The Action Research Thesis Structure
An insider study of organisational change, written up across 447 pages after three and a half years in the field. Its two most copyable features are sub-heading templates that repeat every time a new case or a new intervention appears.
The shape of the document
The thesis runs to 447 pages: about 21 pages of front matter, 273 pages of body, about 13 pages of references and roughly 140 pages of appendices. Appendices are close to a third of the submission.
The front matter order is conventional: title, acknowledgements, contents, a list of 38 figures, a list of 7 tables, a glossary of terms and abbreviations, a summary headed "Abstract" on the page itself, and a signed declaration.
Every figure on this page is a fact about that one submission. None of them is a target, a norm or a benchmark for your own work.
The first three chapters look like any thesis. The next four do not, and the difference is the whole point. Set against the other four examined works in Thesis Structures Compared, this is the only one whose chapter numbering tracks time in the field rather than the logical order of an argument.
Why there is no findings chapter
In a conventional thesis, data is collected, then analysed, then reported in one place. In this one, each learning phase chapter carries its own data, its own analysis and its own conclusions. There is no chapter where the findings are assembled, because the findings arrived in four instalments and each instalment changed what happened next.
The four phases are named for a progressive handover of control from the researcher to the organisation.
- Build up
- Facilitation handover
- Operational management handover
- Diffusion
Read that sequence as a claim about the intervention: at the start the researcher runs everything, and by the end the organisation is running it without them and spreading it. A chapter architecture that names the phases this way makes the handover itself auditable. A reader can check whether phase three actually shows the organisation doing what phase two shows the researcher doing.
Two templates that carry the whole thesis
Inside the phase chapters, the same two sub-heading sequences repeat. This is the single most teachable feature of the work: instead of inventing a narrative shape for each of many episodes, the writer fixed two shapes and reused them.
Template one — every case study inside a phase
Introduction
What this case is, where it sits in the phase, and why it was undertaken at this point.
Process
What was actually done, in sequence. The descriptive spine of the case.
My Role as a Researcher
A standing heading that forces the insider question to be answered every time rather than once in the methodology chapter.
Learning Outcomes
What was learned from this case — the analytic payload, and the hook into the next case.
Template two — every learning-set write-up
Salient Features of the Situation
The state of things before the action. Establishes the baseline the action was responding to.
Reason for This Particular Action
Why this action and not another. This is where the choice is justified, before the outcome is known to the reader.
Procedure
The mechanics. What was run, with whom, over what period.
Difficulties I Encountered
A standing heading for what went wrong. Difficulties are structurally required, not optional confession.
What Happened
The observed result, kept separate from the interpretation of it.
Learning Outcomes — Managerial outcomes / Research outcomes
Split into two named sub-headings. This is the dual reporting device treated in the next section.
Conclusion
What this episode establishes, and what it leaves open for the next cycle.
Three of those headings do work that most theses leave to the writer's discretion. "Reason for This Particular Action" forces justification before outcome. "Difficulties I Encountered" makes failure reportable. "My Role as a Researcher" keeps the insider problem visible in every episode rather than settling it once, abstractly, in chapter three.
Reporting findings twice, for two audiences
A practitioner doctorate has to satisfy two readerships whose interests do not coincide. The organisation wants to know what changed and whether it worked. The examiner wants to know what is now known that was not known before. Most practitioner theses try to serve both in one stream of prose and serve neither well.
This work solves it with two named sub-headings inside every learning outcomes section.
Managerial outcomes
- What changed in the organisation as a result of this action
- Written for the practitioner reader and the sponsor
- Local, specific, tied to this setting
- Justifies the intervention as work
Research outcomes
- What the episode contributes to knowledge
- Written for the examiner and the wider field
- Stated as learning that could travel beyond this setting
- Justifies the intervention as research
The device is cheap to adopt and hard to fake. If you cannot fill the research outcomes heading for a given episode, that episode was consultancy rather than research — and knowing that early is worth more than discovering it at examination.
An emergent literature review and late-arriving questions
The literature review is one chapter of 39 pages with 13 top-level sections, organised thematically. Its own introduction states that it was driven by events as they unfolded in the field, with new bodies of literature added as new problems surfaced. It is emergent, and it says so.
Two structural consequences follow. The chapter has no concluding section, and there is no gap statement in the conventional position at the end of the review. The nearest equivalent is a mid-chapter section that critiques the literature and justifies the method choice; the contribution claim is deferred to the methodology and conclusion chapters. How the other four works handled the same problem is set out in How Real Literature Reviews Are Organised.
The research questions arrive late too. Chapter one carries a single one-sentence "thematic concern". That sentence is restated in chapter three and only there expanded into four numbered research questions — three of the form "to what extent…" and one of the form "are X applicable in Y context?".
Chapter three then adds four reflective methodological questions in a completely different register: why was my research important, how was the research to be undertaken, how could we know when it was complete, and how could we know when it was successful. There are no hypotheses anywhere in the work. The comparison across all five examined works is in Where Research Questions and Hypotheses Live.
First person, and the researcher as a unit of analysis
The thesis is written entirely in the first person. Chapter one defends that choice by citing three earlier action research doctorates as precedent — a precedent argument rather than a philosophical one.
That is worth noting as a technique. The defence is not "first person is appropriate to qualitative research"; it is "examined doctorates in this tradition are written this way, and here are three". A supervisor who is uneasy about first-person prose is usually more moved by examined precedent than by argument.
The first person is not decorative here. The researcher's own leadership was measured before and after using a standardised published inventory, and the result is reported as a finding. The researcher is a unit of analysis in their own study, which is a position the sub-heading "My Role as a Researcher" exists to keep honest.
Rigour, appendices and the missing ethics apparatus
Analysis is qualitative content analysis with coding and categorisation, followed by cross-case analysis. Quantitative data is deliberately limited to frequencies and means. Higher-order statistics are refused on two stated grounds: the sample was too small, and they would have added little to the rich qualitative data.
Rigour is claimed through triangulation of three of four named types — data, methodological and investigator. Theory triangulation is not claimed. A second investigator co-developed the coding. One standardised published instrument was used pre and post; the other instruments were researcher-built and are stated to be non-standardised. That mixture of claimed and explicitly unclaimed rigour is examined further in Rigour, Triangulation and Validity in Practice.
WHAT THE ~140 PAGES OF APPENDICES CONTAIN — AND WHAT THEY DO NOT
| Present | Absent |
|---|---|
| Raw field data, including a list of 211 problems generated at a search conference | Consent forms |
| Researcher-built instruments and their scored results | Any ethics approval certificate |
| Organisational artefacts — process charts, report forms, review forms | Interview or discussion transcripts |
| Summarised correspondence and event chronologies | A list of appendices, or page numbers on them |
| Training slides and per-team feedback | Consistent numbering — items run 4-1 to 7-6 but appear as both 4-1 and 4.3 |
About 23 items, numbered by chapter. Appendix practice varies enormously across the five examined works; see the appendices page for the comparison.
The work has no ethics apparatus at all: no ethics section, no approval statement, nothing. That is a 1999 submission, and the 2016 thesis in the same set handles ethics by embedding it in a participant invitation letter in the appendix. The pair is a seventeen-year before-and-after, and it is evidence about what researchers did, not about what is required of you. The distinction matters and is worked through in Research Ethics as Practised in Published Theses.
Borrowing the structure without borrowing its problems
Whether the phase-chapter architecture fits your project
Before you adopt this structure
- Your phases are distinguished by something substantive, and you can say what in one sentence each
- Your repeated template survives your hardest episode without an empty heading
- Every episode reports what went wrong under a heading of its own
- Managerial and research outcomes are separated everywhere, not only where it is easy
- Your first-person decision is defended — by precedent, by argument, or by your institution's own guidance
- Your ethics approval is documented in the submission, whatever the 1999 exemplar did
- Your appendices have a list, page numbers and one consistent numbering scheme
That last item is not pedantry. The examined work skips a page, mis-prints a chapter start in the contents, and numbers appendices two different ways. It also carries a note explaining that the main text uses one English variant while quoted data is left in the local variant "for accuracy reasons" — a small, good habit of documenting a data-fidelity decision rather than silently normalising the record.
What to carry forward
- When the research cycle is the contribution, let the chapter numbering follow the cycle — four learning phases, no separate findings chapter.
- Name your phases for what changed between them. A progressive handover of control is a claim a reader can check.
- Fix repeated sub-heading templates before drafting. Standing headings for reasons, difficulties and researcher role force honesty that discretion does not.
- Split learning outcomes into managerial and research outcomes. If the research half is empty, the episode was consultancy.
- An emergent review is defensible over three and a half years in the field and indefensible over a twelve-month project.
- The absence of ethics apparatus in a 1999 thesis is evidence about past practice, never a precedent for your own submission.
Frequently asked questions
Can I really write a thesis with no findings chapter?
One examined doctorate in this set does exactly that: each of its four learning phase chapters carries its own data, analysis and conclusions, and chapter eight is conclusions and discussion rather than findings. It works because the phases differ from one another in a way that is itself part of the contribution. If your phases differ only by date, a single findings chapter will serve you better.
What are managerial outcomes and research outcomes?
They are two named sub-headings under Learning Outcomes in the examined work's repeated template. Managerial outcomes report what changed in the organisation; research outcomes report what is now known that was not known before. The split gives the sponsor and the examiner each a place to read, instead of asking one stream of prose to serve both.
Is first person acceptable in a research thesis?
The examined action research thesis is written entirely in the first person and defends the choice in chapter one by citing three earlier action research doctorates as precedent. The supplied teaching source takes no position on person or voice. Check your own institution's guidance, and if you need to argue the case, examined precedent in your own tradition is the more persuasive form of argument.
How many participants does an action research study need?
The supplied source gives no guidance and the examined work offers no justification of its own sample size. Its participation grew across phases from 6 core team members to 26, then 60, then 108, plus 50 on the customer side, because each phase drew more of the organisation in. Those are counts from one study over three and a half years, not a target.
Why does that thesis have no ethics section?
It was submitted in 1999 and contains no ethics section, no approval statement, no consent forms and no ethics certificate in its appendices. The 2016 thesis in the same set embeds ethics language in a participant invitation letter reproduced in an appendix. That contrast is evidence about how practice changed, not a statement of what your institution requires — obtain and document your own approval regardless.
How much of the thesis should be appendices?
There is no answer in the supplied material. In this one submission appendices ran to roughly 140 of 447 pages, close to a third, and held raw field data, researcher-built instruments and organisational artefacts. Across the five examined works appendix practice varies from about a quarter of the document to none at all, so treat the proportion as a consequence of what your evidence is, not as a rule.
References and source attribution
- An insider action research doctoral thesis in an engineering organisation, 1999, 447 pp. — referred to throughout this page as Exemplar A. Structural observations from front matter, contents, methodology chapter, abstract and appendix listing; chapter bodies were not reproduced. Author, supervisor and awarding institution withheld.
- Clarke, R. J. 2005, Research methods and methodologies, pp. 35–45 — the work the supplied teaching source adapts for its one-sentence definition of action research.
- O'Leary, Z. 2017, The Essential Guide to Doing Your Research Project, 3rd ed., Sage Publications, London.
- Bryman, A. 2016, Social Research Methods, 5th ed., Oxford University Press, Oxford.
- The supplied teaching source: consolidated weekly teaching notes and slide material on research methods, the research process and reviewing the literature, used here for the taught conventions this thesis departs from.
Suggested questions for Ask KEVOS
- Draft a repeated sub-heading template for the case studies in my action research write-up.
- Split these learning outcomes into managerial outcomes and research outcomes for me.
- Are my project's phases distinct enough to justify a chapter each?
- Help me defend writing my thesis in the first person to my supervisor.
- What did the 1999 action research thesis put in its 140 pages of appendices?
- Write the 'My Role as a Researcher' section for a case where I was also the project sponsor.