2021-06-08-皮尤-Measuring_the_Risks_of_Panel_Conditioning_in_Survey_Research_-_Pew_Research_Center_Methods_26页_1mb

Summary

The Pew Research Center study on panel conditioning in its American Trends Panel (ATP) found that conditioning does not contribute significant error to panel estimates. Key findings include minimal evidence of harmful bias in areas such as news consumption, political partisanship, and voting, though a slight uptick in voter registration was observed. Some reporting changes were noted, such as reduced media consumption, which may indicate improved data quality through better respondent honesty and rapport. Experiments and cohort comparisons showed minimal conditioning effects, with most changes being small and not statistically significant. The study concludes that current methods mitigate large-scale harmful effects and may enhance data quality over time.