Data Analysis and Related Applications, Volume 2

19Assessing the Predictive Ability of Subjective Survival Probabilities

Subjective survival probabilities vary based on socio-demographic and health factors, and reflect our own views of future survival. The objective of this study is to investigate whether subjective survival probabilities can predict actual mortality. To achieve this objective, we introduce the concept of “Force of subjective mortality” and use it to estimate subjective survival probabilities related to a prediction time interval of two years. We then use a longitudinal dataset from the 6th and 7th Waves of the Survey of Health, Ageing and Retirement in Europe. For the statistical analysis, we employ binary logistic regression and Cox regression models. Our results show that subjective survival probabilities contain supplementary information about our own survival, in addition to socio-demographic, health and lifestyle factors, which are known to predict mortality. Moreover, the incorporation of “Force of subjective mortality” in the analysis allows us to estimate subjective survival probabilities for different prediction intervals, and therefore facilitates the use of data from consecutive SHARE Waves.

19.1. Introduction

Subjective survival probabilities are numbers reflecting individuals’ views on likely future survival and vary considerably based on the specific circumstances of respondents (Hamermesh 1985). Griffin et al. (2013) note that individuals consider their own experiences, health status, social ...