1. An MCDA framework for evaluating and appraising government policy for psychoactive drugs
Professor Larry Phillips
London School of Economics
and
Facilitations Limited
OR58 Annual
Conference
7 September 2016

2. Context Collaboration: DrugScience + Frisch Centre Purpose
DrugScience (Professor David Nutt is founder and Chair)
Ragnar Frisch Centre for Economic Research (Ole Rogeberg is the lead researcher)
Funded by the Norwegian Research Council.
Purpose
Develop an analytic framework for describing, measuring, assessing and discussing drug policy
Decision conferences
10-11 September 2015 and January 2016
18 participants, various backgrounds
Phillips & Nutt facilitating
Three models to test framework: alcohol, cannabis and heroin

3. Policy (Strategy) Table
Production
Sale/distribution
Purchase
Purchase volume (for legal users)
Possession
Use
Illegal (strong sanctions)
None - illegal
Illegal (weak sanctions - de jure or de facto decriminalized)
Per person quotas
Legal with restrictions for individuals (e.g, legal to brew wine, not spirits; grow cannabis, not more than 5 plants, grow "magic mush-rooms" in limited volume)
Only gifting between individuals (no money involved)
Only adult licensed users (e.g., tested for knowledge of harms and principles of safe use)
Per purchase quotas
Limited quantity per person (e.g., for personal use)
Only in specific, certified venues (e.g. pubs or on-site consumption with monitoring such as user rooms or use retreats)
Legal with restrictions for groups (e.g., cannabis clubs where users combine to finance growing for members)
State-run retail stores (e.g., state alcohol monopolies)
Only adults (age restrictions)
Self set quotas changeable with lag
No restrictions
Only in private homes
State controlled
State-licensed retail stores or pharmacies
Only in specific licensed venues or private homes
No restrictions for companies
Only mail-order
Anywhere except certain public spaces (e.g, indoor smoking bans)
No restrictions for companies or individuals
No restrictions - any retail store
34,300 combinations

4. Policy (Strategy) Table
Production
Sale/distribution
Purchase
Purchase volume (for legal users)
Possession
Use
Illegal (strong sanctions)
None - illegal
Illegal (weak sanctions - de jure or de facto decriminalized)
Per person quotas
Legal with restrictions for individuals (e.g, legal to brew wine, not spirits; grow cannabis, not more than 5 plants, grow "magic mush-rooms" in limited volume)
Only gifting between individuals (no money involved)
Only adult licensed users (e.g., tested for knowledge of harms and principles of safe use)
Per purchase quotas
Limited quantity per person (e.g., for personal use)
Only in specific, certified venues (e.g. pubs or on-site consumption with monitoring such as user rooms or use retreats)
Legal with restrictions for groups (e.g., cannabis clubs where users combine to finance growing for members)
State-run retail stores (e.g., state alcohol monopolies)
Only adults (age restrictions)
Self set quotas changeable with lag
No restrictions
Only in private homes
State controlled
State-licensed retail stores or pharmacies
Only in specific licensed venues or private homes
No restrictions for companies
Only mail-order
Anywhere except certain public spaces (e.g, indoor smoking bans)
No restrictions for companies or individuals
No restrictions - any retail store

5. Purchase volume (for legal users)
Policy options
Absolute Prohibition
State Control
Decriminalisation
Free Market
Production
Sale/distribution
Purchase
Purchase volume (for legal users)
Possession
Use
Illegal (strong sanctions)
None - illegal
Illegal (weak sanctions - de jure or de facto decriminalized)
Per person quotas
State controlled
State-licensed retail stores or pharmacies
Only adults (age restrictions)
Per purchase quotas
Limited quantity per person (e.g., for personal use)
Only in specific licensed venues or private homes
No restrictions for companies or individuals
No restrictions - any retail store
No restrictions

6. Drug Harm Policy Value Tree
7 impacts
Health
Social
Political
Public
Crime
Economic
Costs
27 evaluation criteria (with clear definitions)

7. Cluster
Criterion
Definition
Health
Harm to user
Prevents medical harms to a user resulting from consumption of intended substance; includes blood-borne viruses (BBV)
Harm to others
Prevents health harms (including BBVs) to third parties due to either indirect exposure (e.g., second hand smoking) and behavioural responses to consumption (e.g., injury due to alcohol induced violence)
More harmful substances
Decreases consumption of more harmful substances or increases consumption of less harmful substances (e.g., cannabis prohibition leading to synthetic cannabinoids)
Encourages treatment
Encourages treatment of substance-use problems
Product quality
Assures the quality of products due to mislabelled or counterfeit/adulterated product, unknown dose/purity
Social
Education
Improves education about drugs
Medical use
Policy does not impede medical use
Research
Policy does not impede research
Human rights
Policy does not interfere with human rights as distinct from the individual’s right to use.
Individual liberty
Policy does not interfere with individual liberty (individual’s right to use)
Community cohesion
Policy does not undermine social cohesion in communities
Family cohesion
Policy does not undermine family cohesion
Political
International development & security
Policy does not undermine international development and security
Industry influence on governments
Impedes drug industry influence on governments (less lobbying is preferable)
Public
Promotes well-being
Promotes social and personal well-being
Children and young
Protects children and young people
Protects vulnerable
Protects vulnerable groups other than children and young people
Religious/cultural value
Respects religious or cultural values
Crime
Criminalises users
Does not criminalise users
Reduces acquisitive crime
Reduces acquisitive crime to finance use
Reduces violent crime
Reduces violent crime due to illegal markets
Prevents corporate crime
Prevents corporate crime, e.g. money-laundering, tax evasion
Prevents criminal industry
Extent to which the policy discourages illegal market activity
Economic
Generates state revenue
Reduces economic costs
Reduces public financial costs not directly related to the enforcement policy (e.g., spillover effects on health policy budgets)
Cost
Introduction
Financial costs of introducing the policy
Maintenance
Financial costs of enforcing the policy

8. Scoring the options Direct scoring of harm 100 Least harm 80 60
Least harm 80 60
Relative
Strength of
Harm 40 20
Most harm

9. There are no data about the future.
But, the four policies are all hypothetical states about the future.
There are no data about the future.
So, how reliable (repeatable) and valid (represent actual harm) are direct preference judgements?

10. UK drug harms: 2010
ISCD results 2010

11. Europe drug harms: 2013
Europe results 2013

12. UK 2010 vs. Europe 2013
Europe UK
Direct preference judgements are reliable and valid in a decision conference if:
Criteria are defined clearly
Group members represent differing perspectives
Peer review occurs face-to-face
Group is properly facilitated
r = 0.993

13. Swing-weighting the criteria
“How big is the difference between most and least preferred options and how much do you care about it?”
1. Weight swings below each node, e.g., Health Impacts.…….Economic Impacts.…………………..Costs
2. Weight most important swings across the nodes.
Purpose of weights:
To ensure comparability of units of preference across the model. (They do not represent absolute importance.)

14. MCDA results Alcohol Cannabis
For both drugs, a legal but strictly regulated market is judged to yield the best outcomes overall.

15. Decriminalisationbetter
Alcohol
State control vs.
Decriminalisation
State control better
Decriminalisationbetter

16. Decriminalisationbetter
Cannabis
State control vs.
Decriminalisation
State control better
Decriminalisationbetter

17. Sensitivity analyses at each node
cannabis
State Control remains most preferred option over a wide range of weights.
Also for alcohol.

18. Current state
Alcohol results confirm current public health and medical opinion. Not so for cannabis.
For both substances, a legal but strictly regulated market is judged to yield the best outcomes overall.
Was there a ‘reformist bias’ in the group?
Need further research by other teams.
We now have the beginnings of a coherent analytic framework for describing, measuring, assessing and discussing drug policy

19. A guide to further reading
Multi-criteria analysis: a manual, 2000 (Dodgson, Spackman, Pearman & Phillips) Chapter 6 is an MCDA tutorial.
Cambridge University Press, 1993 The book that introduced MCDA in 1976 (Wiley).
UIT Cambridge Ltd, 2012 Explains the harms of misusing psychoactive legal & illegal drugs.