When survey respondents give more consistent answers after deliberation, the spread of their answers can shrink even if their opinions remain as far apart as before. The paper examines this possibility in a reanalysis of Deliberative Polls. In a sensitivity analysis, a before-to-after reliability ratio of 0.91 would account for the average observed convergence. Evidence from three UK Health attitude scales is mixed and does not establish whether reliability changed across the full sample.
Cite: Gaurav Sood. "Steadier, Not Closer: What Response Error Can Explain in Deliberative Polls." https://gsood.com/research/steadier-not-closer-separating-convergence-from-crystallization-in-deliberative-polls/
BibTeX:
@misc{soodsteadier,
author = {Gaurav Sood},
title = {{Steadier, Not Closer: What Response Error Can Explain in Deliberative Polls}},
note = {Working paper},
url = {https://gsood.com/research/steadier-not-closer-separating-convergence-from-crystallization-in-deliberative-polls/}
}