## Chapter 1 — Advanced research integrity and methods ### Learning outcomes You will (1) frame a doctoral question about data governance; (2) set integrity rules for working with linked administrative data; (3) plan methods that others can reproduce. ### 1.1 The candidate and the question Nadia Rahman begins a two-year doctoral research track at the fictional Halvorsen University, supervised by Prof. Lars Eriksen. Her partner organisation is the fictional Westmark Statistics Office, which links a business register with employment records to publish statistics on regional jobs. Her question is how governance decisions, such as who sets data quality rules and who chooses record linkage thresholds, change the statistics that ministers and the public read. Khatri and Brown describe governance as deciding who holds decision rights over data, and Nadia wants to measure the consequences of those rights. ### 1.2 Integrity rules Prof. Eriksen asks Nadia to write integrity rules before touching any data. All analysis uses a synthetic replica of Westmark's register built from published distributions, never confidential records. Every result is produced by a script kept under version control, and every figure in the thesis must be traceable to a script run. She will record decisions that change results, such as dropping a field, in a decision log with dates and reasons. Nadia notes that integrity in governance research matters doubly, because the research itself makes claims about the integrity of other people's data. ### 1.3 Methods others can repeat Nadia chooses mixed methods. Quantitative work measures data quality dimensions and linkage error on the synthetic replica. Qualitative work uses interviews with Westmark staff about who decides quality rules and thresholds, conducted with informed consent and anonymised transcripts. She pre-registers her main hypotheses with her supervisor: that unclear decision rights lead to inconsistent thresholds across releases, and that inconsistent thresholds produce measurable shifts in regional employment counts. Pre-registration protects her against reshaping hypotheses after seeing the results. Interview guides are piloted with two retired staff. ### 1.4 Sources and standards Nadia builds her reading list around established sources. DAMA-DMBOK2 describes data governance as the exercise of authority and control over data assets and lists its relationship to other data management functions. ISO/IEC 38505-1 applies the ISO/IEC 38500 governance model to data, placing accountability with the governing body. Wang and Strong offer a consumer-oriented framework of data quality dimensions. She records which claims each source supports, so that her thesis cites sources for definitions and evidence rather than for authority alone. Definitions come first. ### Worked example Nadia frames a question about decision rights and published statistics, writes integrity rules for a synthetic replica, pre-registers hypotheses and maps sources to the claims they support. > **Synthetic case.** Halvorsen, Westmark and all people are synthetic; DAMA-DMBOK2, ISO/IEC 38505-1, Khatri and Brown (2010) and Wang and Strong (1996) are real. ### Errors and limits A synthetic replica cannot reproduce every quirk of real records; state this limit early. ### Exercises - **EX001 — Research question.** What is Nadia's doctoral question? *Rubric:* 3 = how governance decisions change published statistics;; 2 = mentions decisions;; 1 = vague;; 0 = none. - **EX002 — Decision rights.** How do Khatri and Brown frame governance? *Rubric:* 3 = deciding who holds decision rights over data;; 2 = mentions rights;; 1 = vague;; 0 = none. - **EX003 — Replica rule.** What data may Nadia analyse? *Rubric:* 3 = a synthetic replica from published distributions;; 2 = partly correct;; 1 = real;; 0 = none. - **EX004 — Traceability.** How must every figure be traceable? *Rubric:* 3 = to a script run under version control;; 2 = mentions scripts;; 1 = vague;; 0 = none. - **EX005 — Pre-registration.** Why pre-register hypotheses? *Rubric:* 3 = to avoid reshaping them after results;; 2 = mentions bias;; 1 = vague;; 0 = none. - **EX006 — Two hypotheses.** State one pre-registered hypothesis. *Rubric:* 3 = unclear rights cause inconsistent thresholds; thresholds shift counts;; 2 = one;; 1 = vague;; 0 = none. - **EX007 — ISO model.** Where does ISO/IEC 38505-1 place accountability? *Rubric:* 3 = with the governing body;; 2 = partly correct;; 1 = managers;; 0 = none. ### Continue studying — lesson 01 (chapter 01 of 08) [← Course home](../../README.md) · [Course chapter index](../SUBJECTS.md) · [Chapter 02 →](#ch-02) Continue to the next authored lesson. ## Worked answers **EX001.** How decisions about quality rules and linkage thresholds change the statistics people read. **EX002.** As deciding who holds decision rights over data. **EX003.** Only a synthetic replica built from published distributions. **EX004.** Each figure must trace to a version-controlled script run. **EX005.** So hypotheses cannot be reshaped after seeing results. **EX006.** Unclear decision rights lead to inconsistent linkage thresholds across releases. **EX007.** With the organisation's governing body.