Short on time? This page is the recap. For explanations and worked examples, use the full study guide.
Recap
- Crowdsourcing: obtaining input or information from a large number of people via the Internet.
- Citizen science: research done in whole or in part by distributed individuals, many not scientists, who contribute data using their own devices.
- Wide access to information and public data helps identify problems, develop solutions and share results at scale.
- Human capabilities can be enhanced by collaboration via computing; crowdsourcing offers new models, such as connecting businesses or social causes with funding.
- Challenges: data quality, uneven participation (links to the digital divide and bias), privacy, fair use and credit.
- Check and clean crowdsourced data, and check coverage as well as quality.
Key relationships
| Idea | How to use it |
|---|---|
| Quality control | Require several independent contributors to agree |
| Share of submissions | submissions in group ÷ total submissions |
| Coverage check | compare share of submissions with share of population or area |
| Over- or under-representation | share of data ÷ share of population (above 1 is over-represented) |
| Who takes part | Depends on device, connection, time and interest |
Assumptions and conventions
- Give the working for shares and check they add to 100%.
- When judging a data set, comment on both how much data and who or where it covers.
- A threshold rule (such as at least 3 reports) has a benefit and a limit; state both.
Mistakes to avoid
- Assuming that more contributions always mean better data.
- Ignoring who is not taking part.
- Describing crowdsourcing as only fundraising.
- Forgetting privacy risks to contributors.
- Using counts instead of shares when comparing groups of different size.
- Not saying what the organisers could do to improve coverage.
Two quick worked examples
1. A town app has 400 reports: 240 from the centre, 100 from the north and 60 from the south. The shares are 60%, 25% and 15%. If the south holds 30% of the town’s residents, it is under-represented (15% of reports for 30% of residents).
2. To check a citizen-science sighting of a rare bird, the project asks for a photograph and a location and accepts the sighting only when two independent volunteers report it. This improves quality, but a sighting in a place few volunteers visit may never be confirmed.
A revision routine
Define crowdsourcing and citizen science in one sentence each and give an example of every type of crowdsourcing: data, micro-tasks, ideas, problem solving and funding. Then invent a data set of 300 reports from three areas, calculate each area’s share and compare it with the area’s share of residents. Finish by writing two ways to improve quality and one way to improve coverage.
Quick self-check
- 600 of 800 reports come from city areas. What share is that? (600 ÷ 800 = 75%.)
- Give one way to improve reliability of reports. (Require several independent reports before confirming.)
- Is a volunteer who counts garden birds on a phone an example of citizen science? (Yes: distributed individuals contribute data using their own devices.)
Next: practice questions.