
Four steps or three stages? The source does not say
These two models sit on consecutive slides, describe the same process, are attributed to different authorities, and are never related by any sentence in the week. The attribution on the four-step slides says "4 stages" while every item there is labelled "Step" and the slide title says "Stages"; the diagram slide's caption then calls its three elements "steps".
WHERE THE TWO MODELS PARTLY ALIGN - AN INFERENCE, NOT THE SOURCE'S CLAIM
| Three-stage item | Nearest four-step action |
|---|---|
| "Data organised" | Step 1: "Group cases into categories, sub categories" |
| "Data reduced" | Step 3: "combine similar codes - condense" |
| "Themes emerge" | Step 3: "Indicate themes" |
| "Interconnections noticed" | Step 4: "Look at interconnections between codes" |
| "Theoretical understanding" | Step 4: "Relate general theoretical ideas to text" |
| "Raw data" | Step 1: "Read the text as a whole" |
This mapping is the reader's inference and is offered nowhere by the source. The alignment is imperfect: nothing in the three-stage model corresponds to step 2 at all.
The nine-item ordering exercise in the study notes
The study notes carry a third account, and it is the only place in the week where a count is both stated and correct: "There are nine steps for you to allocate a number or order." The reader is told to type a number against each, from one for the first through to nine for the last, and that "There are no right or wrong answers".
THE NINE ITEMS, AND THE POSITION EACH GLOSS PINS IT TO
| Item, as named | The constraint its own gloss states |
|---|---|
| Transcription of notes | "usually the first action after you have collected your data" |
| Initial Processing | "once you have transcribed your notes" |
| Summary sheets for each response | "after you have transcribed your notes" |
| Identify categories relating to patterns or themes identified | "after you have identified the core categories of your study" |
| Coding | "after you have identified categories relating to the patterns or themes identified" |
| Return to observe or ask further questions | "after you have done some initial processing, and can then be done at any time" |
| Discussion | "after you have done some analysis of your data" |
| Conclusion | sums up "the analysis you have done of your data and any interesting discussions" |
| Recommendations | "tend to come at the end of you study" - the typographical error is in every supplied capture |
Source: the week's study notes, section 3, printed in an order the source states is not the intended one.
What this list has that no other account does is the tail end of a project. Discussion, conclusion, recommendations and returning for more data are post-analysis activities, and this is the only place the study notes break writing up into named parts. Two of the four map onto the deck's second half; recommendations and returning to the field map onto nothing, and the word "recommendation" appears on no slide.
Five accounts, and the ground they share
Counting the two models above, the ordering exercise, the six reasons in the notes and the ten-bubble figure that forms a whole section of them, the week gives five separate accounts of what qualitative analysis consists of. They sit in three documents, differ in length and count, and no sentence anywhere relates any two of them. The six reasons are set out on What Analysing Qualitative Data Involves.
THE FIVE ACCOUNTS OF THE ANALYSIS PROCESS IN ONE WEEK
| Account | Where | Items | Attributed to |
|---|---|---|---|
| Six reasons that read as activities | Study notes, section 2 | 6 | The third-party study-skills site the notes are copied from |
| Nine-item ordering exercise | Study notes, section 3 | 9 | The same third-party site |
| Ten speech bubbles, no order | Study notes, section 6 | 10 | None given |
| Four steps of five actions | Slides 6-7 | 4 x 5 | Part one of an online video series, not supplied |
| Three-stage chevron diagram | Slide 8 | 3 x 2 | A 2010 methods text, page 263 |
The ten-bubble figure is reproduced in full on Ten Questions to Ask of Your Data. Counts are derived by enumeration except the nine, which the source states.
The five points all five accounts agree on - safe to cite
- Reading and re-reading the data comes first. Present in the notes' ordering exercise, in step 1 and step 2 of the four-step model, and as "Raw data" in the diagram.
- Categories and codes are developed from the data rather than imposed on it. Present in the notes' inductive-process item, in step 1, in the slide glossary's "Inductive codes", and in the worked example's "Inductive Categories".
- Codes are reviewed, condensed and grouped after a first pass. Present in step 3, in the first speech bubble, and as "Data reduced" in the diagram.
- Relationships between codes or categories are examined late, not early. Present in step 4, in two of the speech bubbles, and as "Interconnections noticed".
- Themes emerge from that examination rather than being fixed at the start. Present in the diagram, in one speech bubble, and in the notes' sixth reason.
Everything else differs, including how many stages there are - three, four, six, nine or ten - whether categories precede or follow codes, whether the process is linear or cyclical, and what the endpoint is. The endpoint is stated four ways across the week: theoretical understanding, recommendations, a logical chain of evidence, or nothing at all. The vocabulary problem underneath all of this is treated on Terms Used in Qualitative Data Analysis.
Using this on a project, and what the week does not deliver
One more thing before you lean on this week. It states two learning objectives, the first being "Evaluate presentation styles for qualitative data". Not one presentation style for qualitative data is named anywhere in it, and no two methods, models or coding types are ever compared on any criterion, which is what the verb requires. Across the six weeks now on record, fourteen objectives are stated, none is fully delivered, four are partially delivered and ten are not delivered at all. See The Six-Week Objective Audit.
What to carry forward
- Four steps of five actions on one slide pair; three stages of two items on the next. Same process, different authorities, no sentence relating them. Both are reproduced above; neither is the week's model.
- The three-stage model - raw data, organised, reduced, interconnections noticed, themes emerge, theoretical understanding - is invisible to text extraction. If you are working from a text copy of the deck you will not know it exists.
- The four steps do not distinguish their two markings of the text, produce groupings twice without relating them, and leave fifteen nouns undefined across twenty actions.
- Data reduction is named in three separate documents in this subject and no procedure for it appears in any of them.
- Five points of agreement run across all five accounts. Cite those; cite a named authority for anything more specific.
Frequently asked questions
Which model should I follow - the four steps or the three stages?
The material does not say, and it never relates them. The four steps are at the right grain to be run as a procedure; the three-stage model is better as a way of describing where you are up to. If you use either in a methodology chapter, attribute it to the authority the slide cites rather than to the teaching material, which reproduces both without choosing.
Why is the slide 8 diagram missing from my copy of the deck?
It is not missing, it is unreadable to text extraction. The diagram's content is stored as an embedded diagram object rather than as slide text or an image, so every automated reading of the file reports the slide as a caption over nothing. The six items are reproduced in the table above, recovered from the diagram data inside the file.
How do I actually reduce data?
This material does not say. "Data reduced" appears in the three-stage diagram, in the evidence-assembly figure elsewhere in the subject, and in the qualitative-research week's notes, and in all three places it carries no procedure. You will need a methods text for it, and you should cite that text rather than this material.
Can I complete the nine-step ordering exercise?
No. It was interactive on the web page the notes were copied from and there is no field to type into in any supplied capture, no answer key and no correct ordering stated anywhere. It also contradicts itself: it says there are no right or wrong answers and then pins every item's position in its own gloss. Use it as an inventory of analysis actions.
Do codes come before categories or after them?
The material answers both ways. The study notes make categories the input to coding - data is coded according to categories identified by reading and re-reading - and the previous week defines coding as creating categories, making them its output. The two weeks reverse the relationship and neither acknowledges the other. Say which order you used and why.
Is qualitative analysis linear or does it loop?
The week asserts both. The four-step model and the three-stage diagram are linear, and the slide glossary defines interim analysis as the cyclical process of collecting and analysing data during a study. The material asserts iteration in three places and shows an iteration nowhere. If your design collects and analyses in cycles, say so explicitly, because the models here will not carry the claim for you.
References and source attribution
- O'Leary, Z 2010, Doing your research project, Sage Publications - the work the diagram slide cites at page 263, and the only reference listed on the week's own closing slide.
- O'Leary, Z. 2017, The Essential Guide to Doing Your Research Project, 3rd ed., Sage Publications, London - the later edition cited elsewhere in the supplied source.
- Bryman, A. 2016, Social Research Methods, 5th ed., Oxford University Press, Oxford - cited elsewhere in the supplied source; a place to obtain the data-reduction procedure this material names and does not supply.
- Quinlan, C. 2011, Business Research Methods, 1st ed., Cengage Publishing.
- The supplied teaching source: the qualitative analysis week's slide deck, slides 6 to 8, and its study notes, sections 2, 3 and 6. The slide 8 diagram was recovered from the diagram data stored inside the deck file; it does not appear in any text extraction of it.
Suggested questions for Ask KEVOS
- Turn the four steps into a working analysis plan for twelve project manager interviews.
- Which of the five accounts of the analysis process should I cite in my methodology chapter, and how do I attribute it?
- What does data reduction actually involve, and which text should I cite for the procedure?
- Draft a paragraph describing my analysis process that is honest about where the model came from.
- How do I show in my write-up that themes emerged from the data rather than being decided in advance?