Recoveries from Mobility Difficulty or Reporting Errors? A Revisit on Measures of Mobility Limitations using the Health and Retirement Study

Objectives: Clark et al. (1998) have noticed that the incidence of recovery from mobility limitations is substantially greater (more than twice) than that of onset among older Americans aged 51-61, using the first two waves of data from the health and retirement study (HRS). This observation is a bit counter intuitive given the age of the respondents their study targeted at. We try to understand the reason behind it by differentiating between real recoveries from mobility limitation and measurement issued caused by reporting bias.

Methods: Our panel data analysis is performed using a sample of more than 5500 respondents from the HRS data set, which covers a period from year 1992 to year 2002. Measures suggested by Wallace et al. (1995) on individual mobility limitations are constructed based on individual self-reports on limitations to activities of daily living (ADL).

Results: After controlling for individual SES, health status and health behaviors, we find an increasing trend in the incidence of onset of mobility limitations with individual aging. We have also found the unusual incidence of “recoveries” only occurred between the first and second wave, but did not last in the following waves. This could be explained either by measurement errors caused by reporting bias due to confusion with certain survey questions, or by changes in individual healthy behaviors resulted from better health knowledge after interviews.

Conclusion: Our study calls caution on the use of appropriate measures on mobility limitation when longitudinal data are available.