Proxy respondents in longitudinal ageing studies: Can they help to reduce bias in ageing research?

Ageing surveys often face the challenge of potential underestimation of the health status of older populations, particularly in panel studies and for the 'oldest old'. This bias arises from the over-representation of healthier individuals, who typically exhibit higher participation rates and lower attrition, skewing the data towards a healthier-than-average depiction of the population. As a result, the data may not accurately reflect the true health status of the population, leading to under-reporting of health decline or disability.
To mitigate this problem, proxy respondents are often employed when the primary respondent is unable to provide answers – typically due to cognitive or physical limitations. The use of proxies allows the continued participation of individuals who would otherwise be excluded from the study, maintains the integrity of longitudinal datasets, and reduces selective attrition. However, concerns persist regarding the validity of proxy-reported data. The extent to which proxy responses diverge from self-reported data remains uncertain, with mixed evidence suggesting that for instance proxies may either overestimate or underestimate certain aspects of respondents’ health status. A deeper understanding of how proxies affect different types of data, and addressing the ethical implications of excluding vulnerable populations, will be essential to improving methodologies and ensuring the quality of research on older adults. 
Against this background, we are using data from Wave 9 of the Survey of Health, Ageing and Retirement in Europe (SHARE) to analyse the role of proxies on data quality. Preliminary results show that the socio-demographic and health characteristics of respondents who require assistance to answer questions in an ageing survey differ from those of regular respondents. Moreover, excluding proxy responses can alter the interpretation of key findings, such as the explanation of self-perceived health—a critical construct in quantitative health research. This study highlights the importance of incorporating proxy data to enhance both the validity and inclusivity of ageing studies.