The Examined Thesis Quick Reference
One page to hold against your own work: the ranges eight examined master's projects actually occupy, what each one uniquely supplies, what checking their figures found, and what none of them contains.
How to read the marks
Claim classes used throughout this page
- Requirement
- Stated as a rule by a source — a committee's form, an institutional guidance document, a national code. Binding on the project it applies to, and on nothing else.
- Example
- One work's own choice: a sample size, a window, a cut-off year. Never a standard, a target or a benchmark, however reasonable it looks.
- Quoted
- A number taken by a work from a cited external source and reproduced. It is that source's number, not the work's finding.
- Derived
- Counted or computed by this library from the documents, where the work itself states no total.
- Verified
- Recomputed from primary material — a raw file, a printed dataset, an embedded worksheet — and confirmed against what the document printed.
The eight works, and what each one uniquely supplies
THE CORPUS AT A GLANCE (n values verified against each work; contributions derived)
| Work | Domain | Design | n | What only this work supplies |
|---|---|---|---|---|
| F | Stakeholder engagement with a community, in anticipation of civil works | Narrative review + one focus group, thematic against four pre-set themes | 4 in one group | A complete focus-group tool with branch logic; a positionality statement converted into a design control; the best methodological sentence in the corpus |
| G | Production pressure and workforce safety in construction | One web questionnaire; the work's own literature review designated its qualitative strand | 100 | Seventy-five verbatim free-text accounts of unsafe work; an eleven-element consent text; a constructed safety index |
| H | Change-management models inside project management | Literature-based study; no search reported | none | A four-source audit of a widely quoted failure rate — the strongest analytical work in the corpus |
| J | Risk management in event projects | Online survey + two in-depth interviews | 46 + 2 | Both instruments displayed; the cleanest objective-to-structure tracking; one survey reported twice with eleven disagreeing quantities |
| K | Agile adoption in a defence organisation's ICT projects | Systematic quantitative review + questionnaire | 17 people; 348 records screened, 33 retained | A review flow that reconciles; twelve search strings with counts; eleven criteria with six written justifications; a worked sensitivity analysis |
| L | Lessons-learned frameworks across two repeat maintenance projects | Two-case comparison: interviews + project metrics | 10 people, 2 projects | Chart data recoverable from inside the file; the fullest consent procedure; two insider-research controls; a stated blinding procedure |
| M | Scope creep in a defence industry | Hybrid Delphi: interviews, then survey | 5 + 16 | The complete raw dataset printed (20 × 16); a target n stated in advance; attributed anonymous quotations; three reported disconfirmations of its own first round |
| The dataset thesis | Change management from an entrepreneurial perspective | Online survey + secondary reading | 50 | The whole apparatus: instrument as issued and as a document, participant information sheet, raw file, ten-sheet workbook, project schedule |
F, G, H and J were examined together; K, L and M together; the dataset thesis separately with its apparatus. Seven of the eight share one section template.
Scale and shape
PROPORTIONS AND APPARATUS (each figure is that work's own — example, never a target)
| Work | Words | Lit. review | Method | Limitations | Reference entries | Uncited |
|---|---|---|---|---|---|---|
| F | 11,775 | 18.1% | 5.9% | 364 words (3.1%), unheaded, in the conclusions | 25 + 5 in a separate bibliography | 0% |
| G | 14,081 | 17.9% | 8.1% | none anywhere | 66 | 6.1% |
| H | 10,576 | 22.7% | 4.7% | 210 words (2.0%), own heading | 50 printed, 49 distinct | 10.2% |
| J | 10,783 | 13.9% | 9.2% | 151 words (1.4%), own heading | 42 | 7.1% |
| K | 10,567 | 22.4% | 2.6% | 295 words (2.8%), own heading | 70 | 45.7% |
| L | 9,004 | 14.3% | 14.0% | 369 words in the introduction + ~200 in the conclusion, no heading | 23 | 17.4% |
| M | 8,804 | 27.9% | 5.4% | 105 words (1.2%), own heading | 18 | 5.6% |
| The dataset thesis | 11,741 across 62 pages, 64 headings | Not measured on this basis | Not measured on this basis | one bullet, and it is external | 64 | 22% |
Combined ranges across the seven on the template: literature review 13.9–27.9%; method 2.6–14.0%; the two sections together 19.8–33.3%. K's method figure depends on where the method sits — counting its search process, criteria and validation as method moves it from 2.6% to 17.8%. Any comparison must say which figure it is using.
The template, and where it breaks
- Introduction
- Research background
- Aims and objectives
- Research Boundaries
- Literature review
- Research method
- Discussion
- Conclusions and further research
HOW EACH WORK CARRIES THE EIGHT HEADINGS (derived, by heading-by-heading comparison)
| Work | Headings carried | Order | Notable |
|---|---|---|---|
| F | On the template | As printed | No results heading at all — the findings cannot be located from the contents page. Limitations, further research and recommendations are visual lines, not headings |
| G | On the template | As printed | A chapter literally headed "Main chapters" — the template's own placeholder, surviving into a passed work |
| H | On the template | As printed | Method chapter renamed "Research plan"; a second work also prints "Main chapters" |
| J | On the template | As printed | Real sub-headings for limitations and further research, in the contents page |
| K | 8 of 8 | Method before literature review | Method chapter of 280 words; the real method is distributed across three chapters |
| L | 6 of 8 | As printed | No Research Boundaries — a limitations section stands in its place; no Discussion heading; a hand-typed contents page that disagrees with the body in nine places |
| M | 8 of 8 | Method before literature review | Chapter 3 is seven words and one figure — and the figure is a full project schedule |
| The dataset thesis | Not on this template | Six numbered chapters | 64 headings, of which 16 are figure captions styled as headings; a Recommendations chapter with no number and no sub-headings |
Where the order breaks it breaks meaningfully: both works that put method before literature review are works in which the literature is the data rather than the background.
Aims, objectives, questions and hypotheses
WHAT EACH WORK STATES, AND WHAT IT THEN DOES WITH IT (counts verified against each document)
| Work | Aims | Objectives | Questions | Hypotheses | The instructive feature |
|---|---|---|---|---|---|
| F | 1 | 3 | 1 | 0 | The aim names a deliverable — develop a checklist — so it is testable: one exists. One of three objectives is a task, not a knowledge outcome |
| G | 2 | 4 (count stated by the work) | 0 | 2 | States no research question and then refers five times to "the research questions", including two numbered ones it never wrote |
| H | 1 | 3 | 0 | 0 | Four rhetorical questions in the background, none labelled, none restated. Tightest objective-to-conclusion tracking of its group |
| J | 1 | 4 | 0 | 0 | The aim is a conclusion — "aims to demonstrate that" — and objectives 1–3 become the discussion's sub-headings verbatim |
| K | 1, stated in the Research Boundaries section | 0 | 4, printed four times in the document | 0 | Refers to "the thesis hypothesis" four times; there is no hypothesis in the work |
| L | Aims and objectives section present; counts not established | Not defined in source | 0 | 0 | Design named four ways in four places — qualitative case study in one section, mixed method seven paragraphs later |
| M | 1 | 4 | 0 | 0 | All four objectives restated verbatim at the head of the conclusions and answered one by one, in order — the cleanest objective-to-conclusion tracking in the corpus |
| The dataset thesis | 1 | 3 | 1 (two clauses) | 2 | States all four kinds against guidance that offers them as a four-way choice. Neither hypothesis is ever tested |
Instruments, scales and their recorded defects
EVERY INSTRUMENT IN THE CORPUS (structure verified against the displayed instrument where one exists)
| Work | Instrument | Structure | Displayed? | Recorded defects |
|---|---|---|---|---|
| F | Focus-group facilitation tool | 9 printed prompts; 3 branch instructions; ground rules; consent-to-record language; participants asked how they wish to be acknowledged | Yes, in full | The work reports "four main questions" and prints nine, and does not reconcile the two counts |
| G | Web questionnaire | 15 numbered items; item 14 is a 16-statement Likert matrix × 5 points; item 15 free text. The work's claim of "30 questions" reconciles exactly — 13 + 16 + 1 — by a counting convention it never declares | Yes | 9 of 16 Likert statements double-barrelled or containing an absolute; item 13 conditioned on the wrong item; two negatively worded statements with no reverse-scoring stated; no "not applicable" exit |
| H | None | — | n/a — correct for the design | The single appendix is a promised glossary containing no glossary |
| J | Online survey + interview schedule | 13 survey items; 23 interview prompts | Both | Events-per-year bands overlap at 4 of 5 boundaries; the reported bands are a repair made after collection and undeclared; one item of thirteen is never reported; both appendices carry an earlier thesis title |
| K | Search protocol + questionnaire | 12 search strings with counts; 5 inclusion and 6 exclusion criteria; questionnaire structure not recoverable | Protocol yes; questionnaire no | The boundaries narrow "Agile" to one framework; the questionnaire measures three different objects under that term |
| L | Interview schedule | 11 questions, three conditional, no branch logic; phone; ~30 minutes; semi-formal with a stated digression rule | Yes | Experience bands overlap at 3 of 4 boundaries; no coding procedure, frame, theme list or coder count anywhere |
| M | Interview schedule + survey | 12 interview questions, items 4–8 being the five literature themes each ending "Why?"; 20 survey items | Both | Items 4–8 are leading — each names a cause and asks whether it is one; no dispersion statistic reported for any item |
| The dataset thesis | Online survey form | 8 questions, 8 of 8 required, single-choice throughout; 37 response options; 0 free-text; 0 opt-outs; consent by a header sentence with no tick box | Yes — as issued and as a document, and the two diverge in five places | Q5's five-point scale is non-monotonic; bands overlap on Q3 and Q4; the document version invites a skip the form makes impossible; no pilot recorded |
Five of the eight display at least one instrument; three display two; one displays none. Every defect in this column is visible only because the instrument was printed.
Figures, data and what checking them found
THE AUDITABLE WORKS AND THEIR RESULTS (all outcomes verified by recomputation from primary material)
| Work | What could be checked | Outcome |
|---|---|---|
| The dataset thesis | Six figures traced end to end: instrument → raw file → tabulation → chart → sentence | Five survive intact. One fails: 36% where the count gives 30%, repeated in a second chapter, carrying a derived total of 84% where 78% is correct — and the chart printed beside the prose is labelled 30%. Separately, four age counts are exactly right and three of four category labels read the same band |
| M | The complete printed dataset, 20 questions × 16 respondents, 320 cells with 4 marked not applicable | 20 of 20 means recompute exactly. 17 of 20 modes recompute; the other three are ties a spreadsheet's tie-breaking rule resolved silently, one of them on a perfectly flat distribution with no mode at all. Of 18 checkable claims, 12 are fully correct, 2 correct on the mean and wrong on the mode, and 4 wrong |
| L | Ten of twelve figures are live charts carrying their own values; one embedded worksheet survives | Three percentage claims verify exactly, one of them confirming the sample size twice over. The headline safety claim matches none of six candidate computations. The cost section disagrees with itself four ways. One response chart sums to 11 against a sample of 10. One chart from which a two-part finding is drawn contains no plotted data |
| K | The review flow, stage by stage | It reconciles: 27 + 321 = 348 screened; 27 + 288 = 315 excluded; 33 retained. The only reconciling review flow in this material — and the work then breaks four of its own six exclusion criteria in its own body, including excluding the founding document of its own topic, which is its most-cited source |
| G | Prose against prose; the printed free-text appendix against the reported rate | One questionnaire item reported at 87% and 99% one page apart, the second offered as support for the first. The headline quantity reported at 75%, 42% and roughly 40% in three consecutive chapters. The work reports 100 respondents; the appendix cited in support prints 75 rows |
| J | One survey reported twice, in two sections | Eleven quantities disagree. One pair reverses the finding: not-reviewed outnumbers always-reviewed two to one in the first account, and always-reviewed outnumbers not-reviewed two to one in the second |
F and H are not in this table. F's raw material is a 24-or-25-item feedback table printed twice with different contents; H collected no data and its numbers are quoted from cited sources throughout.
Defects by kind, and where each was found
THE RECORDED DEFECT TYPES (each belongs to the document named; no rate across works is implied)
| Kind | Worked instances in this corpus |
|---|---|
| Transcription into prose | A verified percentage retyped wrongly and propagated into a second chapter and a derived total (dataset thesis); four correct counts attached to one wrong label (dataset thesis) |
| One quantity, several values | 87% and 99% for one item; 75%/42%/~40% for one headline (G); eleven disagreements between two accounts of one survey (J); a register counted at 162 and 161 two pages apart (L); three figures for one cost (L) |
| Apparatus without output | A correlation formula, a scoring scheme, an interpretation key, two plots — and no coefficient anywhere (G); thematic analysis named and never operationalised (G); a promised glossary that is empty (H) |
| Unmeasured variable | Hypotheses about a dependent variable that no instrument item measures and that the conceptual framework leaves unconnected (dataset thesis) |
| Instrument band overlap | Events per year, 4 of 5 boundaries (J); experience bands, 3 of 4 (L); enterprise size and experience (dataset thesis) |
| Criteria broken by the work itself | Four of six exclusion criteria broken in the body, including the date bound that excludes the topic's founding document (K) |
| Unresolvable citation | Twenty distinct in-text sources across seven works cannot be reached from the documents citing them; in six of seven at least one is load-bearing — a focus-group size rule, three sources carrying a central finding, a gap-argument citation |
| Reference-list mechanics | Two entries carrying a neighbouring entry's web address (dataset thesis); three entries whose location is a local hard-drive path (K); an entry with no author, an entry with no year, an entry printed twice (H); alphabetical order broken in three works |
| Document production | A hand-typed contents page disagreeing with the body in nine places and listing one section that does not exist (L); appendices carrying an earlier thesis title (J); a corrupted appendix heading (F) |
| Claim exceeding evidence | "Disprove" against evidence that establishes only that four sources do not evidence a rate (H); a headline safety reduction no computation supports (L); a null hypothesis rejected in a sentence that gives the null as its reason (dataset thesis); a recommendation nobody was asked about (J) |
The standing absences
WHAT THIS CORPUS DOES NOT CONTAIN (derived counts across the works assessed on each dimension)
| Absent | Count | Note |
|---|---|---|
| A declared ontology, epistemology or paradigm | 28 examined works, none declaring | The dataset thesis was not assessed on this dimension and is not in the 28. Do not round the figure up. Present this as a property of the training, not of any researcher |
| A named sampling technique | 0 of 7 | The words purposive, convenience, snowball, stratified, random and saturation were searched exhaustively across all seven |
| A reported pilot of any instrument | 0 of 7 | The pilot study is named four times in the teaching material and run by none of these works |
| A participant information sheet | 0 of 7 | The only one in this library is the dataset thesis's, supplied separately with its apparatus |
| Limitations placed in the discussion | 0 of 6 works that have limitations | All sit in the conclusions, so each arrives after the claims it should have bounded. The guidance the works were written under places them in the discussion |
| A computable response rate | 1 in the library | It has a printed numerator and denominator, and the arithmetic printed beside it is wrong. No other work prints a denominator |
| A statement that ethics approval was not required | 0 of 7 | In a document, an absence and an exemption are indistinguishable |
Where to read each of these in full
NAVIGATION — THE PAGES BEHIND EACH BLOCK OF THIS REFERENCE
| If you want | Go to |
|---|---|
| The eight works compared design by design, with claimed design against design used | Eight Examined Theses Compared |
| The section template, its evidence and where it breaks | The Thesis Pro-Forma |
| The complete apparatus project walked end to end | A Complete Research Project, End to End |
| The six traces, link by link | Tracing a Figure Back to Its Source |
| The focus group in full, including what it does not report | Running a Focus Group |
| The systematic review, its flow and its four broken criteria | A Worked Systematic Literature Review |
| The Delphi rounds and the printed dataset | The Delphi Method in Practice |
| The two-project comparison and its recovered chart data | Case Study Design in Practice |
| The instruments compared across four works | Survey Design Across Four Examined Theses |
| One survey reported twice with different numbers | When a Thesis Disagrees With Itself |
| The correlation apparatus that reports no coefficient | Reporting a Statistic You Did Not Compute |
| Conclusions that travel past their findings | Overreach in Conclusions |
| Where integrity actually fails in passed work | Research Integrity in Examined Work |
| What the whole corpus teaches about writing one | What Eight Theses Teach About Writing One |
What to carry forward
- Use this page as a comparator, not a target. Every number belongs to the study that produced it, and eight works are eight worked examples rather than a population.
- Scale varies by a factor of twenty-five on one template for one degree, so the form your work is written to tells a reader almost nothing about what kind of work it contains. Say what kind it is.
- The method chapter is the smallest analytical chapter in almost all of these works, and the standing absences — sampling technique, pilot, coding procedure, response-rate denominator — sit inside it.
- The defects that could be found are the defects in works that printed enough to be checked. Publishing your instrument makes your design auditable; publishing your data makes your analysis auditable.
- Label the values on your charts. In the one project traceable end to end, the only prose errors are on the four figures that carry no data labels.
- Cite 28 examined works declaring no paradigm, and say the ninth was not assessed on that dimension.
Frequently asked questions
Can I use the sample sizes on this page to justify my own?
No. Each n is that study's realised sample, not a target or a minimum, and only one of the seven works on the template states a target in advance at all. Use them to see the range a passed master's project can occupy — four people to one hundred — and then justify your own n against your design and your population.
Why is one work's uncited reference rate 45.7% and another's zero?
Because the metric measures different things across designs. In a systematic review, a reference list may be a corpus rather than a citation list, and entries listed but not quoted are a genre property rather than a fault. In a conventional thesis the same number would mean something else entirely. Always say which kind of document you are measuring.
Should I follow this eight-heading template?
Only if your design fits it. Two of the works that carry all eight headings had to invert the order to put method before literature review, because their literature was their data. One work carries six of eight. The template is a container; decide the design first and say where the container does not hold it.
What does the paradigm count of 28 mean?
That across twenty-eight examined or published works assessed on this dimension, none declares an ontology, an epistemology or a named paradigm. The ninth work in this corpus was not assessed on it, so the figure is 28, not 29. It is best read as a property of the training these works came out of rather than as a criticism of any researcher.
Why do the defect lists differ so much between works?
Largely because of what each one published. The longest lists belong to the works that printed a raw dataset, left chart data inside the file, or supplied a complete apparatus. A study that shows nothing cannot be shown to have miscounted, which is an argument for showing more, not less.
Which single check does this page most support?
Trace every number in your discussion and conclusions back to a tabulation. Two of the four figure failures recorded here — and both failures in the one project traceable end to end — occur at the moment a verified number is retyped into a sentence.
References and source attribution
- Eight examined master's works in project management, supplied as student work and used here as observed practice rather than as model answers. Researchers, supervisors, institutions, employers, jurisdictions, industries, communities and programmes scrubbed. Seven are written to one section template; the eighth was supplied with its complete research apparatus.
- The consolidated extract of four of those works — a focus-group study, a construction-safety questionnaire study, a change-management literature study and an event risk survey — with every heading measured, every reference list audited entry by entry and every quantitative claim checked against the printed evidence.
- The consolidated extract of three further works — a systematic quantitative literature review, a two-project case study and a hybrid Delphi study — including a complete raw dataset of twenty questions by sixteen respondents recomputed mean by mean and mode by mode, and chart values recovered from inside a delivered file.
- The complete research apparatus of the eighth project: survey instrument as issued and as a document, participant information sheet, raw response file, ten-sheet analysis workbook with eight pivot tables and eight charts, and a project schedule, from which six figures were traced end to end.
- The supplied teaching source: weekly study notes, slide decks and assessment activities for a master's-level research methods subject in project management, which places limitations in the discussion, names the pilot study four times, and teaches no focus-group, paradigm or systematic-review content. Author, institution and year not stated in the supplied files.
Suggested questions for Ask KEVOS
- How large a sample do examined master's projects in project management actually use?
- What proportion of a thesis is usually the literature review and the method chapter?
- Which section template do these examined theses share, and where does it break?
- What kinds of defect are most common in passed master's research documents?
- Why do more thorough theses appear to have more defects?
- What should I display in my appendices so my work can be checked?