M5 – Collecting the Right Data

Introduction

Understanding where data comes from, how it is collected, and why certain choices matter is fundamental to good research practice.
This module introduces the principles and practicalities of collecting high-quality, ethical, and relevant data to support effective analysis.

Learning Outcomes

Note

By the end of this module, you will be able to:

  • distinguish between primary and secondary data sources
  • plan data collection in line with a clear analytical purpose
  • select appropriate formats and tools for structured data capture
  • identify sources of bias and assess fairness in datasets
  • relate these principles to equivalent workflows in Python

Key Terms

Tip

Primary data — data collected first-hand for your research purpose
Secondary data — pre-existing data collected by others
Bias — systematic error that distorts representation
Relevance — the degree to which data addresses the research question
Completeness — presence or absence of required variables
Consistency — uniform formats, coding and measurement standards

1. What Is Data and Why Does It Matter?

Data refers to raw facts, observations, or measurements describing phenomena in the world.
From these, we derive information, insights, and evidence.

Data enables:

  • identifying patterns
  • supporting evidence-based decisions
  • optimising processes
  • predicting future trends
  • driving innovation in research and practice

2. Sources of Data

2.1 Primary Data

Data that you collect directly.

Examples

  • Surveying classmates about favourite snacks
  • Measuring heights in a room
  • Counting steps using your phone
  • Observing activity at a campus gate

Strengths: high control, tailored, customisable
Limitations: slower, resource-intensive

2.2 Secondary Data

Data collected by someone else, usually for a different purpose.

Examples

  • National statistics portals
  • Open weather datasets
  • Published research data
  • University demographic reports

Strengths: fast, low cost, broad coverage
Limitations: relevance constraints; fixed variables

2.3 Comparison

Feature Primary Data Secondary Data
Control High Low
Cost Higher Lower
Speed Slow Fast
Relevance Tailored Variable
Richness Flexible Fixed

Activity: Data Hunt

Task 1 — Collect primary data

Gather from 3 peers: - Name
- Height
- Favourite snack

Task 2 — Compare to secondary data

Use the provided dataset containing: - Gender, Age, Home Country
- Steps Walked (yesterday)
- Library Usage (per week)
- Commute Time + Transport Method
- IMD Score
- Hours Slept
- Student Status
- Club/Society Participation

Task 3 — Reflect How does your primary data differ in completeness, relevance, and consistency?

3. Planning for Good Data

3.1 Why Planning Matters

Good data demonstrates:

  • Relevance — aligned with the research question
  • Accuracy — faithfully measured
  • Completeness — minimal gaps
  • Consistency — uniform formats and coding
Warning

Collecting unnecessary or inconsistent variables leads to poor-quality insights and higher cleaning burden.

3.2 From Vague Ideas to Measurable Questions

Vague idea:
“Improve campus transport.”

Analysis purpose:
Examine how student demographics relate to commute burden.

Measurable questions:

  • What % of students cycle to campus?
  • Do PG students have longer commute times than UG students?
Tip

Measurable questions tell you which variables are needed and how to structure the data.

Activity: Question Shuffle

Instructions

  1. Choose a project idea
    • Improve student wellbeing
    • Optimise campus resources
  2. Identify:
    • analysis purpose
    • refined research question
  3. Select variables from the list:
    Name, Gender, Age, Favourite Snack, Height, Home Country, Steps Walked, Library Usage, Commute Time, Transport Method, Hours Slept, Student Status, Club Participation, Healthy Meals Frequency, Postcode, IMD Score.

Example Solutions

Student Wellbeing

  • Purpose: link sleep to social engagement
  • Question: Do club-active students sleep more on average?
  • Variables: Hours Slept; Club Participation

Resource Equity

  • Purpose: assess usage by deprivation
  • Question: Do students from low IMD areas visit the library less often?
  • Variables: IMD Score; Library Usage

4. Tools and Data Formats

4.1 Data Formats

  • Numeric: age, height
  • Text: name, home country
  • Categorical: snack type, transport method
  • Dates/times: survey timestamps, durations

Example

Age → numeric → supports mean, min, max
Favourite Snack → categorical → supports mode, frequency

4.2 Tools for Data Capture

Tool / Platform Best For Notes
Excel / Google Sheets numeric, text, categorical Good for small datasets
Google Forms surveys, structured input Automatic collection, export to CSV
Phone Sensors / Apps time-series data Automated timestamps, high fidelity

Activity: Sorting Game

Categorise data cards into:

  • Primary data (you created it)
  • Secondary data (already existed)
  • Ambiguous cases (depends on context)

Primary examples:
Surveys you collected; height measurements; observed traffic.

Secondary examples:
Government statistics; downloaded weather data; published research.

Ambiguous:
A friend’s snack preference; a classmate’s dataset; Fitbit challenge uploads.

5. Trust, Bias, and Fairness

5.1 Common Bias Types

Bias Type Example Effect
Sampling bias Only surveying morning students Misrepresents population
Measurement bias Using an uncalibrated scale Inaccurate estimates
Selection bias Surveying only volunteers Inflated engagement measures
Confirmation bias Leading questions in satisfaction surveys Skewed responses

5.2 Principles for Fair Data

Note
  • Include diverse groups
  • Use clear, neutral questions
  • Document data collection procedures
  • Reflect on missingness and under-representation

Activity: Biased Sampling Game

Each group receives coloured counters representing a student population:

  • Pink = female
  • Blue = male

Tasks - Estimate % female
- Compute average age of females
- Compare with known full-population statistics

Discussion prompts

  • How does sampling error distort estimates?
  • Which biases can be reduced, which are unavoidable?
  • How can we report uncertainty transparently?

Summary

Note

Key Takeaways

  • Know your data source — it affects trust and interpretability.
  • Good questions drive efficient, relevant data collection.
  • Tools and formats shape data quality and analytical ease.
  • Bias is unavoidable but manageable with planning.
  • Ethical, transparent data collection strengthens research validity.