SlideShare utilise les cookies pour améliorer les fonctionnalités et les performances, et également pour vous montrer des publicités pertinentes. Si vous continuez à naviguer sur ce site, vous acceptez l’utilisation de cookies. Consultez nos Conditions d’utilisation et notre Politique de confidentialité.
SlideShare utilise les cookies pour améliorer les fonctionnalités et les performances, et également pour vous montrer des publicités pertinentes. Si vous continuez à naviguer sur ce site, vous acceptez l’utilisation de cookies. Consultez notre Politique de confidentialité et nos Conditions d’utilisation pour en savoir plus.
The Mind of the UserWhen 98% is more than 100%:How number format affects judgment & decisions Colleen Roller UX Matters Columnist On Decision Architecture
Do you know… How well do people understand and interpret numeric data? Does the format of numeric data have an impact? How should we display numeric data?
This presentation Reviews research that reveals how people perceive and use data Suggests UX design principles and best practices
A study Rate the attractiveness of a simple gamble (on a scale from 0-20) 1st group (rating = 9.4) 7 out of 36 chance to win $9 2nd group (rating = 14.9) 7 out of 36 chance to win $9 29 out of 36 chance to lose 5 cents Bateman/Dent/Peters/Slovic/Starmer - 2007
Judgment & decision making Must be able to attach meaning to data Meaning is determined via Context – a reference point or upper/ lower bound Comparison – against the contextThe ease with which meaning can be deriveddictates the extent to which data will be used.
A study Which bowl did people prefer: A. 100 beans, 7 of which are red B. 10 beans, 1 of which is red Denes-Raj/Epstein - 1994
A study Which disease is more dangerous? A. Kills 1286 out of every 10,000 people B. Kills 24 out of every 100 people Yamagishi 1997
A study Will a mental patient commit a violent act within 6 months of being discharged from the hospital? A. 20 out of every 100 patients commit a violent act (41% refused to discharge) B. 20% chance that the patient will commit a violent act (21% refused to discharge) Slovic/Monahan/MacGregor - 2000
Quiz Which is more easily understood: A. A 30% chance of rain B. A 3 in 10 chance of rain
30% chance of rain Common misinterpretations: It will rain in 30% of the area It will rain 30% of the time It will rain on 30% on days like this Gigerenzer/Edwards - 2003
Problems with probability What the doctor said: You have a 30% - 50% chance of developing a sexual problem What the patient heard: In 30% - 50% of your sexual encounters, something will go awry Gigerenzer/Edwards - 2003
Summary People tend to Comprehend frequencies better than probability/percent 20 out of 100, rather than 20% Focus on the numerator (and ignore the denominator) 9 out of 100 is bigger than 1 out of 10 Different expressions of equivalent data - e.g., 30 out of 1000 is more than 3 out of 10
A study Purchase equipment for use in the event of an airline crash landing: A. Chance of saving 150 lives B. Chance of saving 98% of 150 lives Slovic et al. - 2002
85% is more than 100% Even 85% of 150 is more than “150”! Slovic et al. - 2002
CalculationsProbability Frequency 1% of car trips result in an 100 out of 10,000 car trips accident. In 55% of the result in an accident. trips that result in an Among the 100 trips that accident, the driver is result in an accident, the drunk. In 5% of the car driver is drunk in 55 of trips that do not result in them. Among the 9900 car an accident, the driver is trips that don’t result in an drunk. If the driver is accident, the driver is drunk drunk, what is the in 500 of them. How many probability of an accident? car trips where the driver is drunk result in an accident?
Study: Rates of return Allocate money across two investment funds – stocks v. bonds Group A: shown 1-yr rates of return 63% Group B: shown 30-yr rates of return 81% Benartzi/Thaler, 2001
Data on the page 61% Versus75% Hibbard/Slovic/Peters/Finucane - 2002
No neutral design What information – and how it is presented – drives decision outcomes
Numeric ability Almost half of the general population has difficulty with simple numeric tasks National Adult Literacy Survey
Numeracy Those who are numeric Readily understand and use numeric data effectively Those who are non-numeric Informed less by numbers, and more by other non-numeric sources of info
Determine the right criteria Determine what decision criteria people should be using Highlight them (salience) Make it easy to evaluate, compute, & attach meaning
Frequency v. probability Convey absolute risks over relative risks 3 out of 1000 will have a stroke is better than 50% higher chance of stroke Don’t use decimals (.03)
Apples to apples comparisons When presenting various probabilities, keep the denominator consistent 20 out of 1000 compared to 1 out of 1000 is better than 1 out of 50 compared to 1 out of 1000
Attach meaning Use labels to show standard of performance Example: unacceptable, acceptable, excellent Labels provide expert guidance and easy mental processing
Mapping A higher number means better quality Reduces cognitive load Subtle, but influential
Quiz Which one results in better comprehension and better choices: A. Number of patients per registered nurse B. Number of registered nurses per 100 patients
Conveying small risks To help people understand the meaning of small risks Show context by providing a range of probabilities and risks for comparison Example: being hit by a car v. x-rays v. lighting v. asbestos
Emotion v. probability When consequences are marked by strong emotion All or none – sensitive to the possibility rather than the probability
Final thoughts Determine what info is most important What should people base the decision on? Design for meaning and ease Via context and comparison Facilitate easy computation Test – multiple methods Test drive, A/B testing, website metrics, think aloud, observe/probe in usability testing
Reference articles Simple Tools for Understanding Risks: From Innumeracy to Insight – G. Gigerenzer & A. Edwards, 2003 Numeracy and Decision Making – E. Peters et al., 2006 Numeracy and the Perception and Communication of Risk – E. Peters, 2008 Strategies for Reporting Health Plan Performance Information to Consumers: Evidence from Controlled Studies – J. Hibbard, et al., 2002 Risk as Analysis and Risk as Feelings: Some Thoughts about Affect, Reason, Risk, and Rationality – P. Slovic et al., 2004 Numeracy Skill and the Communication, Comprehension, and Use of Risk- Benefit Information – E. Peters et al., 2007 Reducing the Influence of Anecdotal Reasoning on People’s Health Care Decisions: Is a Picture Worth a Thousand Statistics? - A. Fagerlin et al., 2005 Bringing Meaning to Numbers: The Impact of Evaluative Categories on Decisions – E. Peters et al., 2009 When a 12.86% Mortality is More Dangerous than 24.14%: Implications for Risk Communication – K Yamagishi, 1997