Increasing Response Rates and Representativeness of Online Panels Recruited by Mail: Evidence from Experiments in 12 Original Surveys

Increasing Response Rates and Representativeness of Online Panels Recruited by Mail: Evidence from Experiments in 12 Original Surveys

By Alan Yan, Joshua Kalla, David Broockman
March 8,2017Working Paper No. 3653

Scholars have long noted the marked advantages of panel surveys for understanding the nature and causes of shifts in public opinion. Despite these advantages, existing panel survey approaches are often regarded as prohibitively costly or unrepresentative. In this paper, we present new evidence on best practices for an emerging panel survey approach: recruiting online panels by mail from defined sampling frames. We present results from 12 original such panel studies involving over 1 million individuals in total. Each of these panels contained randomized experiments on incentive levels, survey framing, and other features. These experiments allow us to present new recommendations about how to recruit panels using this mode that are more representative and more cost effective. We conclude by discussing limitations of our findings and their implications for other survey modes.