本文以加拿大安大略省漢密爾頓地區為實例，研討分析小地方之間死亡率之差異。因為官方的死亡資料不夠詳細，我們花了很多時間首先把個人的死亡登記依地點、性別、和年齡分類，然後把得到的死亡數與相關的人口配合而算出各地的死亡指數。我們也利用非線性的統計方法來解釋漢密爾頓地區內的死亡差異。 我們的主要體現如下：(1)小地方間之死亡差異很大；(2)中位家庭所得能解釋此差異之百分之四十六；(3)其他被我們探用的社經因素，與中位家庭所得合用於多變數模式時，都顯得不重要。 本文報告的只是初步的結果。為促使小地方死亡差異研究的加強，本文最後提了一些實際的意見供同行的學者參考。

This paper deals with the problems of measuring and explaining local mortality variation, based on a case study of the Hamilton region in the Province of Ontario, Canada. Due to lack of detailed official data, the local mortality pattern is established after we have spent much time in recording information directly from individual death registrations and in matching the number of deaths by age and sex with the relevant at-risk population for each of the 96 census tracts in the Hamilton region. To explain local mortality variation by socioeconomic variables, we use the logit model and the maximum quasi-likelihood estimation method. We find that local mortality variation is substantial in our study area; that median family income can explain nearly half of the mortality variation among the census tracts; and that other socioeconomic variables (e.g. widowhood and divorcehood), which appear to be significant variables of mortality variation in bivariate and non-spatial analyses, turn out to be insignificant in our multivariate analysis using the logit model. Our results are by no means definitive. To help promote further research on local mortality variation, we conclude our paper by providing what we hope to be helpful and practical suggestions.