One aim of statistical genetics is to determine the rough location of a disease causing gene relative to a genetic marker, a DNA sequence that has its position already known. This quest can be facilitated by genetic linkage analysis which infers relative positions of loci by examining patterns of inheritance in families. This thesis makes two contributions to linkage analysis. My initial work developed two visualization tools to explore the linkage likelihood data. The visualization tools are necessary because linkage data is enormous and 7-D making it difficult to understand. The first visualization tool displays 3D slices of the data by mapping three variables in three axes and fixing the other four variable values. A user can change both which variables to map on the axes and the values of other four variables. The second tool adopted a nested dimension technique called "Worlds-within-Worlds". A user first sees a point cloud of three dimensions, then clicks on any one point to see a new 3D view specified by the clicked point. This nested dimension technique provides intuitive navigation through the high-dimensional space. These visualization tools help researchers understand or reveal relations among parameters that otherwise would not have been easy to see. The second contribution of this thesis is a method to find subsets of pedigrees that share the same genetic mechanism. The visualization tools mentioned above helped statistical geneticists note the fact that the maximum HLOD values of individual families cluster around different locations in the likelihood space. Moreover, the general shapes of the likelihood surface for some families are quite similar while the shapes of the likelihood surface for other families are quite different. This suggested a new clustering algorithm to identify subsets of pedigrees sharing the same underlying genetic mechanism. I developed a pedigree clustering method called "Total Incremental Linkage Likelihood" (TILL) that sequentially adds the pedigrees to the existing pedigree set while maximizing the integral of the likelihood surface of newly formed pedigree cluster. Four classification schemes were examined and a series of simulations were designed to compare the performance of each classification. The TILL method performs better than the others in terms of sensitivity and specificity. A simulation also showed that the TILL method performs about 90% to optimal sensitivity and specificity. Some existing linkage detection methods that use the subset during the detection process require extra information such as age at onset or birthplace of each pedigree, but TILL does not require such information for finding pedigrees that share the same genetic mechanism.