Why50
- 1. The f ollowing slides show how many participants are needed to get a reliable visual analysis of one design
- 2. 1 Person
- 3. Large improvements in data quality 2
- 4. Large improvements in data quality 3
- 5. Large improvements in data quality 4
- 6. Large improvements in data quality 5
- 7. Large improvements in data quality 8
- 8. Large improvements in data quality 10
- 9. Large improvements in data quality 12
- 10. Large improvements in data quality 15
- 11. Moderate improvements in data quality 17
- 12. Moderate improvements in data quality 20
- 13. Moderate improvements in data quality 22
- 14. Moderate improvements in data quality 25
- 15. Moderate improvements in data quality 27
- 16. Moderate improvements in data quality 30
- 17. Moderate improvements in data quality 32
- 18. Small improvements in data quality 35
- 19. Small improvements in data quality 37
- 20. Small improvements in data quality 40
- 21. Small improvements in data quality 42
- 22. Small improvements in data quality 45
- 23. Small improvements in data quality 47
- 24. Small improvements in data quality 53
- 25. Some claim 8 people are enough to analyze visual attention. Following slides show random sets of 8 people from the same study shown on previous slides.
- 26. 8 people Sample A
- 27. 8 people Sample B
- 28. 8 people Sample C
- 29. 8 people Sample D
- 30. 8 people Sample E
- 31. Heatmaps of random 8 people out of all 53 participants were very different. It means sample of 8 can give a very misleading picture of true visual attention.
- 32. Others claim 15 people are enough for visual analysis. How about that?
- 33. 15 people Sample A
- 34. 15 people Sample B
- 35. 15 people Sample C
- 36. 15 people Sample D
- 37. Variation is still too large for samples of 15 people. What about samples of 30?
- 38. 30 people Sample A
- 39. 30 people Sample B
- 40. CONCLUSION Reliable visual analysis demands samples larger than 30 people