| Panel Conditioning and Scale Reliability: Evidence from the British Household Panel Study (2004) | |||||||||
Abstract | |||||||||
| Department Sociology Conference papers from the Department Sociology University Surrey Year Panel Conditioning and Scale Reliability Evidence from the British Household Panel Study Patrick Sturgis Nick Allum This paper posted UniS Scholarship Online http epubs surrey publsoc Panel Conditioning and Scale Reliability Evidence from the British Household Panel Study Patrick Sturgis and Nick Allum Department Sociology University Surrey Theoretical Background Panel conditioning can lead biased estimates marginal and structural parameters responses questions later rounds panel may influenced those given earlier waves participating income consumer studies changes consumer behaviour Pennel and Lepkowski people appear more consistent the responses over time due memory effect Jagodzinski Generally these effects have been found rather modest Holt Panel conditioning item reliability Less attention has been paid the effect panel participation the reliability scale items Good reason suspect reliability may not constant over time Survey interview may focus people attention the issues question People remember previous responses and strive for consistency over time Why does this matter Correction for error single item measures Setting Error Variance single item indicator will contain some degree random error common approach use previous research reasonable estimates specify error variance Hayduk The variance the item then corrected this amount each wave Should assuming stability these estimate | |||||||||
Publication details | |||||||||
| |||||||||