The impact of attrition and non-response in birth cohort studies: a need to incorporate missingness strategies

Tarek Mostafa, Richard Wiggins

Abstract


This paper reveals the extent of attrition in the British Cohort Study begun in 1970 (BCS70) and how it affects sample composition over time. We examine the determinants of response and then construct inverse probability weights (IPWs) to adjust for sample loss. Secondly, we create a hypothetical substantive data set from BCS70 across data collection waves 3 and 4 to illustrate the effectiveness of the use of weights and multiple imputations (MI) in handling the impact of unit non-response and item missingness respectively. Our findings show that when the predictive power of the response models is weak, the efficacy of non-response weights is undermined. Further, multiple imputations are effective in reducing the bias resulting from item missingness when the magnitude of the bias is high and the imputation models are well specified.


Keywords


BCS70, attrition, unit non-response, item non-response, weights, imputation.

Full Text:

PDF


DOI: http://dx.doi.org/10.14301/llcs.v6i2.312

Copyright (c) 2015