Group 3 Tmp 1 - First Impression
1 like590 views
This document summarizes a proposed study on how first impressions and errors impact user ratings of websites. The study would use an experimental design to test how a very good or average first impression followed by an error affects appeal ratings. It would also test whether errors on similar-looking websites impact ratings. The study aims to provide insights on managing user expectations and avoiding copying other websites that may fail. Potential risks include confounding variables and the need for qualitative follow up.
Group 3 Tmp 1 - First Impression
- 1. TwinTide AUtumn Training SchOol 2013: REsearch Methods for Human-Computer Interaction (TUTOREM 2013) „FOOL ME ONCE, SHAME ON YOU...“ Group 3 Kirsten Christiana Christiane Kwamena
- 4. THE FIRST IMPRESSION How crucial is it to make a good first impression? Primary proposer: Dr. Alexandre Tuch studies showed: 50 ms is enough to make a reliable first impression of a website http://www.designdecoy.com/5-ways-smart-businesses-make-great-first-impression
- 5. RESEARCH QUESTIONS How does a first impression with a website effect how users rate it after an error? How does prior error with a website effect user‘s first impressions of a visually similar website?
- 6. METHODOLOGICAL APPROACH positivist experimental approach (quantitative) establish whether the effect occurs variables have been proven to be measurable interpretist approach (qualitative) establish causes why the phenomena occurred explain any effect found
- 7. Good impression WS very good appeal scale very good first impression, could lead to high expectations which then might lead to big disappointment after an error occurs very bad
- 8. Average impression WS very good appeal scale moderate first impression, could prevent the development of high expectations which then might lead to less disappointment after an error occurs very bad
- 9. Good impression WS Average impression WS appeal scale very good very bad Phase 1: between-subjects
- 10. HYPOTHESES H0: an error has no effect on the appeal rating H1: very good impressions will lead to a large drop in appeal rating after an error H2: very good impressions will lead to a significantly larger drop in appeal rating after an error than average impressions
- 11. appeal scale very good Good impression WS Similar good impression WS Average impression WS Similar average impression WS very bad Phase 2: within-subjects
- 12. HYPOTHESES H0: an error has no effect on ratings of visually similar websites H1: following an error in a good impression website, a visually similar website will be rated lower than the initial one H2: following an error in an average impression website, a visually similar website will not be rated significantly lower than the initial one
- 13. STUDY Participants n = 100, randomly recruited, familiar with IT, no experience with web design Materials desktop computers, 12 websites (collected through a prestudy of appeal rating and visual similarity) Design Phase #1: between-subjects Phase #2: within subjects independent variable: appeal of website with error (good / average) dependent variable: appeal rating (first impression, after error, first impression of visually similar WS) Analysis ANOVA
- 14. PROCEDURE appeal scale random WS 60 s appeal scale random WS 50 ms
- 15. appeal scale PROCEDURE 3 random WS (rated for appeal) appeal scale 60 s, then error appeal scale 50 ms 5 random WS (rated for appeal) 50 ms
- 16. CONTRIBUTIONS TO HCI „Keep your promises!“ first impression may not be the most important thing to strive for - one should live up to the expectations one evokes „Don‘t be a copycat!“ if the copied (popular) website fails, this might impact the first impression of your website your website might promise more, than you can deliver
- 17. RISKS confounding variables: error was not attributed to the website worse appeal of visually similar website was not caused by the prior error scale might be inappropriate to measure appeal after interaction Qualitative follow-up study may clarify this