92526 Chap.6 Meaningful Data: Quality, Questions and Bias
Meaningful Data: Quality, Questions and Bias
Well-organised data can still be meaningless, and this chapter covers the Week 2 tools for telling the difference.
It begins with the claim that the way data is obtained matters as much as the data itself, then shows how the form of a survey question shapes what comes back: open questions climb from fill-in metadata to deep qualitative answers, and closed questions from yes-or-no counts to sliders that give deeper quantitative data.
The chapter sets out the seven quality dimensions the slides list from IBM, namely completeness, uniqueness, validity, timeliness, accuracy, consistency and fitness for purpose, and audits a season of GPS files against them with real arithmetic.
It closes with Patton's question types and with bias, the systematic error that enters when questions or technologies carry the collector's assumptions.
What this chapter covers
- 01
Why collection methods matter as much as data
- 02
Open and closed question forms
- 03
Patton's question types
- 04
Completeness, uniqueness and validity
- 05
Timeliness, accuracy and consistency
- 06
Fitness for purpose
- 07
Bias from question framing and technology
Worked example · free
Rank four survey questions by the depth of data they return
- 1(a) A fill-in open question that returns metadata, identifying the respondent rather than describing the session.
- 1(b) A closed alternative-answer question that returns shallow quantitative data.
- 1(c) A closed Likert question that returns deep quantitative data on one specific dimension, difficulty.
- 1(d) A long-answer open question that returns deep qualitative data, including reasons the designer did not anticipate.
Key terms
- Completeness
- How much of the expected data is present and usable.
- Validity
- How well data captures the construct it is supposed to measure.
- Timeliness
- Whether data is ready within the time frame in which a decision needs it.
- Bias
- Systematic error introduced into collection or analysis that distorts interpretation, often through question framing.
- Likert scale
- A closed rating question on a fixed scale, such as 1 to 5, giving quantitative data on one dimension.
Meaningful Data: Quality, Questions and Bias FAQ
What are the dimensions of data quality?
The slides list seven, credited to IBM: completeness, the amount of usable data; uniqueness, the amount of duplication; validity, whether data matches the construct it measures; timeliness, whether it is ready in time; accuracy, correctness against an agreed source of truth; consistency, how stable a measure is across observations; and fitness for purpose, whether it meets a real need.
What is the difference between accuracy and consistency?
Accuracy asks whether values are correct against an agreed source of truth, such as a laboratory force plate. Consistency, also called reliability, asks whether a measure gives stable results from one observation to the next. A device that always reads 3 cm high is consistent but not accurate.
How do open and closed questions differ in the data they give?
Open questions let people answer in their own words: a short answer gives shallow qualitative data and a long answer that asks why gives deep qualitative data. Closed questions offer fixed responses, from yes-or-no items through checklists, rankings and Likert ratings to sliders, returning progressively deeper quantitative data on chosen dimensions.
What are Patton's question types?
Patton's typology, shown on a Week 2 slide, sorts questions by what they ask about: behaviour, experience, opinion, value, feelings, knowledge, sensory impressions, demographics and background. The slide asks about one opposition attack in soccer in each of these ways, showing that one event can produce very different data.
How does bias get into sport data?
Bias is systematic error in collection or analysis. It enters when collectors assume a technology measures what they believe is real, or when questions are framed around problems the collector already expects, so issues nobody thought to ask about never appear in the data. Mixing question types and opening with open questions, as this chapter suggests, is one way to reduce that risk.
Exam move
Learn the seven quality dimensions with one sport check for each, then practise diagnosing which dimension fails in short scenarios, paying particular attention to accuracy versus consistency and validity versus accuracy. Work one numerical audit yourself, computing completeness and duplication from expected and actual counts.
For question design, take one problem, such as fatigue in a squad, and write one question for each rung of the open and closed ladders, then say which you would ask first and why. Finally, rewrite one leading question so that it no longer assumes the answer, which is the simplest defence against bias.
Working through Meaningful Data: Quality, Questions and Bias in 92526? Sia is AskSia’s AI Information Technology tutor — ask any 92526 Meaningful Data: Quality, Questions and Bias question and get a clear, step-by-step explanation grounded in how 92526 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.
