Upload
quasim
View
36
Download
0
Tags:
Embed Size (px)
DESCRIPTION
Building Evidence in Education: Conference for EEF Evaluators 11 th July: Theory 12 th July: Practice www.educationendowmentfoundation.org.uk. Panel session 3: Working with schools. Creative solutions: lessons learnt from evaluating the LIT programme Sarah Haywood NatCen. “Mind the Gap”. - PowerPoint PPT Presentation
Citation preview
Building Evidence in Education:Conference for EEF Evaluators
11th July: Theory12th July: Practice
www.educationendowmentfoundation.org.uk
Panel session 3: Working with schools
Creative solutions: lessons learnt from evaluating the LIT programme
Sarah HaywoodNatCen
Richard Dorsett, NIESREEF Evaluators Conference
12 July 2013
“Mind the Gap”
What is being tested?
• a parental engagement intervention – parents work with their children to create a short
animated film– series of facilitated sessions
• whole-school intervention – Metacognition: training teachers in the principles of
“learning to learn”• Targeted at year 4 pupils in academic year 2012/13• Predicted effect size of 0.35 - 0.45
Randomisation designAll schools:
NS=50
TreatmentNS
1=25
CPD & PENC=25
CPD NC=25
ControlNS
0=25
ControlNC=25
School randomisation
Class randomisation
A B C
Treatment effect 1: CPD & PEAll schools:
NS=50
TreatmentNS
1=25
CPD & PENC=25
CPD NC=25
ControlNS
0=25
ControlNC=25
School randomisation
Class randomisation
A B C
Treatment effect 2: CPDAll schools:
NS=50
TreatmentNS
1=25
CPD & PENC=25
CPD NC=25
ControlNS
0=25
ControlNC=25
School randomisation
Class randomisation
A B C
Treatment effect 3: CPD & PE vs. CPDAll schools:
NS=50
TreatmentNS
1=25
CPD & PENC=25
CPD NC=25
ControlNS
0=25
ControlNC=25
School randomisation
Class randomisation
A B C
Recruitment and randomisation• Birmingham, Devon, Haringey, Manchester• Drop-out is a worry
– 2 controls did so before knowing treatment status– 3 controls, 1 treatment dropped out & substituted– 2 controls dropped out & were not substituted
• Substitute schools – take treatment status of dropouts they replace– excluded from the impact estimates– provide potentially useful supplementary data
• Wanted 2-form entry but not always achieved
Achieved sample
NS=43
TNS
1=24
CPD/PENC=24
CPD NC=15
CNS
0=19
CNC=19
Some lessons• RCT Design is relatively easy – practical issues are more
complicated• The process of inducting schools is important to secure full
engagement pre-randomisation• Having something to offer schools control schools in particular
may help with drop out• Minimising drop out is best. But some drop-out is inevitable –
need for a protocol?• Some implications for analysis
– Helpful to understand reasons behind dropout– Can consider nonexperimental techniques– NPD analysis may be unaffected by drop-out of controls
EEF Conference 2013Towards a Protocol for Effective Recruitment
Mary SheardJuly 12, 2013
Recruitment as a problematic and complex relationship
“Recruit schools to the evaluation not the project”
Contexts
EEF projects: Project and design, challenges and solutions
EEF Protocol and Survey Outcomes:Effective recruitment; what has worked well and what have been the challenges
Non-EEF projects: Experience across a wide range of research studies and evaluations
What do we mean by ‘effective recruitment’?
Terminology: What is meant and understood? Programme Intervention Initiative Project Evaluation …
What has worked well• Relationships: schools, LA/parent organisations, programme
developers, evaluators trainers; test providers [Ethics]• Partnerships with schools: key personnel in school; lead project
contact; teacher implementers; technical support.
• Roles: clarity, responsibilities, expectations, inclusiveness [Ethics]
What has also worked well• Information: quality, clarity, conciseness, sufficiency, inclusivity;
suitability, accessibility (audiences, ethics]
• Examples of documentation; inviting initial expression of interest; school agreement form/contract; pupil data; data protection
Challenges and resolutions
Identifying and linking with key personnel
Senior leadership involvement
Lines of communication
Information overload
Saturation of constituency/schools asparticipant partners
More challenges and resolutions
Defining/explaining and the relationship between school, programme developer (trainer) and evaluation team
Timing
The concept of random assignment
Participation as control
Testing preparation and procedures
What we have learnedNeed to systematise a comprehensive recruitmentstrategy,to establish a recruitment protocol or checklist as the prequel to a project data management plan
Need to create a recruitment database
Need to consider equity/equal opportunity and fairnessin recruitment approaches: hard to reach schools and schools that are missed out
Developing a consistent recruitment strategy
Creating a protocol/checklist for effective recruitmentin future large-scale evaluations