| Abstrakt: |
The article presents a comment on the work of the sociologists J.R. Udry C. Teddlie, and C.M. Suchindran, developing a model of random variation of vital rates in small geographical areas. The model is based on the division of the variance of vital rates into three parts; that which would occur in a homogenous population; that due to population heterogeneity; and that due to yearly fluctuations in the underlying basis for the mean value of the probability of the vital event in question. Using birth rates presented in Udry et al., a test of the new model demonstrates the importance of the third factor, yearly fluctuations, in determining year-to-year variation in birth rates. In spite of the substantial effect of the first term, the observed annual variation of the birth rate is mostly due to effects associated with the time heterogeneity of the underlying stochastic process. The proportions do not depend to any great extent on the exact estimate of childbirth intervals. The high variability of birth rates over time, then, should not primarily be attributed to the intrinsic random behavior of estimators based on relatively small population sizes, but rather to short-term changes in either the determinants of regional fertility or the completeness of the registration system. |
