Method
We collect a monthly convenience sample of at least N=1000 from US Prolific workers (participants 18 years and over).
For January-June 2023 for each data collection wave we excluded all potential participants from the pool that participated in one of the previous waves (e.g., for March participants from January and February cannot participate).
Starting from July 2023, we only exclude participants from the wave before (e.g., for July participants from June cannot participate).
Prior to analysis, we exclude duplicate IPs from our sample to reduce the risk of bots and fraudulent answers (Waggoner et al. 2019).
For the results directly reported on the website we employ US representative sample weighting (post-stratification; Valliant 1993) based on the following variables obtained from the US Consensus Bureau (www.census.gov): Age, gender, and region (as of July 1, 2022).
References:
Waggoner, P. D., Kennedy, R., & Clifford, S. (2019). Detecting fraud in online surveys by tracing, scoring, and visualizing IP addresses. Journal of Open Source Software, 4(37), 1285
Valliant, R. (1993). Poststratification and conditional variance estimation. Journal of the American Statistical Association, 88(421), 89-96.