Coping with policy resistance in a complex world: not simply BENU

Cristiano Codagnone IPTS Behavioural Economics Seminar Series Session 4 Seville, 9 June 2011

To guide the motions of the human puppet, it is necessary to know the wires by which he is moved
M. Helvetius, 1810, A treatise on Man, his intellectual faculties and his education
Not only BE and Nudging

Explaining the title

Policy resistance : tendency for interventions to be defeated by the systems response to the intervention itself (*) Main causes: a) intrinsic complexity (*); b) inadequate policy making intellectual framework (*), (**); c ) ineffective use of, often, weak evidence (***) For over sixty years generations of policymakers have been raised to have a mechanistic view of the world, and a checklist mentality: to achieve a particular set of aims, draw up a list of policies and simply tick them off (**)
(*) Sterman, J.D., Learning from evidence in a complex world. Am J Public Health, 2006. 96(3): p. 505-14 (**) Ormerod, P. N Squared: Public Policy and the Power of Networks, RSA 21st Enlightenment (***) Ongoing work for forthcoming monograph on impact evaluation in the domain of Information Society
Not only BE and Nudging

Contents
Part I Part II Three Disclaimers Obesity as anchoring in a social network

Part III Privacy: survey results and BE concepts Part IV ICT for better health

Not only BE and Nudging

Part I: 3 disclaimers
Syncretism Questions rather than answers Criticism

Not only BE and Nudging

syncretism: melding eclectic thoughts

Innovative modelling and the data deluge Obesity, smoking and binge drinking as anchoring in networks

Complexity and system thinking

?
foundationalism vs pragmatism in evidence based policy-making Network analysis Ontological certainty vs ontological unpredictability

Understanding privacy through behavioral hypotheses and survey data

Behavioural economics and nudging

Envisaging the future: ICT meets science toward better health

Not only BE and Nudging

castles in the air or truth to power?

Reforms as experiments in an open society (Popperian Inspiration): Initial foundationalism with its claim on privileged knowledge based on methodological strategic choices

HIGH E V ID E N C E V A LID IT Y / R E LIA B ILIT Y

3.5B

Pragmatism Ideas become true just in so far as they help us to get into satisfactory relations with other parts of our experience (William James, Pragmatism: A New Name for some Old Ways of Thinking, 1907 Pragmatism in policy evaluation User friendly and user oriented aggregation of facts using with craftsmanship all available tools Policy makers and politicians do not read or read at most 11 minutes a day, they form their ideas in exchanges with lobbyists, think tanks, and consultants, so lets give them a few factoids for the press release: eHealth to potentially increase GDP by X%
Not only BE and Nudging

2.6B

LOW
1.4B

HIGH

LOW EVIDENCE USABILITY

Spotting the blind spots

Beobachter [...sind] blind fr ihren blinden Fleck Niklas Luhman Literal translation The observer is blind for his blind spot Or a different way to look at it: The system does not see what it does not see

Asking the right questions instead of providing one fit all answers can help spotting the blind spots of policy making Including disentangling what is fad and fashion about BENU from how it can contribute, among all other approaches and instruments, to improve evidence use in policy making

Not only BE and Nudging

Critical frame of mind

But, if constructing the future and settling everything for all times are not our affair, it is all the more clear what we have to accomplish at present: I am referring to ruthless criticism of all that exists, ruthless both in the sense of not being afraid of the results it arrives at and in the sense of being just as little afraid of conflict with the powers that be
Karl Marx , Letter to Arnold Ruge, September 1843 (first published in Deutsch-Franzosische Jahrbucher, February 1844)
What is new about BE? David Hume: we favour the present over the future Adam Smith: disproportionate aversion to loss Karl Marx: False consciousness

Big Society and Nudging a nice way for the Cameron Cabinet to say small state and distract public attention from spending cut?

The we know better: First came the Jacobins Then the Bolshevik And now the Nudgers?

POLICY RESISTANCE INADEQUATE FRAMEWORK INEFFECTIVE US OF EVIDENCE

BE not really a revolution: So different from Game Theory? Still agents maximising a preference relation over some space of consequences and the solution in most cases still involves standard equilibrium concepts

Who nudge the nudger? Policy makers also prone to same biases and not necessarily motivated by maximization of the public good

BE intrinsic limits: Only the individual, groups&institutions? Anchoring from the Framingham Heart Study (3 cohorts over 31 years) versus experiments mostly with students, who know they are observed, have small stakes, etc

Not only BE and Nudging

Part II: Obesity as anchoring

Some background on obesity Framingham Heart Study data analysis With a little help from my friend(*)? N Square
(*) Using headline of New York Time article on obesity and network ( 18 September 2010)

Not only BE and Nudging

A Tsunami in 3 decades
Not predicted nor prevented a Tsunami of poor health, entirely mediated by modernised lifestyles, has hit our shores on both sides of the Atlantic
( + 125.3 % ) ( + 250.0% ) ( + 85.5% ) ( + 151.5% ) ( + 131.4% ) ( + 61.9% ) ( + 41.4% ) ( + 39.1% ) ( + 93.1% ) ( 137.9% ) ( 45.9% )

OECD Health Data ( * ) data comes from measurement, all other are self-reported

Not only BE and Nudging

Modernity is outpacing biology

5-6 Million Years

Not only BE and Nudging

3-4 Decades

Determinants of obesity: standard view

Dietary Input + Energy intake + Glycemic indices + Ingestion Frequency -- Dietary Fibers Lipid profile

Thrifty Genotypes/ Phenotypes

Environmental factors Food relative prices Retail /logistic (fresh food) Image of, exposure to, food Information, education Build environment Family structure TV and digital leisure Healthcare delivery

Adverse outcomes Diabetes II Syndrome X Obesity Hypertension Hyperlipidemias Hyperinsulinemia CDV diseases

Energy Output - Transportation costs - Food preparation / subsistence costs - Micro environmental thermal costs - Leisure activity costs - Work/ occupation costs

Not only BE and Nudging

The Framingham Heart Study: basic information (*) Interconnected social network of 12,067 people, assessed repeatedly from 1971 till 2003:
1948 5029 enrolled 1971: most of the children of original cohorts and their spouses enrolled 2002: third generation enrolled

All sort of network related information available and analysed:

Ego: person whose behaviour is analysed Alter: A person connected to the Ego who may influence him/her Degree of separation: social distance between two people measured by the smallest # of intermediaries between an ego and others in the network

(*) Christakis, N.A. and J.H. Fowler, The spread of obesity in a large social network over 32 years. N Engl J Med, 2007. 357(4): p. 370-9

Not only BE and Nudging

Hypotheses and methods

Homophily: obese people simply stick to each other Confounding: obese people share SES and other environmental factor Induction: alters influence and/or have peer effects on egos Method:Longitudinal logistic regression model where obesity status at any examination point T+1 is a function of
Egos attributes (age, sex, education) Egos obesity at previous time T Alters obesity at time T and T+1 Time lagged dependent variable to control for Egos genetic endowment and any other intrinsic predisposition to obesity Time lagged independent variable for Alter obesity controlled for homphily:
Significant coefficient for this variable would mean that an alters weight would affect the egos weight or that the two would experience contemporaneous events affecting both their weights Model was estimated in varied ego-alter pair types

Also looked at smoking cessation and at geographic distance

Not only BE and Nudging

Visualisation

http://www.nejm.org/action/showMediaPlayer?doi=10.1056%2FNEJMsa066082&aid=NEJMsa066082_attach_1&a rea=
Not only BE and Nudging

Findings
Directional influence in friendship ties
If an ego stated that an alter was his/her friend, the egos chances of becoming obese appeared to increase by 57%, if the alter became obese Between mutual friends the egos risk of obesity increased 171% if an alter became obese No statistically significant relation when the friendship was perceived by the alter but not the ego

Degree of separation, quality of tie, and geographical distance

Among pairs of adult siblings: increased chance by 40% Risk that a friend of a friend of an obese person would be obese was 20% higher than in a random network (10% at 3 degree of separation) Whereas degree of separation decreased the effect of an alter on an ego, geographic distance did not

Not only BE and Nudging

Discussion
Authors conclude they have compellingly shown that obesity may spread in social networks in ways shaped by the nature of social ties Assuming data have been correctly tortured, why is it so?
Alter directly influences egos behaviour Alter changes egos norms of acceptability:
Special case of anchoring to your alter as implicit reference point So as increased BMI spread our anchor may also increase

Other interesting implication

Assume that for any X people who stop smoking, there will be an additional percentage of X that eventually will quit as well(*) Assume now you run a randomised control trial using mainstream HTA to evaluate a lifestyle intervention helping X people to stop The cost per QALY measure from the above will underestimate the actual net effect of the intervention The more so if among the X you catch one who is an influential alter for many egos

(*) Christakis, N.A. and J.H. Fowler, The collective dynamics of smoking in a large social network. N Engl J Med, 2008. 358(21): p. 2249-58
Not only BE and Nudging

Binge drinking in the UK

Ormerod P and Wiltshire G, Binge drinking the UK: a social network phenomenon. Mind and Society 2009. 8(1): p. 135152

Not only BE and Nudging

N Square: Nudging plus network (*)

(*) Ormerod, P. N Squared: Public Policy and the Power of Networks, RSA 21st Enlightenment. Quoting D Watts. A simple model of global cascades on random networks,in Proceedings of the National Academy of Science, 99, 5766 5771, 2002.
Not only BE and Nudging

Different types of networks

Random network Scale free network: few nodes have many links (tipping points, influencers, etc) Small world

Not only BE and Nudging

Networks effect and policy

Nudge a few to obtain success But very complex
If the world is like: increased uncertainty Unintended consequences and resistance What worked? The treatment or the network?

Ormerod syncretic approach:

Traditional survey information on network Agent Based Modelling Simulation Nudging focal nodes in the network
Not only BE and Nudging

Part III: Privacy Industry knows it better Selective survey results Typical case for prospect theory Information ineffective Privacy 2.0: ownership and PDV

Not only BE and Nudging

Floating data for price discrimination (*)

We leave traces of our behaviour everywhere that are more or less legally accessible: Online shopping details Catalogue purchases, Magazine subscriptions, Leisure activities Information from social-networking sites Health Insurers are testing modelling and data mining technique to : Identify customers risk profiles Segment their offer and/or select their customers (avoiding adverse selection through price discrimination) All of the above, if it works, at substantially lower costs and in much shorter time if compared to traditional methods

(*) Insurers Test Data Profiles to Identify Risky Clients Wall Street Journal, 19 November 2010
Not only BE and Nudging

marketing data to predict life spans (*)

(*) Insurers Test Data Profiles to Identify Risky Clients Wall Street Journal, 19 November 2010
Not only BE and Nudging

Selective results from survey (*)

For instance
Some feel entitled to protection of information about themselves and then end up trading away that same information for little reward People change idea on whom to grant access to their credit card transactional data, if they are asked about repeatedly Feeling their privacy protected by merchants offering SSL connections (security or privacy) Presence of privacy policy taken by many to mean protection regardless of content Privacy seal interpreted as guarantee of trustworthy website Even well presented warning about spyware not always leads individuals to abort installations Users of social sites do not change their default settings Rapresentativeness heuristics : neat appearance of website=trustworthiness Users think that providing personal information on social networks could cause privacy problems to other users but not to them

(*) Most reported in Acquisti, A., & Grossklags, J. (2008). What can behavioral economics teach us about privacy? In Digital Privacy: Theory, Technologies, and Practices (pp. 363-377). New York and London: Auerbach Publications; some confirmed by Eurobarometer Survey on eID (Joint IPTS/DG INFSO, DG Justice project)
Not only BE and Nudging

Why? objective difficulties

Data subjects know less than data holders about magnitude of data collection and possible secondary usage: become subject to externalities

Complex life-cycle of information in digital society has multitude of consequences individuals can hardly imagine
Two unknowns: what privacy-relevant outcomes, with what consequences Even when aware about risks, they miscalculate probability and magnitude of occurrence Privacy risks by-product : privacy good attached to other goods in a bundle Privacy costs and benefits difficult to estimate, they are immaterial: ambiguity about probability and value of outcomes

Not only BE and Nudging

Why? Cognitive biases

Even with full information and cognitive power to process it, we fail to do so due behavioural anomalies and biases:
No consistent preferences between alternatives Anchor value of personal information to arbitrary value: difficult to set a price, but once one accepts reward offered by merchant that may become the anchor Tend to discount as improbable an identity theft Valence effect: overestimation of the likelihood of favourable effects. Rational ignorance: cost of learning higher than potential benefits Status quo bias: default privacy settings Who can ex post, and even the more so ex ante, assess the costs and benefits of privacy decision making based on simplified heuristics? Behavioural economics offer a number of tools of better understanding privacy decision making

Not only BE and Nudging

Limits of information (*)

Commission find it easier to go for information requirements rather than for harder interventions:
i.e. Directives 2002/65 on distance marketing of consumer financial services (OJ 2002 L 271/16) imposes obligation to provide consumer with information on some 40 pieces of information

A win-win approach for 2 out of 3 (output in place of outcome game)

Policy makers tick their boxes, Traders are often happy to cover their backs by over-supplying information Consumers, limited by the magic number 7, are forced into rational ignorance

Conclusion
Information ok Nudging also good (change default values: cooling off periods, etc) But they cannot substitute liabilities, bans, etc

Not only BE and Nudging

(*) Howells, G. , The Potential and Limits of Consumer Empowerment by Information. Journal of Law and Society, 2005. 32 (3): p. 349-70

Privacy 2.0 (*)

Codes of Fair Information Practice:
data privacy practices by government and corporations no longer adequate: distributed data collection, data mining, and easy dissemination Personal participatory sensing: sensing devices of all kinds deployed by individuals change expected flows of information of both pubic and private spaces

Community-Based Participatory Research (CBPR):

Success in health and environmental research: validity of data marginalised groups to act on the data they helped collect and analyse Users see data as theirs, care about them, change attitude: ownership

Personal Data Vaults (Google Health, Microsoft Health Vault)

Owner is the user Displace ineffective policy efforts guided by wrong intellectual framework ( i.e. on electronic identity, inter-operable personal electronic health records) and the likes

Not only BE and Nudging

(*) K. Shilton et al, 2009, Designing the Personal Data Stream: Enabling Participatory Privacy in Mobile Personal Sensing

Part IV: The future: ICT for better health

Barabasi: advent of network medicine The view of the clinician Crowds, modelling, persuasive technologie

Not only BE and Nudging

Network medicine (*)

(*) A. Barabasi, Network Medicine From Obesity to the Diseasome, N Engl J Med, 357, 4: 404-407
Not only BE and Nudging

Example from COPD (*)

Courtesy of Prof Josep Roca (from his presentation Predictive Medicine: the view of the clinician)
Not only BE and Nudging

More knowledge needed (*)

Courtesy of Prof Josep Roca (from his presentation Predictive Medicine: the view of the clinician)
Not only BE and Nudging

Knowldege density biased

Predictive Capacity

Little/ no information Zone

Low Wellness
Authors visual essay
Not only BE and Nudging

Illness Zone

Power law distribution

Immagine nodes are health data and links are their co-occurrence with other health data for each of the N=individuals

And immagine a third (N=individuals in good or poor health) ax We may verify that 20% of key health data have 80% of co-occurences explaining poor or good health!
Not only BE and Nudging

<<

Crowdsourcing, living epidemiology, and persuasive technology

Community based crowdsourcing: high yield health data
Individuals input a few high yield data (nutrition, activity and key physiological parameters) for themselves and their children They use ubiquitous mobile technologies to communicate and receive data They are provided with new functionalities to detect and input physiological parameters

Crow dsourcing of simple but High Yield Health Data data intensive predictive modelling

Prediction & Prevention

Living epidemiology: real time analysis of cooccurrences

Researchers (granted access to the data) identify the per unit and crowds co-occurrences explaining wellness or illness Evidence used to evaluate and support public health investments Evidence mapped against characteristic of community help shape other interventions

Personalized feedback and nudges through persuasive technology

Persuasive technologies: change behaviour

Can integrate classical interventions Can overcome and avoid the resistance that these interventions encounter It is a quintessential nudging approach leveraging crowds participation and network effect

Not only BE and Nudging

Nudging to avoid hard battles?

Impose Taxes (on individuals, on junk food) Alter relative prices (agricultural policy) More effective and spread food labeling Menu labeling in fast food No vending machines in schools Restriction in advertising of nutrition poor food Information / educational campaign Intervention on the built environment Integrated health and social care prevention New battle like with Tobacco?

The bad guys

Industry script in the US is clear Americans need to be more active and take greater responsibility for their diets M. Kent, Coca-Colas CEO (Wall Street Journal Oct 7 2009, op-ed in response to envisaged taxes on soft drinks)

NEITHER
Individual responsibility (you get what you deserve)

OBESITY

NOR
State Big Brother (I care and choose for you)

BUT
Libertarian paternalism (architect/ steer choices)

Conclusive thoughts

Not only BE and Nudging

Back to the future

The Greek provided the universal analytical proof that the square on the long side of every right angle triangle has the same area as the sum of squares on the other two sides Babylonian engineers simply measured the sides of a thousand right triangular and heuristically induced the same conclusions Science paradigms Past 3000 years: Past 400 years: Past 30 years: Tomorrow: Empirical Theoretical-analytical Computational Data intensive eScience
Needle in the haystack: benign and malign co-occurrences explaining good or bad health

Not only BE and Nudging

Final considerations
One, certainly useful, thing is BE and its experiments

Another one is integrating BE with other insights and tools to:

Reinforce the evidence base for policy making Help drive a paradigm shift in policy making intellectual framework Spotting and correcting blind spot

Refining evidence and approach can lead us closer to the never ending process of approximation to know the world

Not only BE and Nudging

At the end of the day, however

what counts as evidence or what counts as research involve not just technical objective judgments but also subjective and contextualised assessments. The attaching of labels such as evidence or research to particular types of knowledge are in fact a political actWe need to recognise then that knowledge, evidence and research are all privileged terms that reflect the perceptions, priorities and power of those who use them (Polanyi, 1967; Foucault 1977; Giddens, 1987). Thus the playing out at ground level of debates about what counts as research is by no means always a rational/technical matter, but instead involves a complex deployment of both technical expertise and power dressed up in the guise of rationality
S Nutley, I Walter and H. Davis, Using evidence: how research can inform public service, Bristol, The Polity Press, 2007, p. 25

Not only BE and Nudging

Thanks for your attention

Not only BE and Nudging

<<

How It may works

change

Evidence based choice architectures & public health policies High yield Health data

Personalised co-occurences + Network based nudges

Social network analysis +

Ubiquitous mobile media and sensors

Persuasive technologies and serious gaming

community data commons

t ed bu tri dis ure & t ta uc da astr r ic nt inf ma ting se u p ed ar com Visualised Sh

Institutional legacy data

Visual analytics and simulation

Visualised effects of public health policies

Complex modelling

effects of benign & malignat Co-occurences Visual analytics

and simulation

Worldwide scientific and practitioners community

Not only BE and Nudging