OpenCyber Academy
Doctoral research · Chapter 2 · English
Course code: OCA-187Chapter 02: OCA-187-C002

Chapter 2 — Data governance: research frontier

Open, free learning for anyone who wants to protect people and technology. No account is required to read this text or its worked answers.

Lesson map

Lesson 02 published. Map, example, EX008.

flowchart TD
  N0["Learning outcomes"]
  N1["2.1 The state of the field"]
  N2["2.2 Measuring a register"]
  N3["2.3 Rules are decisions"]
  N4["2.4 Positioning the contribution"]
  N5["Lab — Quality dimensions of a synthetic business reg"]
  N0 --> N1 --> N2 --> N3 --> N4 --> N5
  N5 --> EX["Worked example"]
  EX --> Q["Exercise EX008"]

Logical drawing

flowchart LR
  N0["Learning outcomes"]
  N1["2.1 The state of the field"]
  N2["2.2 Measuring a register"]
  N3["2.3 Rules are decisions"]
  N4["2.4 Positioning the contribution"]
  N5["Lab — Quality dimensions of a synthetic business reg"]
  N0 --> N1 --> N2 --> N3 --> N4 --> N5 --> Q["EX008"]

[Learning outcomes]
    |
    v
[2.1 The state of the field]
    |
    v
[2.2 Measuring a register]
    |
    v
[2.3 Rules are decisions]
    |
    v
[2.4 Positioning the contribution]
    |
    v
[Lab — Quality dimensions of a synthetic business reg]
    |
    v
[Worked example] --> [Exercise EX008]

Learning outcomes

You will (1) review current debates in data governance research; (2) measure data quality dimensions on a synthetic register; (3) identify a gap for original work.

2.1 The state of the field

Nadia reviews the literature. Early work treated data quality as accuracy alone; Wang and Strong showed that data consumers also care about timeliness, relevance, accessibility and interpretability. Batini and colleagues compared assessment methodologies and found that most define dimensions, measure them and propose improvements, but differ in which dimensions they choose and how they weigh them. ISO 8000-8 sets out concepts and prerequisites for measuring information and data quality. Nadia notes that measurement methods are well developed, while research on who decides quality rules is thinner.

2.2 Measuring a register

To ground her review, Nadia measures three dimensions on a synthetic register of 5,000 businesses. The lab reports completeness of 0.798, validity of 0.937 and uniqueness of 0.966. Completeness is lowest because many records lack a sector code or postcode. Validity failures come from postcodes with letters and an unknown sector code. Uniqueness falls below one because some registration numbers appear twice. Each figure depends on a rule someone chose, such as whether a blank sector counts as missing or as a legitimate value. Rules matter as much.

2.3 Rules are decisions

Nadia shows Prof. Eriksen how the completeness score changes if blank sectors are allowed for newly registered firms. The number rises, although no record has changed. She argues that quality scores are partly governance outputs: they reflect decisions about rules, thresholds and exceptions. DAMA-DMBOK2 places data quality management within a governance framework for this reason. Nadia's literature map shows few empirical studies that trace how such rule decisions are made and how they shift published figures, which is the gap she will address.

2.4 Positioning the contribution

Nadia writes a positioning statement: existing research measures quality and proposes governance structures, but rarely links the two quantitatively in an official statistics setting. Her contribution will be a method that records decision rights, applies alternative rules on a synthetic replica and measures the effect on outputs. Prof. Eriksen challenges her to show why this matters beyond Westmark. She answers that any organisation publishing indicators from administrative data faces the same question of who decides what counts as good enough data. Westmark is the test case.

Lab — Quality dimensions of a synthetic business register

"""Score completeness, validity and uniqueness of a synthetic business register (stdlib only)."""
import random, re
random.seed(3)
rows = []
for i in range(5000):
    reg = f"BR{i:06d}"
    if random.random() < 0.03: reg = f"BR{random.randint(0, i):06d}"      # duplicate ids
    post = f"{random.randint(1000, 9999)}" if random.random() > 0.08 else ""
    if random.random() < 0.04: post = "ABCD"                                # invalid
    sector = random.choice(["A", "C", "F", "G", "J", "M", ""]) if random.random() > 0.02 else "Z9"
    rows.append((reg, post, sector))
n = len(rows)
complete = sum(1 for r in rows if r[1] and r[2]) / n
valid = sum(1 for r in rows if re.fullmatch(r"\d{4}", r[1] or "0000") and r[2] in {"A","C","F","G","J","M",""}) / n
unique = len({r[0] for r in rows}) / n
print(f"records {n}")
print(f"completeness {complete:.3f}  validity {valid:.3f}  uniqueness {unique:.3f}")

Run with python3 lab_quality.py. Verified output:

records 5000
completeness 0.798  validity 0.937  uniqueness 0.966

Worked example

Nadia reviews quality literature, measures completeness 0.798, validity 0.937 and uniqueness 0.966 on 5,000 synthetic records, shows that rule choices move scores and positions her contribution.

Synthetic case. The register is synthetic; Wang and Strong (1996), Batini et al. (2009), ISO 8000-8:2015 and DAMA-DMBOK2 are real.

Errors and limits

Dimension scores depend on rule definitions; publish the rules with the scores.

Exercises

Worked answers

EX008. Dimensions such as timeliness and relevance that consumers value.

EX009. 0.798.

EX010. Many records lack a sector code or postcode.

EX011. Some registration numbers appear twice.

EX012. Allowing blank sectors for new firms changed the rule, not the data.

EX013. Concepts and prerequisites for measuring information and data quality.

EX014. Empirical links between quality rule decisions and published outputs.