1. How to design and deliver a successful evaluation 19 th October 2015 Sarah Lynch, Senior Research Manager National Foundation for Educational Research (NFER), UK Email: s.lynch@nfer.ac.uks.lynch@nfer.ac.uk NFER website: www.nfer.ac.ukwww.nfer.ac.uk

2. Are you ready to measure impact? How is the programme used in schools? Is it used as intended? If not, why? What challenges are faced? What changes should be made? Feasibility Study/ Pilot

3. Focus the research questions Link to the main aims of the programme: Age of first drink? Frequency of drinking? Increased knowledge? Link to the main aims of the programme: Age of first drink? Frequency of drinking? Increased knowledge? Most important outcomes to measure For example, background questions (gender, ethnicity etc.), where they get alcohol from, who they drink with and so on… Other questions/outcomes Is the programme working as planned? What do teachers think about it? What are the practical challenges? Process & ‘fidelity’

4. Strength of evidence Level 1 Descriptive Level 1 Descriptive Level 2 Correlation/ Comparative Level 2 Correlation/ Comparative Level 3 Before and after (Users only) Level 3 Before and after (Users only) Opinion of “does it work”? Small-scale Lower cost Very limited evidence Users compared with non-users After use only (no baseline) Limited/emerging evidence One group; users Before and after Limited/emerging evidence € € - €€

5. Strength of evidence Level 4 Before and after (Users and Comparison) Level 4 Before and after (Users and Comparison) Level 5 Level 4, with multilevel model analysis Level 5 Level 4, with multilevel model analysis Level 6 Randomised Control Trial Level 6 Randomised Control Trial €€ €€ - €€€ €€€ Users and a matched comparison group Before and after Moderate evidence Add more sophisticated analysis Most credible for peer-reviewed publication (Level 5&6) Considerable evidence Example: Talk about Alcohol, England Random allocation to groups Most robust Most credible for peer-reviewed publication (Level 5&6) Considerable evidence Example: Steps Towards Alcohol Misuse Prevention Program (STAMPP), NI

6. Sample size considerations Number of schools using programme Number of students per school Size of effect/expected change If a small change is expected, need larger sample to detect it Representative of the population

7. More schools or more students? Increasing number of schools is better for a school-based programme

8. Requirements for intervention group Talk About Alcohol evaluation Year 14 lessons 1 hour on website Year 22 lessons Talk About Alcohol evaluation Year 14 lessons 1 hour on website Year 22 lessons Specify minimum use of programme Written/online instructions But only what will be available in reality to all schools Written/online instructions But only what will be available in reality to all schools Clear guidance on programme Implement a realistic, consistent approach Fidelity - is the programme used as intended?

9. Realistic and practical Robust, systematic approach Maintaining balance

10. Avoiding school drop-out Be clear about: –purpose and aims of the evaluation –expectations and evaluations tasks –timescales/number of time points Offer incentives: –Feedback on own results compared to whole sample –Prize draws –Offer comparison schools materials later Maintain contact; avoid burden; make it simple

11. Data collection Parents Young people (Age?) Young people (Age?) Teachers Attitudes Knowledge Behaviour Characteristics Attitudes Knowledge Behaviour Characteristics Process Impact Process Impact Attitudes, role models, perceptions of norms Questionnaire (online or paper?) Interviews Questionnaire (online or paper?) Interviews It’s a challenge in England! Face-to-face interviews Telephone Questionnaire It’s a challenge in England! Face-to-face interviews Telephone Questionnaire Before and after questionnaire Identical questions Comparability with other studies Before and after questionnaire Identical questions Comparability with other studies Who What How

12. Analysis Statistical models to allow for any differences in characteristics of groups Need to collect data on student characteristics Is any difference between groups statistically significant? Need a larger sample to do reliable analysis

13. Adding to the evidence-base Publish Other coverage Report Peer-reviewed academic articles Media, newsletters, magazines Blogs Influence Stakeholders Policy Strength of evidence Levels 5&6

14. Discussion What challenges do you face? What solutions have you found? How can you strengthen your evidence?