| dc.description.abstract | The concepts of longevity and life expectancy are fairly easy to understand. The longest the individual lives and typical age at death of individual are frequently used in such fields as Biostatistics, Demography, Economics, Engineering and Sociology. The challenges faced by researchers are that there are many different methods of estimating life expectancy. However, the different methods can give widely different results. Although there is rarely a correct method of brief demographic information, it is often possible to show methods that are clearly incorrect or give misleading results. The problem of life table is that following cohorts for a long period of time is not common, which prevents cohort analysis and the critical assumption of a stable-age distribution so difficult to meet. The alternative is to use survival analysis in the computation of life expectancy. Both life table and the alternative model are inconsistent and under-estimate life expectancy.
In this study, a preliminary analysis was made using SPSS to determine the mortality and survival profile of retired academic staff of universities. Samples of 225 and 304 retired staff members from Obafemi Awolowo Univesity (OAU) and University of Ibadan (UI) retirees were surveyed respectively. Early retirees were dropped from the sample, leaving those who retired at age 60 to 65 between the study dates of January 1977 to December 2012. Life table method was use to analyze the pattern of mortality for these data set. Life table analysis provides estimates of probabilities of surviving a given number of years after retirement. This technique allows subjects to enter (i.e., retire at age 60) or leave the study (i.e., die) at different points in time and it utilizes all the data on partial exposure to the risk of dying. It is non-parametric and requires no assumptions about the distribution of the survival function. The median survival life time of respondents from OAU, UI and combined data of both institutions are 18.78, 17.72 and 18.22 respectively. We can conclude that academic retirees’ median life expectancy from these institutions is around 18.24 years.
The survival probability and duration of service before retirement for UI and combined Institutions are significantly different whereas for OAU they are not significantly different. Survival probability and duration of service by gender before retirement is not significantly different. Meanwhile, there is no significant difference between survival rates of academic retirees and gender after retirement as it was supported by the estimate from the analyses of the study. We observed that there is significant difference between survival experience of university academic retirees and cadre on retirement as it was estimated in the study. In addition, there is no significant difference between post retirement occupation and survival probability of OAU academic retirees. But the reverse is the case for UI and combined UI and OAU data. We discovered that survival probability of the retirees and age on retirement are significantly different. In the study we noticed too that survival experience and spouse of university academic retirees are not significantly different. We noticed that survival trend of universities academic retirees follows the same pattern. The first two years after retirement probability of surviving was unity. At the 8th to 11th years of post retirement, probability of surviving were stable and there after fluctuating as post retirement year were increasing.
Application of the derived non-parametric and semi-parametric longevity models to cohort groups of academic retirees’ data were carried out. The estimated mean of post retirement years for academic retirees from derived longevity using Kaplan Meier model is 21.59 years and for Cox proportional model is 21.67. Meanwhile, the mean life expectancy from life table model is 18.33. Based on the analyses, we can say that Life table model is inappropriate for estimating life expectancy.
Derived parametric longevity models were applied to cohort groups of academic retirees. The estimated mean for post retirement years of the retirees from derived longevity using Exponential, Weibull and Gompertz Models without the effect of explanatory variables are 22.66, 22.35 and 22.08 years respectively. The estimated mean for post retirement years for these institutions with the effect of explanatory variables using Exponential, Weibull and Gompertz Models are 22.78, 22.82 and 22.48 years respectively. The outcomes of analysis from derived longevity models using parametric models correspond with the outcomes of derived longevity model using Kaplan Meier and Cox proportional models. | en_US |
