近年來，基因與疾病之關聯分析 (association analysis) 越來越受到研究學者重視，因為在複雜性疾病與易感性基因之探討中 傳統的連鎖方法 (linkage method) 已不適用，所以複雜性疾病與易感性基因的關聯分析也蓬勃發展起來。在本文中我們主要是在探討 關聯分析中以家庭為研究資料與以群體為研究資料之間的優缺點，進而取長補短提出結合兩種資料之新的關聯分析方法 來增加估計與檢定之效力。我們同時考慮環境因素，探討基因因素與環境因素之交互作用。 本研究共分為三部份。第一部份探討如何整合病例-父母/病例-同胞 (case-parent/case-sibling) 與病例-對照 (case-control) 研究。我們提出一個加權最小平方 (Weighted Least Squares) 的方法將病例-父母/病例-同胞與病例-對照分析之估計式加以結合，以增進統計檢定之效力。 第二部分旨在探討基因-環境之交互作用。我們提出一個二階段研究設計法。在第一階段研究中，先收集病例資料; 在第二階段研究中，再收集其相對應之控制組資料。我們提出一個迴歸估計式以結合第一階段之單純病例分析(case-only analysis) 與第二階段之病例-對照分析。此建議之估計式即使在基因因子與環境因子 獨立之條件 (此條件為單純病例分析所必需) 不成立的情形下，依然可得出正確之統計推論。 第三部份旨在探討群體分層 (population stratification) 存在 之情形下，基因-環境之交互作用。我們提出一個二階段研究設計，以病例資料為第一階段資料， 再從病例資料中隨機抽取一部份病例患者之父母資料為第二階段資料。我們提出一個迴歸估計式結合單純病例研分析與病例-父母分析之估計式。 此新估計式即可整合單純病例分析與病例-父母分析，同時在群體分層存在之情形下，仍可得出有效之統計推論。 / In recent years, there are increasing attention to association studies, because linkage method will not be suitable under complex disease and susceptible genes. In the thesis, we are probing into association of family study and population study. And we combine family study and population study for increased efficiency of association method. We also consider interesting studies about gene-environment interactions. The thesis contains three projects. The first project focuses on examining when and how the two sources of information offered by such studies, one from the case-parent/case-sibling analysis, and the other from the case-control analysis with data from affected subjects and unrelated controls, can be integrated to enhance statistical power. We propose a weighted least-squares approach to linearly and optimally combine separate estimators from the case-parent/case-sibling and the logistic regression analysis for the association parameters. In the second project, we focus on examining the situation of gene-environment interaction. We propose a two-stage design. In the first stage, we collect patient data, and we seek out control data with respect to cases in the second stage. We propose regression analysis estimation in order to combine the case-only analysis in the first stage and the case-control analysis in the second stage. This estimation earns the correct statistical inference when genes and environment factors are not independent. In the third project, we explore gene-environment interactions under population stratification. We propose a two-stage design. In the first stage, we collect patient data, and we randomly collect a partial data of patient's parent from the cases in the second stage. We propose regression analysis estimation in order to combine the case-only analysis and the case-parent analysis. This estimation can combine the case-only analysis and the case-parent analysis, and attains effective statistical inference under population stratification.