The Dissertation Sequence — From Proposal to Defense
The dissertation is the degree. Everything before it — the courses, the apprenticeship, the comprehensive examination — is the run-up. This document is the full map of the dissertation road: every stage, its gate, its deliverable, its advisor, its common failure mode, and its path to publication.
Level: doctoral · Applies after: advancement to candidacy · Supervision: dissertation committee · Credits: DL 990 (dissertation research).
The sequence follows the doctrine's own shape — a pipeline with gates, criteria written before the work begins (doctrine/04). Six stages, each ending in a formal gate owned by the committee:
proposal → literature review → methodology approval → data collection → analysis → defense
│ │ │ (incl. IRB/ethics) │ │ │
└─ gate ─────┘ └─ gate ──────────────────┘ │ │
└── gate ────┘The dissertation in one paragraph
A dissertation in this field is an original, defensible contribution to the empirical base of the pursuit discipline. It takes one open question from the research agenda (modules/doctoral/research-agenda.md) — a calibration question, a factor-validation question, an agency-pattern question, a governance question, or a learning question — and answers it with a rigorous design, real data, and honest analysis. It is not a longer term paper, and it is not a practitioner's handbook. It is a piece of research that a skeptical committee accepts as a contribution and that a peer-reviewed journal will publish.
Stage 1 — Proposal
The work. The proposal is the design of the whole dissertation written down: the research question, the gap it fills, the hypotheses, the design, the measurement plan, the data plan, the ethics plan, and the intended contribution. It is assembled from the DL 901 design, the DL 902 preregistered analysis plan, and the DL 903 seminar design — the three courses were designed so their deliverables assemble into the proposal.
The gate. The committee approves the proposal when it can answer yes to three questions:
- Is this a researchable question? Specific enough to design for, with a falsifiable answer possible.
- Is this a contribution? A question the field needs answered, not already answered in the literature (hence the literature review runs with the proposal).
- Is this feasible? The data are obtainable, the sample is big enough, the ethics are clear, and the student can finish.
Deliverable. The written proposal (typically 30–60 pages: question, literature, design, measurement, sampling, ethics, analysis plan, timeline) plus a proposal defense — a public seminar where the committee and guests interrogate the design.
Common failure modes. A question too big to finish; a design that cannot support the intended claim; a data plan that assumes access the student does not have. The fix for all three is the same: the committee pushes early, and the student narrows.
Committee involvement. The chair is the primary design partner; the methodologist member signs the design's soundness; the domain member signs the contribution's importance.
Stage 2 — Literature review
The work. The literature review runs with the proposal (the proposal cites it, and the review matures as the dissertation proceeds). The student maps what is known across the discipline and its neighbors:
- The practice literature — the capture-management and business-development canon (Shipley, APMP, the proposal-professional bodies): mostly prescriptive, mostly unempirical. The student reads it to know what the field claims.
- The scholarly neighbors — economics of procurement and auctions, public administration and grants management, organizational learning, decision science and judgment/calibration research, small-business policy. This is where the methods and the theories come from.
- The empirical record of this field — thin by design, but growing. The student must know exactly what has been measured, and what has only been asserted.
The gate. The committee approves the review when it demonstrates command — the student can situate their question in the literature, name the closest prior work, and say precisely what their study adds. The review is not a bibliography; it is an argument that the gap is real.
Deliverable. The literature review chapter (or a review article, if the program's publication pathway allows the review to double as a publishable piece).
Common failure modes. The librarian's trap — a comprehensive-but-aimless survey that never argues a gap. The fix is the thesis sentence: "The literature establishes X and Y, but nobody has measured Z — this dissertation measures Z."
Stage 3 — Methodology approval
The work. The design is frozen: the measurement protocol final, the sample plan final, the analysis plan preregistered and dated, the robustness checks listed. The student secures what the design requires:
- Data access. For public data: the extraction plan and provenance log. For organizational data: the agreements, consent, and confidentiality protocols.
- Ethics approval. IRB review where human subjects or identifiable decision-makers are involved — the field's own gate discipline applied to research (doctrine/04).
- Feasibility proof. A pilot where possible — a small test of the measurement protocol (e.g., two coders scoring ten proposals to check inter-rater reliability) so the full study does not fail at scale.
The gate. The committee approves the methodology when the design is specific enough that the student cannot quietly change it later — the analysis plan is preregistered, and the measurement protocol is reproducible from the written definition alone.
Deliverable. The frozen methodology chapter: measurement protocol, sample construction, data-access and ethics approvals, preregistered analysis plan, robustness-checks list.
Common failure modes. The "the data will tell us" trap — an analysis plan so vague that any finding can be post-hoc justified. The fix is the preregistration habit from DL 902: decide what the data must show before seeing it.
Stage 4 — Data collection and IRB/ethics
The work. The evidence is gathered. Two regimes, both governed by the integrity standard from DL 901:
- Public data — award records, solicitations, published review criteria, agency data. No consent needed, but provenance discipline is absolute: every source documented, every extraction logged, every transformation recorded, so another researcher can verify the data is what the dissertation says it is.
- Organizational data — decision logs, win-probability estimates at decision time, budgets, deliberations. Sensitive even when lawful: informed consent, confidentiality, anonymization where feasible, and a clear boundary for a professional studying their own firm.
The ethics non-negotiables. The non-fabrication standard is absolute — fabricated awards, invented figures, or misattributed data fail the degree. The discipline's own research corpora model the rule: documented facts and clearly-labeled illustrative material are kept apart, and invented specifics are never attributed to real organizations.
The gate. The committee approves the data collection when the dataset is built, documented, provenance-registered, and clean — and when the ethics trail (consent, IRB, confidentiality) is complete.
Deliverable. The dataset, its provenance and codebook, the ethics record, and a data-collection chapter (or appendix) describing both.
Common failure modes. Sloppy provenance that collapses under scrutiny; a sample that turns out to be smaller or more biased than planned; a confidentiality breach with organizational data. The fixes are all habits installed before the work starts: log everything, plan for attrition, and treat the data with the seriousness the field gives to a customer's secrets.
Stage 5 — Analysis
The work. The preregistered analysis is executed. The student:
- fits the models and computes the prediction-quality measures the design promised;
- runs the robustness checks the plan listed;
- reports honestly — the null results, the design's limits, the survivorship caveats, the identifying assumptions;
- writes the findings chapters (results, then discussion of what the findings do and do not mean for the doctrine).
The gate. The committee approves the analysis when it answers the question without overclaiming. The tests: Does the analysis follow the preregistered plan? Are the bias traps (selection, survivorship, confounding, clustering, tiny-N instability) addressed or bounded? Would a replication-minded reader be able to reproduce the numbers from the methods?
Deliverable. The results and discussion chapters, the replication appendix, and — if the publication milestone is not yet met — the first article draft.
Common failure modes. The overclaim — a relational finding dressed up as a causal one; the p-hunt — running analyses until one flatters the narrative; the buried null — reporting what worked and hiding what did not. The field's young empirical base will only be trusted if its dissertations refuse all three.
Stage 6 — Defense
The work. The dissertation is assembled into a complete document, rehearsed, and defended in a public defense before the committee (and, where the program uses one, an external examiner from outside the home institution). The defense is not a formality — it is the final gate, and it tests whether the contribution stands on its own merits.
The defense panel's questions. The panel pushes on exactly the places the dissertation must survive:
- What exactly are you claiming — and what are you not claiming? (The scope discipline.)
- What is your identifying assumption, and why is it plausible? (The causal-claims discipline.)
- Where is the survivorship hazard in your data, and what did you do about it? (The field's central trap.)
- Why should a practitioner trust this — and what would make it untrustworthy? (The contribution's usefulness.)
- What is the next study? (The research-agenda posture — a dissertation ends a question and opens the next.)
The gate. The committee awards the degree when the dissertation is an original, defensible contribution and the student has demonstrated command of it under questioning.
Deliverable. The complete dissertation, revised per the defense, and the approved final version.
Advisor roles through the sequence
| Role | Who | Through the sequence |
|---|---|---|
| Chair | The primary intellectual partner | Reads every draft; argues the design; owns the schedule; is the gatekeeper who says the hard thing early. Present at every gate. |
| Methodologist member | A statistics/design expert | Signs the soundness of the measurement and the statistics (Stages 1, 3, 5 especially). |
| Domain member | A capture/grants/procurement expert | Signs the contribution's importance and the doctrine's accuracy. |
| Outside reader (optional) | A scholar from a neighboring field | The defense's insurance against insular thinking; tests whether the contribution stands beyond the discipline's own vocabulary. |
| External examiner (defense) | A scholar outside the home institution | Tests the work on its merits; a fresh skeptical set of eyes. |
| Ethics / IRB liaison | Institutional or program-level | Approves and monitors the consent, confidentiality, and provenance plan (Stage 4). |
The supervision structure is itself a gate discipline: the student and chair agree at each gate on what "done" means before the work, so no expectation surfaces for the first time at the defense.
The publication pathway
A dissertation in this field is written to be publishable, and the program's publication milestone (at least one article submitted to a peer-reviewed venue before the defense) is the bridge from degree to contribution. The typical pathway:
- The literature review becomes a review article — valuable in a young field, and it establishes the student's command publicly.
- The empirical core (one study, cleanly bounded) becomes the flagship journal article — the calibration study, the factor-validation study, the agency-pattern study, or the learning-loop study. The dissertation's single cleanest study is the article.
- A second study (the robustness extension, the cross-agency replication) becomes a second article or a conference paper.
- The methodology contribution — e.g., a validation framework for pipeline verification — becomes a methods paper, valuable precisely because the field has no standards for "how to know a pursuit pipeline works."
Where to publish. The discipline's own venues are the proposal-professional bodies (APMP, NCMA, the capture-management associations) and their journals and conferences. The scholarly neighbors are procurement-and-auctions economics, public administration, grants-management and evaluation journals, and decision science. A strong dissertation targets one of each: the practitioner venue for the field's attention, the scholarly venue for the methods and the evidence.
The replication duty. A published contribution that cannot be reproduced is not a contribution. The dissertation's data and methods must support reproduction — public data published with the article where possible; confidential data accompanied by a documented analysis path that a reviewer could execute.
Milestones and timeline
| Stage | Typical duration | Gate | Gate owner |
|---|---|---|---|
| Proposal | 4–8 months | Proposal approved at proposal defense | Chair + full committee |
| Literature review | runs with the proposal and matures throughout | Review chapter approved | Chair + domain member |
| Methodology approval | 2–4 months | Design frozen; ethics + data access secured | Chair + methodologist |
| Data collection | 6–18 months | Dataset built, clean, provenance-registered | Chair + ethics liaison |
| Analysis | 4–10 months | Preregistered analysis executed, honestly reported | Chair + methodologist |
| Defense | 2–6 months | Degree awarded at public defense | Full committee + external examiner |
Total from candidacy to defense: two to three and a half years, with the timeline dominated by data collection and honest analysis — the two stages that cannot be rushed.
The dissertation in two sentences
The dissertation is the discipline's first rigorous answer to one of its own hardest questions — the calibration of its estimates, the validity of its factors, the reality of its agency patterns, the discipline of its gates, or the truth of its learning loop. The six-stage sequence exists so that answer is designed before it is executed, honest after it is analyzed, and published after it is defended. Test the doctrine so the doctrine can be trusted — the dissertation is where that test becomes a contribution.
Reflection questions for the candidate
- Which stage of the sequence is most likely to be your slow stage — and what will you do in year one to de-risk it?
- What is the single sentence of the dissertation — the claim — that must survive the defense, and have you written it yet?
- Who on your committee will tell you the hard truth early, and have you agreed with them what "done" means at each gate?