You are not currently logged in.
Access JSTOR through your library or other institution:
Demographic Time Series Modelling of Total Deaths in Nigeria
D.B. Adekanmbi, F.J. Ayoola and A.O. Idowu
Southern African Journal of Demography
Vol. 15, No. 1 (January 2014), pp. 21-48
Published by: Population Association of Southern Africa
Stable URL: http://www.jstor.org/stable/soutafrijourdemo.15.1.21
Page Count: 28
You can always find the topics here!Topics: Time series models, Demography, Mortality, Time series, Analytical forecasting, Statistical models, Modeling, Forecasting models, Time series forecasting, Death
Were these topics helpful?See somethings inaccurate? Let us know!
Select the topics that are inaccurate.
Preview not available
ABSTRACT This study presents an approach to modelling total deaths based on statistical time series techniques as proposed by Box and Jenkins, for demographic analysis of total deaths in Nigeria, spanning the period 1995 to 2007, with a view of uncovering possible seasonality in the series. The Boot-Fiebes-Lisman first difference (BFL-FD) method was employed in disaggregating the annual total death series into quarterly figures, to uncover the trend, pattern and seasonality of deaths in Nigeria. The disaggregated series became stationary as confirmed by the value of the Dickey-Fuller test of nonstationarity, after subjecting the series to Box-Cox transformation and non-seasonal differencing. The Correlogram visual analysis method of model identification was employed in identifying suitable seasonal ARIMA models for the transformed death figures. Two univariate ARIMA models tagged model A and model B were fitted to the disaggregated death figures. The models were used in forecasting for period 2001 to 2003, and the forecasts were compared with the actual death figures. Satisfactory diagnostics results of the ARIMA models confirmed the adequacy of the models. It was discovered that death cases in Nigeria lack seasonal pattern as generally speculated of ember months. The forecast intervals, though, signalled that further precision was quite spurious; nevertheless the demographic time series models are useful in gaining insight into the pattern and trend of deaths in Nigeria. Recommendations were made on achieving effective health administration in Nigeria, and also future research issues on incidences of death were discussed.
© 2014 Population Association of Southern Africa