What Goes Wrong in a Thesis Schedule
Every fault on this list is visible on the printed chart alone, without knowing anything about the study behind it. Several are artefacts of how the tool was driven rather than faults in the thinking — and one of the nine may not be a fault at all.
How to read a defect list about somebody else's plan
The artefact is one student's schedule for one thesis, transcribed in full at a worked thesis schedule. It is anonymised student work supplied alongside the teaching material, and no week of the subject teaches scheduling, names a tool, defines a dependency or supplies a plan of its own.
The nine defects, and how each one shows
THE NINE, EACH DEMONSTRABLE FROM THE CHART
| # | The defect | How it shows on the chart | Kind |
|---|---|---|---|
| 1 | Children finish after their parent | Synthesize finishes 63 days after the Literature Review it sits under | Tool use |
| 2 | Durations do not match their date ranges | Introduction: 10 days stated across an elapsed span of 25 | Tool use |
| 3 | Every estimated duration is flagged and never resolved | 17 of the 24 tasks still carry the estimate marker, as does the summary line | Tool use |
| 4 | Tasks start and finish on weekends | Six starts and several finishes fall on a Saturday or a Sunday | Tool use |
| 5 | A five-month gap with no task in it | Nothing scheduled between 8 June and 14 November 2020 | Unexplained |
| 6 | Analysis starts before collection finishes | Data Analysis starts 5/12/20; Data Collection finishes 12/12/20 | May be deliberate |
| 7 | Coding starts inside analysis and is shorter than it | Coding: 5 days, starting three weeks into a 15-day Data Analysis | Sequencing |
| 8 | Nothing is allocated to re-writing, editing or polishing | No editing task exists anywhere on the plan | Omission |
| 9 | List order and date order disagree | Title Page is listed second and scheduled first; the abstract is listed 22nd and scheduled near the end | Presentation |
Kind is this library's classification. The source records the nine as defects and separately notes that several are artefacts of tool use.
Defects 1 to 3: the tool was not driving the plan
These three look like separate problems and are one problem. In a scheduling application, a task's dates are meant to be produced — from a duration, a calendar and a set of links to other tasks. A summary line's dates are meant to be produced too: they span the earliest start and latest finish of everything beneath them. When you type dates and durations in as free text, none of that machinery engages, and the chart becomes a drawing of a plan rather than a model of one.
THE ARITHMETIC, WORKED FROM THE TRANSCRIBED TABLE
| Task | Stated duration | Elapsed span of its own dates | Difference |
|---|---|---|---|
| Introduction | 10 days | 2/03/20 to 26/03/20 — 25 days | 15 days |
| Synthesize Stream 1 & Stream 2 | 45 days | 27/03/20 to 27/07/20 — 122 days | 77 days |
| Defining Gaps | 10 days | 20/04/20 to 5/06/20 — 47 days | 37 days |
| Research Sources and Methods | 15 days | 15/11/20 to 3/12/20 — 19 days | 4 days |
Spans are inclusive of both end dates. These four are the instances the source names; the same arithmetic can be run on other lines of the chart.
The roll-up failure is the same fault seen from above. Synthesize finishes 27/07/20 while the Literature Review it sits beneath finishes 25/05/20 — the child ends 63 days after its parent. Defining Gaps also ends after that parent. Data Collection finishes 12/12/20, nine days after the Research Sources and Methods line it sits under. A summary task cannot legitimately finish before its children; that these do means the summary durations were typed rather than calculated.
Defect 3 is the mildest of the three and the easiest to fix. Seventeen of the twenty-four tasks — and the project summary line itself — still carry the scheduling tool's estimated-duration marker, a trailing question mark the application adds by default and expects you to clear as an estimate firms up. None was cleared. That is not wrong in the way a broken roll-up is wrong, but it means the plan never distinguishes the durations its author was confident about from the ones they were guessing at, which is precisely the distinction the marker exists to record.
The calendar assumption, and what it costs
Defect 4 is the one with the largest arithmetic consequence and the smallest visible footprint. Task 8 starts on Sunday 3/05/20. Task 9 starts on Sunday 17/05/20. Tasks 11 and 12 both start on Sunday 15/11/20. Task 13 starts on Saturday 5/12/20. Task 14 starts on Sunday 27/12/20. The Initiation line finishes on Sunday 15/03/20. Others across the chart finish on weekend days too.
The default working calendar was not applied, so the plan runs on seven-day weeks. That inflates apparent capacity by about 40% — five working days become seven — and it does so invisibly, because nothing on the printed chart announces it. A reader glancing at a 51-day literature review will read it as roughly ten working weeks. On a seven-day calendar it is a little over seven.
The gap in the middle, which the source does not explain
Ethical Approval finishes on 7/06/20. The next task, Research Sources and Methods, starts on 15/11/20. Between 8 June and 14 November 2020 — one hundred and sixty days — the chart carries no task at all, in a plan that runs from 2 March 2020 to 19 February 2021.
One caution on how the gap gets read. The same source, comparing the schedule against the subject's teaching, records the schedule as respecting the rule that ethics clearance precedes data collection and describes it as five months of clearance ahead of collection. That is one available reading of the same 160 days — clearance lead time deliberately built in. It is not reconciled anywhere with the finding that the gap is unexplained, and both statements stand in the record. Treat the interval as an unlabelled void that may or may not be intentional, and if it is your own chart, label it.
The overlap that may be good practice
Data Analysis starts on 5/12/20. Data Collection finishes on 12/12/20. Analysis therefore begins seven days before collection ends, and Coding starts on 27/12/20, inside the analysis window.
This matters more than the seven days involved. A reader who takes the overlap as a mistake learns the wrong lesson and goes on to build a strictly sequential plan, which is the pattern the checklist explicitly warns against. The library's treatment of the relationship is at the data collection and analysis sequence, and the checklist itself at the do's and don'ts of social research.
Defect 7 is about proportion, not overlap. Coding is given five days and starts roughly three weeks into a fifteen-day Data Analysis task, which makes it a short subtask inside analysis. The subject's qualitative material treats coding as the mechanism by which qualitative analysis happens — the thing analysis largely is, rather than a step within it. The plan never states what kind of study it is, so whether this is a mis-sequencing or a quantitative study's different use of the word cannot be settled. See coding qualitative data in practice.
No time at all for editing
This is the largest defect on the list and the one least dependent on how the tool was driven. The subject's own writing-up material names three stages — first draft, then re-writing and editing, then polishing into a final draft — and says of the last stage that you should take as much time as is needed. That is the only guidance on editing duration anywhere in fourteen weeks, and it is unquantified.
The schedule contains no editing task at all. Its final tasks are Contents page(s), Summary/abstract and Cover, which are assembly, not revision. Nothing on the chart produces a second draft from a first. The stage the material calls the most demanding part of producing a finished report is the one stage with no line on the plan. The stages are set out at the challenge of writing up.
Defect 9 is smaller and worth stating precisely because it is easy to get backwards. Summary/abstract finishes on 11/02/21, one day before Cover, which puts the abstract second-to-last in date order. That respects the subject's rule that the abstract is written last, because it summarises everything the rest of the document contains — see writing an abstract. What disagrees is the plan's list order: Title Page is task 2 in the list and is scheduled first, so a reader scanning the task names top to bottom and a reader scanning the dates left to right get different accounts of the same plan.
The schedule against the subject's own teaching, and what to check on yours
SEVEN INSTRUCTIONS, AND WHAT THE PLAN DID WITH THEM
| The subject says | The schedule does |
|---|---|
| A proposed timeline for the research is required | Supplies one — the only worked schedule in the dataset |
| Take as much time as is needed for editing | Allocates none |
| Write the abstract last | Schedules it second-to-last by date; lists Title Page first |
| The proposal needs a trial table of contents of one or two pages | Schedules Contents page(s) at 5 days |
| Ethics clearance precedes data collection | Respected — approval is a discrete task well ahead of collection |
| Create a timetable to avoid last-minute panics | This is that timetable |
| Recommendations are a closing step of the report | No recommendations task exists |
The final row matters most for a project audience: recommendations are usually the part of a research report a commissioning organisation reads first.
Run these against your own schedule
- No task starts or finishes on a non-working day, or the calendar is deliberately seven-day and says so
- Every summary line's dates are calculated from its children, and none finishes before a child does
- Every stated duration is consistent with its own start and finish dates
- Estimate flags have been cleared on everything you are now confident about, and remain on everything you are not
- Every gap longer than a fortnight carries a named line explaining it
- Any overlap between collection and analysis is annotated as deliberate, because it probably should be
- There is a re-writing and editing task, and it is not the smallest thing on the chart
- There is a recommendations task if your audience is an organisation rather than an examiner
- The last content task finishes far enough before submission to absorb one bad draft
The same page, run against a second schedule
Every fault on this page's list came from one project file. A second has arrived, from a different project, and it fails almost none of them — which is the most useful thing that could have happened to this page, because it separates the faults that are artefacts of driving the tool from the faults that are about research planning.
THE TWO SCHEDULES, SIDE BY SIDE
| The first schedule | The second schedule | |
|---|---|---|
| Rows | 25 | 29 |
| Working calendar | None applied — weekend starts and finishes | Five-day, applied consistently |
| Durations against date ranges | Disagree on at least four tasks | Match on 22 of 23 |
| Summary roll-up | Broken — children finish after parents | Correct in all eight cases |
| Milestones | None | Four zero-day milestones |
| WBS numbering | None | Full three levels |
| Uncleared estimated-duration markers | 17 of 24 | 1, on the project summary |
| Unexplained gap | 160 days with no task | None |
| Ethics task | Present, before data collection | Absent — among 29 rows |
| Editing or revision task | Absent | Absent |
| Progress tracked | Not recorded | 0% throughout, after seven saves |
Two schedules, two projects. The first was recorded on this page as carrying nine demonstrable defects; three of those nine — children outliving parents, durations not matching date ranges, weekend working — do not occur in the second at all.
What to carry forward
- A second schedule fails almost none of this page's original nine faults — so five of them were about driving the tool. What both plans omit is revision time, and the second also omits ethics approval and fieldwork entirely.
- Three of the nine defects are one defect: the durations were typed, so nothing rolls up, nothing recalculates, and the chart is a picture rather than a model.
- An unset working calendar is invisible on a printed chart and overstates capacity by about 40%. Weekend starts are the tell.
- The 160-day gap in the middle is unexplained. Report gaps like this as gaps; label your own before somebody else reads them as an oversight.
- The collection-and-analysis overlap may be deliberate and, if so, is exactly what the subject's checklist asks for. Annotate deliberate overlaps so nobody has to guess.
- The biggest fault is an omission, not an error: no editing task, and no room before submission for a rewrite that the material itself says is always needed.
Frequently asked questions
Which of the nine defects should I worry about most on my own plan?
The missing editing task. The tool-use defects make a chart unreliable, but they are fixable in an afternoon by setting a calendar and linking tasks. A plan with no capacity to rewrite a poor first draft is a structural problem you can only fix by changing the dates, and by then the dates are usually fixed for you.
How can I tell whether a schedule is calculated or just drawn?
Change one duration and see what moves. In a calculated schedule, lengthening a task pushes everything downstream and updates the summary line and the project finish date. In a drawn one, nothing moves. Two other tells are summary lines that finish before their children and tasks that start on weekends.
Is it wrong to start analysing data before collection is finished?
No — the subject's own final checklist tells you to start analysing as soon as you collect, and not to assume you must collect everything first. The problem with this chart is not the seven-day overlap but that nothing on it distinguishes a deliberate overlap from an accidental one. If yours is deliberate, annotate it.
What explains the five-month gap in the middle of the plan?
Nothing in the source explains it. It could be a candidature break, a part-time study pattern, or a period when the data simply was not available — the supplied material does not say and this library does not guess. The practical lesson is to label such intervals on your own chart, because an unlabelled gap reads as an oversight.
How much time should a research schedule allow for editing?
The supplied material gives no figure. Its only statement is that you should take as much time as is needed to produce the final draft, which is unquantified guidance in a subject that elsewhere requires a timetable. What the material does say clearly is that a great deal of work goes into producing a final draft from a first, so a plan with no revision capacity is planning for an outcome the material says does not happen.
Are these faults the student's fault?
They are faults in the artefact. The schedule was produced by a learner in a subject that requires a timeline and teaches no scheduling at all — no tool, no dependency, no working calendar, no distinction between duration and elapsed time. Several of the nine are what an unconfigured scheduling application produces from perfectly sound intentions.
References and source attribution
- An anonymised student thesis schedule supplied with the teaching material: a commercial scheduling application's project file and its printed Gantt chart, twenty-five lines, dated Sat 8/02/20. Analysed via the printed chart.
- A methods text's student-researcher checklist of do's and don'ts, reproduced by the teaching source from a publisher's website with a copyright line dated 2012, and paraphrased here.
- The supplied teaching source: weekly study notes, slide decks and assessment activities for a master's-level research methods subject in project management. Author, institution and year not stated in the supplied files.
Suggested questions for Ask KEVOS
- How do I tell whether a Gantt chart is calculated or just drawn?
- How much float should a research schedule leave before submission?
- Should data analysis overlap data collection?
- What does an unset working calendar do to a project plan?
- How do I review somebody else's research schedule?