The Dubuque Electricity Portal: Evaluation of a City- Scale Residential Electricity Consumption Feedback System + CHI 2013 -Thomas Erickson / 맹맹맹 x 2013 fall
The Dubuque Electricity Portal:Evaluation of a City-Scale Residential Electricity Con-sumption Feedback System
+ CHI 2013-Thomas Erickson/ 맹욱재x 2013 fall
The Dubuque Electricity Portal:Evaluation of a City-Scale Resi-dential Electricity Consumption Feedback System
2013.10.16Wookjae Maeng
Why this paper?
• ICT policy project• Curious about the way of evaluation
3
4
Dubuque?
http://www.popsci.com/science/article/2011-11/data-transforming-dubuque-and-it-will-transform-your-city-too
5
Author
Thomas Ericksoninteraction designer and re-searcherSocial Computing Group at IBM's Watson Labshttp://researcher.watson.ibm.com/researcher/view.php?person=us-snowfall
6
Abstract
Dubuque Electricity Portalfor volunteering reductionLog, Survey, interview20weeks pilot, 765 household3.7% reductionUnderstanding of usage↑
7
Problem
• Energy consumption is critical global issue
• In developed, developing countries• Side effect : CO2 emission, ecology dam-
age
• Among U.S’s Electricity consumption 37% from home
8
Previous Work
Sustainability
EnvironmentalPsychologyHCI
9
Previous Work 2
ECF(Electricity Consumption Feedback) techniques
1. Feedback (show how much they use)2. incentive (points, pricing, differential pricing)3. Comparisons (between individuals, group)4. Goal-setting 5. Commitment (elicit public commitment)
10
System
• The Sitestrong sustainability agenda citysmart electricity meters are deployed
• The Systemevery 15min transmit data to gatewayData was stored and uploaded in cloudanalyzed, fed back to web portalnot track individual devicebut high, medium, low load device
11
Solution
Figure 2 on this paper
12
Pilot project
• 765 neighborhood – 20 weeks• First log in – fill in profile for cus-
tomizingsize of household, type of device
• Survey, interview for UX qualitative data
13
Survey
• Survey – 32 questions, 10minmultiple choice, Likert scale, open-ended
1. Why?2. How often?
If answer > 1go next question
elsego final question
3. Portal UI – value, degree(cue for picture)4. Effects on understanding and use5. Physical or Behavior Change• Distributed by email when the pilot ended
22% answer (116/561)
14
Survey 2
• Privacy requirement – link survey to log data X
• 34% - logged on > 1 (91 survey, 266 log)
• 53% response rate of logged on > 1• 5% never use (25 survey, 490 log)
15
Interview
• 18( 1 couple, 4 women, 13 men)• 23 ~ 90 years old• Most single-family, living with spouse• 33% having children• Recruited by city employee
most engaged, active participants• 45 min• 15 city office, 2 participant’s office, 1
restaurant
16
Interview 2
• Semi-structured• Informant’s background and household• Same with survey• Discussing their experience with oth-
ers?• How the portal could be improved?
17
Result - Survey
• 35% use > 1 from log(= water portal)opt-in energy conservation program < 10%
• Reasonsover 90%1. interest in reducing electricity costs2. belief in the importance of sustainability of city3. Concern about the environment4. Dislike of waste and a preference for being frugal– Curious about the tech : 84%– Concern about climate change : 72%– Good educational experience : 59%91% conserve electricity before : highly motivated
18
Result 2 - Survey
• Barriers51% no difficultiesproblem with ID or password : 42%difficulty in finding the site : 12%can’t see their data : 4%
• Complexity of initial password from privacy is-sue( mixed number and upper, lower case)
• Forgetting password : 27%no time : 26%integrating into ordinary life is hardtoo difficult (5%), too complicated(4%)
19
Result 3 - Survey
• Degree of uselog in > 5 / week : 12%1 / week : 18%occasional : 31%rare : 25%don’t recall : 14%
• Baseline consumption checkerfrequently at first(30%), regularly(20%)
30%
30%
40%
20
Result
Figure 3 on this paper
21
Result 4 - Survey
• Time based got good score• Alert, Facebook Chat require menu access• Clear correlation between UI component
and understanding how they use• Comparison by Neighbor, Alerts, Facebook
chat were bad.
22
Result 5 - Interview
• Incentivesmany : comparing with themselvesothers : comparing with similar neighborfixed itselfNo points from direction
• Contextaverage temperature, degree of heating/cooling daysvacation or not
23
Result – Impact of portal useExceptional Participants266 participant 3.7% reduction
<= Already taken actions Small support
61% at least one change in infra - unplugging(40%)42% plan for change - energy star appliance(28%)69% at least one change in behav-ior - non peak period(51%) - less use of light(33%) - short shower, dishwasher cycle(32%)
table 1 on this paper
24
Discussion
• Limitation1. Not random samples – smart electricity meter
particular social economic characteristic- single family, blue or white color jobs, retired
- internet literate2. Volunteers
- highly motivated- already taken many actions to conserve
3. Over-represent of participant from survey, interview
25
Discussion 2
• Accessibilitydifficulty with ID and passwordcomplexitycan estimate waking up, sleeping, vaca-tionunfavorable comparison=> biometrics to single sign on
• Engagementcollective attention from campaign
26
Discussion
• Manner of usetime-based visualization – historic feedbackmultiple feedback optionschat failure – face to face conversation, not privacy=> larger set will be solved=> linking to local discussion
• Credibilitybig spike – system errorbut refrigerator defrost cycle, dehumidheaterelectricity consumption is invisible=> list of possible explanationalert should be informative
27
Contribution
• HCI 에서 sustainability design approach a lotfew field study, identifying gap be-tween real and theory
• thorough evaluation - 20 weeks,765 household, log, survey, 18 interview
28
Generalizing to Me
• ECF? -> Electricity Consumption Feed-back!
• HVAC? -> Heating Ventilating Air Condi-tioning
• Environmental Psychology• Good reference for public data visualiza-
tion• Insight from limitation and result• Curious about implementing in Korea
29
Sharing
• What is the bad points of this paper?• How do you think about implement-
ing in Korea, further issues?• Any idea about improving chat, in-
teraction with other users?