Since we wrote the paper for this conference, we have taken two new directions…
Since we wrote the paper for this conference, we have taken two new directions…
iThings 2012: Exploring Acceptance and Consequences of the Internet of Things in the Home
Exploring Acceptance andConsequences of the Internet of Things in the Home Tim Coughlan a, Michael Brown a, Glyn Lawson b, Richard Mortier a, Robert J. Houghton a,b & Murray Goulden a Horizon Digital Economy Research(a) & Human Factors Research Group(b), University of Nottingham, UK IEEE iThings 2012
Overview• Background• Conceptualising the Acceptance of Internet of Things in the Home• A Scenario-Based Survey Study• Further Ongoing Studies• Some Initial Design Guidance
BackgroundFrom building blocks and vision, becoming a technical reality…
BackgroundBut in what forms will it be accepted by the general public? What might be the consequences for our lives?
Background• Focus to date on creative ideas and technical development• Need to pursue a parallel user-centred agenda: – What is acceptable? – Can we predict impact? Potential consequences? – Methods to understand issues before implementation• Here we focus on information sharing in the home
Accepting New Technologies• Technology Acceptance Models – In short: If we perceive that a technology is useful and easy to use, we are likely to decide to accept it. – Various additional elements / weightings can be added to models to reflect data from different settings – But limited understanding of social influences R. P. Bagozzi, The Legacy of the Technology Acceptance Model and a Proposal for a Paradigm Shift. Journal of the Association for Information Systems 8 (4), 2007, 244-254. H. Sun and P. Zhang, The role of moderating factors in user technology acceptance. International Journal of Human Computer Studies, 64 (2), 2006, 53-78.
Accepting New Technologies• Diffusion of Innovations Model – At an individual level: • Process: Knowledge > Persuasion > Tentative Decision > Experimentation > Implementation > Confirmation – At a social level • Groups: Innovators, Early Adopters, Early Majority, Late Majority, Laggards E.M. Rogers, Diffusion of Innovations. 1983, New York: Free Press.
Accepting New Technologies• Domestication – Technology as a part of personal home routines and relationships – E.g. ‘who has the remote control?’ – Unfolding and evolving over the long term – Begin as novelty, becoming ‘unremarkable’ T. Berker, Y. Punie, M. Hartmann and K. J. Ward (eds) Domestication of Media and Technology, 2006, Open University Press. Pink, S., Home Truths: Gender, Domestic Objects and Everyday Life (2004). Oxford University Press. Crabtree, A., Mortier, R., Rodden, T., and Tolmie, P., Unremarkable networking: the home network as a part of everyday life. In DIS12 (2012) ACM. 554-563.
Internet of Things and the Home• Semi-Private• Diverse Relationships and Hierarchies• Relationships with Technology are Complex and Evolving
Internet of Things and the Home• Our current models for privacy may not work for IoT• E.g. Energy Monitoring data collects information about multiple household members.• Personal information can be ascertained through collation and contextual knowledge• Consent to, and knowledge of, what is being captured may be limited. A. Soppera and T. Burbridge, Maintaining Privacy in Pervasive Computing - Enabling Acceptance of Sensor-based Services, Intelligent Spaces: The Application of Pervasive ICT, A.Steventon and .S. Wright, (editors). 2006, Springer.
Methods Situated in Reality, but• Field Trials expensive. Does not scale well to many• Prototype Smart Homes designs or participants• Lab-studies• Surveys and Scenarios Detached from Reality but inexpensive. Potential to test many designs and participants
1st Scenario-Based Survey• Surveys based on presenting designs and scenarios of use can be effective in: – Understanding perceptions and intentions – Evaluating multiple designs / variations• Design some envisioned technologies• Show examples of information sharing with these• 35 Participants viewed 3 scenarios and provided feedback
A 1st Scenario-Based Survey• Scalable approach to collecting quantitative and qualitative data: – Delivered online – Aggregate statistics to closed questions – Open Questions: • Interpretations of presented data and responses • Why you would and would not want this
Quantitative Results Smart Energy Proximity Friedman Statement Fridge Monitor Portrait Test (X2) Q1 I can tell a lot about what people have been doing based on looking at … (the technology). 3.11 3.77 3.37 7.848 Q2 I would like …(the technology) to provide me with information about people that I live with. 2.60 3.60 2.14 39.018a Q3 I would be comfortable with the people I live with using … (the technology) if it included information about me. 3.17 3.80 2.23 39.631a Q4 I would like to see information about the general public collected through … (the technology). 2.66 3.40 2.09 31.88a Q5 I would be comfortable with the general public viewing anonymised information about me that was collected through (the technology). 2.63 3.26 1.89 34.308a Q6 I would be comfortable with commercial organisations viewing anonymised information about me collected through (the technology). 2.14 2.69 1.66 26.247a Q7 I understood what (the technology) does. 4.23 4.43 3.89 9.8 Q8 I understood what was happening in this scenario. 3.97 4.31 3.60 17.797a Q9 … (the technology) is easy to use. 3.94 4.00 3.74 3.155 Q10 … (the technology) is useful. 3.14 4.20 2.60 36.505a Q11 People under the age of 21 would find … (the technology) easy to use. 4.00 4.11 3.83 5.915 Q12 People under the age of 21 would find … (the technology) useful. 3.00 3.29 3.03 3.5 Q13 People aged 22-60 would find … (the technology) easy to use. 3.89 4.09 3.77 3.057 Q14 People aged 22-60 would find … (the technology) useful. 3.29 4.06 3.09 21.121a Q15 People aged over 60 would find … (the technology) easy to use. 3.54 3.63 3.29 0.711 Q16 People aged over 60 would find … (the technology) useful. 3.00 3.77 2.80 24.065a Q17 I would like to have … (the technology) in my home. 2.83 4.09 1.89 49.333a a = significant at df=2, p≤0.05 with Bonferroni Correction
Acceptance• Statistically, the Energy Monitoring scenario was significantly more popular than the Smart Fridge and Proximity Portrait scenarios• However, this was often related to type of household: “I do the shopping and the cooking and am… fully aware ofwhat is in the fridge… in a larger household, with many people helping themselves…there may be some call for it.”
Privacy• The Proximity Portrait was seen as the most invasive and concerning scenario for respondents “I would have to be convinced that it had a useful purpose which outweighed the ‘Big Brother’ scenario” It “would be uninteresting if it did not know about enough activities, and could be uncomfortable if it was too precise”
Usefulness and Action• Usefulness appears linked to provoking action: – 31 respondents said they would act in response to the Energy Monitor scenario e.g.: “Make sure that Jane’s getting enough social contact and physical / mental exercise…in a way not to offend or cause guilt” – 15 for Smart Fridge – 4 for Proximity Portrait
Next Steps: Tailored Scenarios• Scenarios need to present data about people• In two scenarios a fictional household was used• Research suggests that fictional protagonists can affect perceptions of scenarios• Can we make scenarios more ‘situated’? van den Hende, E. A., Dahl, D. W., Schoormans J. P. L., Snelders, D., Narrative Transportation in Concept Tests for Really New Products: The Moderating Effect of Reader–Protagonist Similarity, J. Product Innovation Management, (2012), John Wiley & Sons. 1540-5885.
Next Steps: Tailored Scenarios• New Method: – Ask for information about the household – Use this to generate a version of the scenario with data displayed for a person’s actual household• To be released as a Wordpress plugin for adaptation to various designs and scenarios
Next Steps: Tailored ScenariosHouseholdmember namesinserted here
Next Steps: Tailored Scenarios• Themes: – Most expect competitiveness to emerge, but varies as to whether this is a good or bad consequence – Questioning whether quantity equals value: • Steps taken wearing a sensor does not cover all exercise • Food eaten in the home is not all food eaten – More detail with respect to relationships with other household members • How we expect them to act affects interpretations of data
Next Steps: Tailored Scenarios (Study 1 n=35, Study 2 n=79)
Next Steps: Field Studies• Currently running studies where households share data from ‘Fitbits’ (individual clip on pedometer device)• Display placed in communal area of each home• Individual steps over the day and comparative chart• 1 week installations to see some evolution of opinions and behaviour
Next Steps: Field Trials• Finding similar themes around: – Frustration that Data (Steps) != Value (Exercise) – Competition arises, but sensors not gathering everything, get lost, not ‘fair’. Very disruptive. – Becomes a discussion point over meals – People interpreting and learning more about each other (e.g. amount of exercise done in school day compared to parent’s desk job). – Strong distinctions drawn between sharing in the household and sharing with the wider world.
Initial Design Guidance• Focus on the balance between privacy and utility – Make a case for how useful a system is, how people can act based on the results. – Intrusion with a purpose is far more acceptable – Don’t design to collect or share data without a clear purpose – Don’t expect to share outside the home in the same form as inside
Initial Design Guidance• Leverage relationships with understood issues and technologies – Hard for people to understand ‘Really New Products’ • Lack of connection with their current lives • Lack of ways to envisage using it – Energy Monitor is well understood as an extension of a product some respondents already owned and liked – Proximity Portrait thought of “like Weasley’s clock in Harry Potter” – Fantasy, but also “creepy” S. Hoeffler, Measuring Preferences for Really New Products, Journal of Marketing Research Vol. 40, No. 4 (2003), pp. 406-420. AMA Press.
Initial Design Guidance• Support negotiation and flexibility over sharing and use of data – Hard to support existing models of consent to collect data – Can we focus on evolving consent for different uses of data? “I will let you know some form of (data x) now, for this purpose, but not for other purposes” T. Toscos, K. Connelly, and Y. Rogers. 2012. Best intentions: health monitoring technology and children. ACM conference on Human Factors in Computing Systems (CHI 12). ACM Press, 1431-1440.
Initial Design Guidance• Make use of ambiguity: – To avoid conflict or sense of surveillance – To support various agreed upon levels of sharing• However, there is a social problem when data does not equate to the expected ‘value’ – Need to be clear to viewers what is NOT being captured – Competition relies on fairness W. Gaver, J. Beaver, S. Benford, Ambiguity as a Resource for Design, Proc. of the SIGCHI conf. on Human factors in comp. sys. (CHI), ACM Press, Ft. Lauderdale, Florida, 2003. P.M. Aoki, A. Woodruff, Making space for stories: ambiguity in the design of personal communication systems. Proc. of the SIGCHI conf. on Human factors in comp. sys.(CHI) , ACM Press, Oregon, 2005, p. 181-190.
Upcoming WorkshopMethods for Studying Technology in the Home @Information available soon via http://chi2013.acm.org/
Conclusions• Scenario-based methods can be useful to understand perceptions and potential consequences of many designs, but also awareness that scenarios are not situated events• Potential to innovate methods based on importance of envisaging actual household
Conclusions• Studies suggest preferences for information sharing which: – Clearly connects to actions the viewer can perform – Is well understood – Relates to the needs of the specific household• Privacy inside the household may be very different to that outside, but IoT blurs boundaries that people value• The physical walls of the home are a meaningful boundary for people. Need to distinguish sharing within and without