Chapter 7 — Thesis synthesis and reproducibility archive
Lesson map
Lesson 07 published. Map, example, EX039.
flowchart TD
N0["Learning outcomes"]
N1["7.1 The argument"]
N2["7.2 Reproducibility archive"]
N3["Lab — Manifest digests for the archive"]
N4["7.3 Confidentiality"]
N5["7.4 Coherence checks"]
N6["Synthesis"]
N0 --> N1 --> N2 --> N3 --> N4 --> N5 --> N6
N6 --> EX["Worked example"]
EX --> Q["Exercise EX039"]
Logical drawing
flowchart LR
N0["Learning outcomes"]
N1["7.1 The argument"]
N2["7.2 Reproducibility archive"]
N3["Lab — Manifest digests for the archive"]
N4["7.3 Confidentiality"]
N5["7.4 Coherence checks"]
N6["Synthesis"]
N0 --> N1 --> N2 --> N3 --> N4 --> N5 --> N6 --> Q["EX039"]
[Learning outcomes]
|
v
[7.1 The argument]
|
v
[7.2 Reproducibility archive]
|
v
[Lab — Manifest digests for the archive]
|
v
[7.3 Confidentiality]
|
v
[7.4 Coherence checks]
|
v
[Synthesis]
|
v
[Worked example] --> [Exercise EX039]
Learning outcomes
You will (1) synthesise findings into a coherent thesis argument; (2) build a reproducibility archive with integrity checks; (3) protect partner confidentiality in shared materials.
7.1 The argument
Amara's thesis argues that in an operating health board, calibrated quantitative assessment produced slightly better calibration, more consistent priorities and fewer unresolved ties than an ordinal matrix, at a training cost of ninety minutes per assessor, while board members needed explanation to use the outputs well. The advantage was modest and uncertain, and the partial replication supports its direction. She positions this against analytical critiques of matrices, showing that the predicted problems appear in practice but at smaller scale than critics sometimes imply.
7.2 Reproducibility archive
Amara builds an archive containing the pre-registered protocol, the synthetic scenario set, the analysis code with fixed random seeds and the codebook for bands and ranges. She computes a cryptographic digest for each file and lists the digests in a manifest, so that anyone can confirm the files have not changed since submission. The lab produces the manifest for four files. Reproducibility lets examiners and future researchers rerun her analyses and test alternative assumptions, which is essential when results are modest and open to interpretation.
Lab — Manifest digests for the archive
# Manifest digests for a reproducibility archive (synthetic file contents).
import hashlib
files = {"protocol_v3.md": "pre-registered protocol", "scenarios.csv": "id,prob,loss\n", "lab_mc.py": "seed=5", "codebook.md": "bands"}
for n, c in sorted(files.items()): print(n, hashlib.sha256(c.encode()).hexdigest()[:12])
print("files:", len(files))
Run with python3 lab_archive.py. Verified output:
codebook.md 1a5db2aba54d
lab_mc.py 21df5d9b21fc
protocol_v3.md 0f7bd61c2ae0
scenarios.csv ed97f09e7efe
files: 4
7.3 Confidentiality
The archive contains no real Corvane risks, vulnerabilities or names. Scenarios are synthetic composites agreed with Corvane's information governance committee, and assessor responses are released only as aggregated tables. Amara documents the de-identification steps and has the committee approve the archive before deposit. She notes that a GRC study can leak sensitive information through context even when names are removed, such as a description of an unusual system, so she reviews each scenario for identifying details. Any doubtful scenario is rewritten or removed before deposit.
7.4 Coherence checks
Before submission, Amara checks coherence: every claim in the conclusion is traceable to a chapter and to evidence in the archive; limits stated in the results appear in the abstract; and terminology matches the standards she cites, such as using ISO 31000:2018 terms for risk criteria and treatment. She asks a fellow candidate to trace three claims at random. Both find the trail intact. Prof. Delgado approves submission, noting that a thesis whose claims can be traced and rerun earns trust beyond its numbers.
Worked example
Amara states a modest, traceable thesis argument, archives protocol, data and code with a digest manifest and protects partner confidentiality.
Synthetic case. Archive contents are synthetic; ISO 31000:2018 is real.
Errors and limits
Digests prove integrity, not correctness. Context can still identify partners.
Exercises
- EX039 — Training cost. What training cost does the thesis report? Rubric: 3 = ninety minutes per assessor;; 2 = 90 minutes;; 1 = vague;; 0 = none.
- EX040 — Archive contents. Name the four archived files. Rubric: 3 = protocol, scenarios, analysis code and codebook;; 2 = four;; 1 = vague;; 0 = none.
- EX041 — Digest purpose. What do the manifest digests let others confirm? Rubric: 3 = files have not changed since submission;; 2 = integrity;; 1 = correctness;; 0 = none.
- EX042 — Context leak. How can a GRC study leak information without names? Rubric: 3 = through identifying context such as unusual systems;; 2 = context;; 1 = vague;; 0 = none.
- EX043 — Terminology source. Which standard's terms does Amara use for risk criteria and treatment? Rubric: 3 = ISO 31000:2018;; 2 = ISO 31000;; 1 = COBIT;; 0 = none.
- EX044 — Claim tracing. How did Amara test coherence? Rubric: 3 = a fellow candidate traced three random claims;; 2 = tracing;; 1 = vague;; 0 = none.
Synthesis
A coherent thesis is traceable, reproducible and careful with partner confidentiality.
Continue studying — chapter 07 of 08
← Chapter 06 (draft) · Course chapter index · Chapter 08 (draft) →