GESIS Panel: Collecting and using representative primary data free of charge Prof. Dr. Michael Bosnjak GESIS Team Leader ´Survey Operations´ Full Professor of Evidence-Based Survey Methodology at the University of Mannheim

Personal Background January 2013 – December 2015: Inaugural Team Leader of the GESIS Panel Since January 2016: Team Leader Survey Operations and Full Professor for Evidence-Based Survey Operations

Study planningArchive and registeringResearchData collectionData analysis GESIS services are organized around user requirements

 Pretesting is generally considered to be indispensable for the successful development of a questionnaire  The GESIS Pretest Lab offers the following services:  Conducting cognitive pretests in the lab or „in the field“  Conducting eye movement analyses (eye tracking) in combination with cognitive pretests  Consulting on choosing and independent usage of various pretesting methods  Consulting on questionnaire design  Further information about the GESIS Pretest Lab: Example: GESIS Pretest Lab

Study planningResearchData analysis Focus of today´s GESIS Panel presentation Data collectionArchive and registering

GESIS Panel General Overview 6

7 Deliverables: Roles / Target Groups Providing free data collection services for social scientists within a mixed-mode access panel (offline & online) representative for the German population (18-70 years of age). 1 Providing free access to the data collected within the mixed-mode access panel to the social science community. 2 Enabling knowledge transfer on how to build and operate a representative panel. 3 Generating methodological findings on access panel data quality issues (publications, guidelines, standards). 4 Panelists Primary Researchers Data Users: Secondary Researchers Academic Panel Provider Survey Methodologists

8 Basic Access Panel Characteristics Probability-based access panel of individuals: General population in Germany, German- speaking, years Recruited panelists from population registers (270 sampling points) 7599 face-to-face interviews (CAPI) About 5,000 panelists (2014 starting sample)

9 Basic Access Panel Characteristics Multi-mode surveys: –Web-based surveys (approx. 65% of panelists) –Mail surveys (to reduce coverage error and to address sample composition bias concerns; approx. 35% of panelists) Experiments with 4 groups max. Bi-monthly data collection, approx. 20 minutes of interviewing time each wave.

10 Waves and Accepted Study Designs Single-wave studies: –Cross-sectional designs (e.g., and ) Multiple-wave, longitudinal studies: –Multiple cross-sectional designs; Splitted or partitioned cross-sectional design (e.g., S2 ) –Panel designs (e.g., S3 ) NOT considered: Cohort (sub-sample) studies S2.W1 S2.W3 S2.Wk S1.W1 S4.W2 Sx.Wk 20 minutes Wave 1 Wave 2Wave 3 Wave k time S3.W2S3.W3 S1S4 S2 S3

11 Target Group: Primary Researchers

12 Target Group: Primary Researchers Material to be prepared: 1.Proposal (structured abstract according to template) 2.Survey(s) (GPX- format, one per wave) 3.XLS Codebook(s) (one per wave) Material to be prepared: 1.Proposal (structured abstract according to template) 2.Survey(s) (GPX- format, one per wave) 3.XLS Codebook(s) (one per wave)

13 Target Group: Primary Researchers GESIS Panel Questionnaire Editor (Unipark)

14 Target Group: Data Users: DBK Data dissemination portal for incremental master-dataset, study descriptions, codebook, technical reports

15 Target Group: Data Users: Portal Data dissemination portal to generate customized data sets

16 GESIS Panel: Longitudinal Core Study Data quality indicators & “Evergreen” topics most likely demanded 1. Subjective Well-Being 2. Political and Social Participation 3. Environmental Attitudes and Behavior 4. Personality and Personal Values 5. Media/Communication Technology Usage 6. Socio-Demographic Update 7. Work and Leisure 8. Panel Survey Participation Evaluation & Survey Mode Preferences

GESIS Panel Recruitment 17

18 GESIS Panel: Recruitment Stages of the recruitment process

19 GESIS Panel: Recruitment Sample sizes and response rates

20 GESIS Panel: Recruitment Characteristics of the face-to-face recruitment CAPI Interview at respondent’s home Median interview duration 15 min Fieldwork period: June 2013-December 2013 Fieldwork agency: TNS Infratest 267 interviewers

21 GESIS Panel: Recruitment GESIS Panel recruitment phase

GESIS Panel Representativeness 22

23 GESIS Panel: Representativeness Comparison of the GESIS Panel to the Population

24 GESIS Panel: Representativeness GESIS Panel & German Micro-census (%) Group Population (Micro- census) GESIS PanelBias Age ≥ [8.5; 10.1]+1.2 Gender: Male [46.7; 49.5]-1.9 Urban area (≥ 100,000) [23.6; 26.1]-7.1 Education: upper secondary (Abitur) [30.8; 33.5]+6.1 One-person household [15.1; 17.1]-4.3 German citizen [94.1; 95.4]+4.9 Marital status: single [29.0; 31.5]-4.2

25 GESIS Panel: Representativeness GESIS Panel Onliner, Offliner & Micro-census (%) Group Population (Micro- census) GESIS Panel Online GESIS Panel Offline Age ≥ [5.1; 6.7]14.8 [13.2; 16.4] Gender: Male [49.3; 52.8]43.2 [41.0; 45.5] Urban area (≥ 100,000) [25.3; 28.5]21.5 [19.6; 23.4] Education: upper secondary (Abitur) [39.4; 43.0]17.0 [15.3; 18.7] One-person household [12.8; 15.3]19.4 [17.7; 21.3] German citizen [94.2; 95.8]94.3 [93.3; 95.6] Marital status: single [33.5; 36.9]22.1 [20.2; 24.0]

26 GESIS Panel: Representativeness Nonresponse in the GESIS Panel

27 GESIS Panel: Representativeness Subgroup participation rates (%) Group Stage 1 gross sample→ recruitment Stage 2 recruitment→ consent Stage 3 consent→ active panel Total38.6* Age ≥ (-0.3)77.3 (-4.4)90.7 (+11.2) Gender: Male38.8 (+0.2)80.6 (-1.1)78.6 (-0.9) Urban area (≥ 100,000)30.9 (-7.7)84.9 (+3.2)77.5 (-2.0) Education: upper secondary (Abitur) n/a 86.3 (+4.6)82.6 (+3.1) One-person householdn/a83.0 (+1.3)80.3 (+0.8) German citizenn/a82.7 (+1.0)80.8 (+1.3) Marital status: singlen/a85.6 (+3.9)73.0 (-6.5) Internet usen/a84.2 (+2.5)79.1 (-0.4) *AAPOR RR5

28 GESIS Panel: Wave completion

29 GESIS Panel: Attrition

GESIS Panel Regular Operation 30

31 GESIS Panel: Current Status Data Collection and Data Usage (September 2015) 33 conducted studies of approx. 60+ primary researchers: –25 submitted and 8 core studies –15 cross-sectional and 18 longitudinal studies –several small-scale experiments Upcoming waves are fully booked (until spring 2016) Approx downloads More than a hundred data users Currently ca panelists (refreshment in 2016)

Examples: Fielded Topics (+ LCS) SociologyPro-Environmental Behavior in High-Cost Situations Leisure Travel and Quality-of-Life be, cb bc, bd, be Political ScienceEuropean Election Study Conceptions of Democracy bb, bc, +2 waves bf PsychologyTime Perspective Scale Spatial Cognition PANAS Scale – Norms for Germany Prospective Memory Battery Within-Yearly Dynamics and Cycles in Subjective Well-Being ab ba bb bc, bf cb + 11 waves EconomicsInheritance taxesbf Survey Methodology Cross-National Replication of Question Design Experiments German Panel Comparison Study ba ca Cross NationalInternational Panel Comparison Study Reforms monitor: GESIS Panel, GIP, LISS, ELIPSS bb (2014) cc (2015) 32

33 GESIS FastTrack Procedure

34 GESIS Panel Data Management Workflow of a regular wave: Initialization step [PP 0] Preparation of incoming data, separately for each mode [PP 1 and 2]: –Corrections of ambiguous answers, initially coded as 7777 by data typists (only in offline mode) –Data editing –Implementation of missing value scheme Finalization of wave data [PP 3] –Merging of online and offline data –Data editing, implementation of missings, mode adjustments –Correction of meta information –Anonymization rules: e.g. blackening of open-ended answers, defining data publication status, recoding of data

35 GESIS Panel Data Management Integration of the regular wave to our master/panel data [PP 4]: Implementation and documentation of errata Merging of the current wave Implemetation of AAPOR scheme (broad spectrum of codes, e.g. mail bouncers, bouncers, dropout, etc.) Finalization of two data editions (STATA, SPSS and csv): –Onsite: GESIS Panel Extended Edition –Offsite: GESIS Panel Standard Edition Codebook: Combining meta information using STATA: -Variabel and value labels (data set) -Question text, intro text, item sources, etc. (external data sheets) Codebook is written in LaTeX using STATA

36 Screenshot Homepage hier

Contact: Prof. Dr. Michael Bosnjak GESIS — Leibniz Institute for the Social Sciences 37

38 GESIS Panel: Representativeness GESIS Panel & German Micro-census (%) Group Population (Micro- census) GESIS Panel Online Bias Age ≥ [5.1; 6.7]-2.2 Gender: Male [49.3; 52.8]+1.0 Urban area (≥ 100,000) [25.3; 28.5]-5.2 Education: upper secondary (Abitur) [39.4; 43.0]+15.2 One-person household [12.8; 15.3]-6.4 German citizen [94.2; 95.8]+5.2 Marital status: single [33.5; 36.9]+0.7