1. The Phone/Internet Divergence John Curtice Senior Research Fellow NatCen Social Research and ‘The UK in a Changing Europe’ initiative

2. The Phone/Internet Divergence Based on all polls 1.12.15-19.5.16. Deal = 19.2.16. Last 6 weeks = all polls since 1.4.16

3. The Two Methods Phone (Random) Dialling of Landline and Mobile Numbers Try to fit quotas when phone is answered Internet Stratified (random) selection from a pre-recruited panel Potential Common Pitfall Differential availability and willingness to participate Both rely on weighting achieved data to make it look representative

4. Possible Sources of Difference Administrative Differences Presentation of Don’t Knows Social Desirability Sampling Differences Are internet participants just the politically committed? Are those who can be got hold of by phone in a short period of time unrepresentative?

5. An Experiment with Don’t Knows Source: Populus/Number Cruncher Politics

6. In Practice Since beginning of April phone polls have averaged 11% Don’t Knows, internet polls, 17% But big spread amongst internet polls Remain averages 51% in internet polls with above average % of DKs, and 50% in those with a below average % ORB’s internet polls that do not offer Don’t Know as a response have averaged Remain 51; Leave 49

7. Within Age Group Differences Source: ICM 14-15.5.16

8. Within Social Grade Differences Source: ICM 14-15.5.16

9. The Importance of Education Source: British Social Attitudes 2015

10. Phone Polls Have More Graduates? Source: Populus/Number Cruncher Politics

11. Taking Into Education Into Account Source: YouGov 29.4-12.5.16

12. However, Source: YouGov 29.4-12.5.16

13. Conclusion Explanation probably does not lie (primarily) in the differences in how phone and internet polls are administered More likely to lie in differences in achieved samples Polls are mostly failing to take into account one of the key demographics of the referendum – education But only limited and inconsistent evidence so far on whether phone and internet polls differ in their ability to recruit graduates Leaving us all, none the wiser!