Sarah Cotterill and Hisako Nomura

1
A Randomised Controlled Trial of a Door-to-Door
Recycling Campaign

• Sarah Cotterill and Hisako Nomura
• Institute for Political and Economic Governance
• Rediscovering the civic and achieving better
  outcomes in public policy
• www.civicbehaviour.org.uk

2
Overview

• Context and background
• Research design
• Canvassing and Participation Monitoring
• Results
• Policy Implications
• Further research

3
Background and context

• Government target 50 of household waste to be
  recycled or composted by 2020 (31 2006/7)
• Key actions include Culture Change
• (Defra Waste Strategy for England 2007)
• Citizen behaviour change essential for achieving
  better environmental outcomes
• Many environmental acts are low effort, but need
  reminders, habit or brief contacts to activate
• Kerbside recycling popular .. but not everyone
  does it

4
Encouraging Recycling

• Doorto-door Canvassing
• Street effects household recycling behaviour is
  influenced by neighbours (Tucker 1999 Shaw 2008)
• Habit current participation makes future
  participation more likely
  (Freedman Fraser 1966 Cialdini, Trost
  Newsom 1995)
• Importance of the baseline

5
Who recycles

• Recyclers tend to be older, more affluent, home
  owners and better educated.
• Areas with high ethnic minority population tend
  to have lower recycling rates (but coincides with
  deprivation).
• Case studies suggest no significant difference in
  recycling behaviour of Asian and white households
• Transience?
• (Martin, Williams Clark 2006 Perry Williams
  2007)

6
Research Questions

• Does canvassing lead to a rise in kerbside
  recycling participation?
• Is the rise in participation sustained over time?
• Is canvassing more or less effective when the
  baseline recycling rate is low?
• Does the success of canvassing vary according to
  the characteristics of the population?

7
Old Trafford and Gorse Hill
8
The research site

• Old Trafford
• 53 white
• Relatively deprived (some parts in lowest 6 of
  English neighbourhoods)
• Mean Multiple Deprivation Score (2007) 41.7
• Mix of terraced and semi-detached housing
• Gorse Hill
• 85 white
• Less deprived (in the lowest 30 nationally)
• Mean Multiple Deprivation Score (2007) 30.6
• Mix of terraced and semi-detached housing

9
EMERGE Recycling

• Social Enterprise
• Wide range of materials
• Boxes and bags
• Weekly collection
• Visibly different vehicles, sorted on street
• Promotion of the scheme
• Quality of recycling service affects recycling
  rates
• (Harder et al 2006 Woodward et al 2005)

10
Sample and randomisation

• 6580 households in 194 streets
• Street based design expected street effect
• Streets randomly assigned to
• Canvass group (97 streets, 3468 houses)
• Control group (97 streets, 3112 houses)
• Stratified by district (Old Trafford/Gorse Hill)
  and street length
• Thanks to the York Trials Unit for doing the
  random assignment

11
Canvassing

• May/June 2008 6 weeks
• 4 canvassers recruited and trained
• 3pm-7pm Mon-Fri 11am-3pm Sat
• Whole area canvassed twice
• Spoke to 2129 of the 3468 households (61 of
  households)
• Awareness, Attitudes, Barriers (Recyclers and
  Non-Recyclers) (Shaw et al 2007)
• Leaflets

12
Measurement

• Participation Monitoring (WRAP 2006)
• Same day as recycling collection
• Independent monitor
• 3 weeks
• Participation ratio of households that
  recycle at least once in three weeks
• March/April 2008
• July 2008
• October 2008

13
CONSORT Flow Diagram
14
Blinding

• Households unaware of the research study or the
  participation monitoring.
• Participation monitors not aware of which
  streets in canvass/control groups.
• Recycling collection crew not aware of which
  streets in canvass/control groups.

15
Variables involved in the analysis

• Outcome variable
• Post intervention recycling participation ratio
  ( of households per street observed recycling at
  least 1 in 3 weeks).

16
Variables involved in the analysis

• Explanatory variables
• Street level
• Canvass Group (coded 0Control Group, 1Canvass
  Group)
• Baseline Participation Ratio
• Times (coded 11st follow up 22nd follow up)
• Street size (total number of households per
  street)
• Super Output Area level
• of population that is non-white (2001 census)
• Index of Multiple Deprivation score (2007)

17
Descriptive analysis Frequencies
18
Descriptive analysis Recycling and Deprivation
19
Descriptive analysis Recycling and Ethnicity
20
Descriptive analysis Recycling and Recycling
base rate
21

p.lt.1 plt.05 plt.01 plt.001
22
(No Transcript)
23
Findings

• Recycling participation rose by 5 amongst the
  canvassed streets compared to the control group.
• 3 months after the canvassing there was still a
  significant difference between the canvass and
  control streets, but it was reduced.
• Canvassing has less impact on streets with very
  high or very low recycling rates

24
Findings

• Recycling rates are lower in the more deprived
  areas.
• After controlling for deprivation, recycling
  rates are higher in the areas with a larger
  ethnic minority population.
• Canvassing is particularly effective in the areas
  of medium or high deprivation.
• Canvassing is particularly effective at raising
  recycling in areas with a high proportion of
  ethnic minorities.

25
Policy Implications

• Canvassing campaigns are likely to be most
  successful if targeted in
• Streets with low baseline recycling rates
• Relatively deprived areas
• Canvassing and other promotional campaigns need
  to be repeated regularly to reinforce the
  recycling message.

26
Further research

• Analyse data at a household level?
• Explore street effects and social norms further.
• Is transience an issue here?