본문 바로가기 주메뉴 바로가기
국회도서관 홈으로 정보검색 소장정보 검색

목차보기

Title page

Contents

Abstract 3

1. Introduction 4

2. Invitation Messages for Self-Administered Surveys 7

3. Adaptive Randomization in Experiments 11

4. Experimental Procedures and Data 14

5. Results 21

5.1. Which Invitation Messages Yield More Survey Starts? 21

5.2. Which Treatments Increase Starting Rates? 23

5.3. MAB Optimization vs. Static Experiment 24

6. Supplementary Analyses 25

6.1. Treatment Effects on the Completion Rate 25

6.2. Heterogeneous Treatment Effects 26

6.3. Excluding Prior Participants 27

7. Conclusion 28

References 32

Appendix 37

A. Extensions and Robustness 37

B. Invitation Message Alternatives 44

Tables

Table 1. Overview of Treatments 16

Figures

Figure 1. Stylized Experimental Procedure with Adaptive Randomization 19

Figure 2. Realized Cumulative Distribution Weights of Message Alternatives 20

Figure 3. Prediction of Starting Rates 22

Figure 4. AMCEs on the Starting Rate 24

Figure 5. AMCEs on the Completion Rate 25

Figure 6. AMCEs Depending on a Firm's Size 27

Figure 7. AMCEs Given No Prior Participation 28

Appendix Tables

Table A.1. Conjoint Analysis 37

Table A.2. Static Experiment vs. MAB Optimization 38

Table A.3. Number of Employees by Response Status 39

Appendix Figures

Figure A.1. Balancing Checks 40

Figure A.2. Spatial Distribution of Sent Invitations 41

Figure A.3. Share of Opened Invitations per Message Variant 42

Figure A.4. Predicted Difference in Starting Rates using OLS and Batched OLS 43

Figure B.1. Invitation Message P0A0U0D0M0 (Original Language, German) 44

Figure B.2. Invitation Message P1A1U1D1M1 (Original Language, German) 45

Figure B.3. Invitation Message P0A0U0D0M0 (English Translation) 46

Figure B.4. Invitation Message P1A1U1D1M1 (English Translation) 47