1 Data collection mode change Going from CAPI

1 Data collection mode change: Going from CAPI to CATI the case of Greece WAPOR 72 nd Annual Conference Workshop 2019, - Toronto, ON Canada Katerina Nikolova, Nikola Jovanovski and Carsten Broich www. companyname. com © 2016 Startup theme. All Rights Reserved.

CONTENT OVERVIEW 1 2 3 Introduction Background & why Greece Research Telephone penetration, population data, telephone stems Methodology CATI vs. CAPI: where do they differ? 4 5 6 www. companyname. com © 2016 Startup theme. All Rights Reserved. Sampling Approach Sample design Fieldwork & Results Comparison between CAPI and CATI results Conclusion Suggestions for future research 2

1. INTRODUCTION www. companyname. com © 2016 Startup theme. All Rights Reserved. 3

About Sample Solutions 4 Background Founded in the Netherlands, back in 2009 with the focus on RDD Telephone Sample Dual-Frame RDD sample for more than 100 countries worldwide in past 10 years. Projects Election Polling, Remote Areas, Conflict Zone, Gallup World Poll, PEW Global Attitude Study. www. companyname. com © 2016 Startup theme. All Rights Reserved.

Why Greece? Geography of Greece Difficult to cover whole of Greece with ~7% due to islands and remote areas that are difficult to sample “excluding the least densely populated PSUs” -> PEW Global Attitude Ionian Islands North Aegean Islands South Aegean Islands www. companyname. com © 2016 Startup theme. All Rights Reserved. 5

2. RESEARCH www. companyname. com © 2016 Startup theme. All Rights Reserved. 6

Research question What is the impact of neglecting the islands of Greece during a Nat. Rep study? Using a Nat. Rep CATI sampling design that includes the island population → Differences with regard to attitudes → Differences with regard to demographics → Differences in attitudes due to mode change (trend breaking? ) What can be learned for future mode changes? → Suggested split between mobile and landline to reduce Design Factor → What extra variables to be collected (eg. Postcodes) www. companyname. com © 2016 Startup theme. All Rights Reserved. 7

3. METHODOLOGY www. companyname. com © 2016 Startup theme. All Rights Reserved. 8

Comparison 9 99% coverage People who don’t use telephone Non-Coverage of Target Population mainland + island population CATI Includes the islands population - higher coverage island population CAPI Concentrates in many studies exclusively on the mainland population www. companyname. com © 2016 Startup theme. All Rights Reserved. Source: Special Eurobarometer 462: E-Communications and Digital Single Market, published in July 2018

4. SAMPLING APPROACH www. companyname. com © 2016 Startup theme. All Rights Reserved. 10

Sample design 11 Raw sample & activity check Creating stems Deep analysis of the official Generating phone numbers numbering plan document and pulsing 2 4 1 3 5 Feasibility check Estimations & sampling methods RDD Sample Penetration, frame sizes LL: Stratified Random www. companyname. com Cell: Simple Random © 2016 Startup theme. All Rights Reserved. Final sample

12 Sample composition 40% Landline N = 1000 Based on telephone market penetration www. companyname. com © 2016 Startup theme. All Rights Reserved. 60% Cell

5. FIELDWORK & RESULTS www. companyname. com © 2016 Startup theme. All Rights Reserved. 13

Fieldwork Stats CATI CAPI • Fieldwork duration: 13 days • Fieldwork duration: 27 days • Average interview duration: 14. 5 • Average interview duration: 42. 5 minutes • Questionnaire length: 27 questions • Questionnaire length: 32 questions → Follow-up: Why the large difference in duration (individual questions, time stamps per question) www. companyname. com © 2016 Startup theme. All Rights Reserved. 14

Mode differences Time pressure Memory effect Social Desirability bias www. companyname. com © 2016 Startup theme. All Rights Reserved. 15 Interviewer bias

Landline & cell respondents vs. age groups and gender Landline Type Cell. Slidefusion copyright Male Female Male 18 -24 1. 50% 0. 70% 4. 00% 3. 20% 25 -39 1. 40% 3. 40% 15. 00% 9. 90% 40 -59 5. 80% 10. 40% 15. 10% 9. 40% 60 -79 5. 90% 6. 20% 3. 80% 2. 40% 80+ 1. 10% 0. 70% 0. 00% 0. 10% 15. 70% 21. 40% Total www. companyname. com © 2016 Startup theme. All Rights Reserved. 37. 90% Female 25. 00% 16

17 CAPI Disposition Results RR 3 = 36, 4% Completed interviews 35% Refusal households 42% No answer 2% Language barrier 4% www. companyname. com © 2016 Startup theme. All Rights Reserved. Non-residential 1% Other 16%

18 CATI Disposition Results RR 3 = 16, 75% Sample used Completed interviews 16500 1000 Partial interviews 72 Eligible, noninterview 4138 Unknown eligibility 2105 Non-eligible 9185 www. companyname. com © 2016 Startup theme. All Rights Reserved. AAPOR Outcome Rate Calculator (DFRDD phone)

19 www. companyname. com © 2016 Startup theme. All Rights Reserved.

20 www. companyname. com © 2016 Startup theme. All Rights Reserved.

21 CONCLUSION www. companyname. com © 2016 Startup theme. All Rights Reserved.

Research question What is the impact of neglecting the islands of Greece during a Nat. Rep study? → Island population (landline part) is overall more positive but sample size not large enough to be significant. → Composition of landline population seems to be different: Younger people have access to landlines (less mobile only? ) Other lessons learned: → Response rate 16. 8% quite high for CATI (USA/Germany 5 -6%) → Dual Frame necessary not for non-coverage but for demographics → Attitudes between CATI and CAPI similar But: → Need for research on mobile-only population on Islands www. companyname. com © 2016 Startup theme. All Rights Reserved. 22

23 SAMPLESOLUTIONS WWW. SAMPLE. SOLUTIONS ADDRESS Stationsplein 45 3013 AK Rotterdam E-MAIL info@sample. solutions TELEPHONE WEBSITE Direct: 123 -4567 -9898 www. sample. solutions Fax: 123 -4567 -9898 The Netherlands www. companyname. com © 2016 Startup theme. All Rights Reserved.