Chapter 14 — Data quality dimensions

Chapter 14 — Data quality dimensions

Lesson map

Lesson 14 published. Map, example, EX040.

flowchart TD
  N0["Learning outcomes"]
  N1["14.1 Dimensions"]
  N2["14.2 Fitness for use"]
  N3["14.3 Measuring dimensions"]
  N4["14.4 Trade-offs"]
  N0 --> N1 --> N2 --> N3 --> N4
  N4 --> EX["Worked example"]
  EX --> Q["Exercise EX040"]

Logical drawing

flowchart LR
  N0["Learning outcomes"]
  N1["14.1 Dimensions"]
  N2["14.2 Fitness for use"]
  N3["14.3 Measuring dimensions"]
  N4["14.4 Trade-offs"]
  N0 --> N1 --> N2 --> N3 --> N4 --> Q["EX040"]

[Learning outcomes]
    |
    v
[14.1 Dimensions]
    |
    v
[14.2 Fitness for use]
    |
    v
[14.3 Measuring dimensions]
    |
    v
[14.4 Trade-offs]
    |
    v
[Worked example] --> [Exercise EX040]

Learning outcomes

You will (1) describe common data quality dimensions; (2) choose dimensions for a use; (3) recognise that quality depends on use.

14.1 Dimensions

Amara studies Wang and Strong's research, which asked data consumers what quality meant to them and grouped answers into intrinsic, contextual, representational and accessibility categories. Common dimensions include accuracy, completeness, consistency, timeliness, validity and uniqueness. Batini and colleagues review how methods measure these. She notes that the research started from users' views rather than technical definitions, which supports the idea that quality depends on what data are used for at Harbourline. Amara lists each dimension with a short Harbourline example so that owners can relate them to their work.

14.2 Fitness for use

A member's birth year may be unimportant for sending receipts but essential for age-related offers. Store code accuracy matters for regional reports but not for points calculation. Amara explains that the same data can be good enough for one purpose and inadequate for another. She lists the main uses of loyalty data and the dimensions each depends on most, producing a matrix that shows where quality improvements would bring the most benefit to the co-operative. The matrix is shared with owners for comment.

14.3 Measuring dimensions

For each dimension Amara proposes a measure. Completeness is the share of records with a value. Validity is the share passing format rules. Uniqueness is the share of members without duplicates. Timeliness is how soon sign-ups appear in the system. Accuracy needs comparison with reality, such as asking a sample of members to confirm addresses. She notes that accuracy is the hardest and most expensive to measure, so it is usually sampled. She suggests sampling one hundred members each quarter by survey.

14.4 Trade-offs

Improving one dimension can harm another. Requiring email at sign-up would raise completeness but might lead staff to enter fake addresses, lowering accuracy, or discourage members from joining. Amara recommends asking for email with a clear benefit, such as digital receipts, rather than making it mandatory. She learns that quality targets should consider behaviour, because people respond to rules in ways that may undermine the goal behind them. The loyalty owner agrees, and the digital receipt option increases email completeness without any reports of fake addresses at stores.

Worked example

Amara studies quality dimensions from Wang and Strong and Batini, maps loyalty data uses to dimensions, defines measures and avoids mandatory email that would damage accuracy.

Synthetic case. Measures are synthetic; Wang and Strong (1996) and Batini et al. (2009) are real.

Errors and limits

Mandatory fields can produce fake values; consider behaviour.

Exercises

• EX040 — Research basis. Whose views did Wang and Strong start from? Rubric: 3 = data consumers;; 2 = partly correct;; 1 = other;; 0 = none.
• EX041 — Hardest dimension. Which dimension is hardest to measure? Rubric: 3 = accuracy;; 2 = partly correct;; 1 = other;; 0 = none.
• EX042 — Email approach. How should Harbourline encourage emails? Rubric: 3 = offer a clear benefit such as digital receipts;; 2 = partly correct;; 1 = mandatory;; 0 = none.

Continue studying — lesson 14 (chapter 14 of 40)

← Chapter 13 · Course chapter index · Chapter 15 →

Continue to the next authored lesson.