CMStatistics 2017: Start Registration
View Submission - CMStatistics
Title: The intra and interclass correlation coefficients for measuring the familial relatedness in genetic studies Authors:  Ozge Karadag - Hacettepe University (Turkey) [presenting]
Serpil Aktas - Hacettepe University (Turkey)
Abstract: Most of the traditional statistical methods have been developed making certain assumptions about the dependency structure of the observations, such as independently identically distributed response variables. However, in many fields of applied sciences, especially in biomedical or epidemical researches, it is usually difficult to hold this assumption. In such fields, usually correlated data occurs. In the presence of correlated data, regarding the dependency between the observations may cause inappropriate analysis results. The objective is to measure the familial relatedness among the family members by using intra and interclass correlation coefficients. These correlation coefficients play an important role in estimating the degree of resemblance among family members with respect to the some characteristics, such as weight, blood pressure, age or cholesterol level. Basically, these correlations are Pearson type correlations, however in genetic studies, the degree of relationship is integrated to the estimators by assigning appropriate weights. A real family data set is used to investigate the familial relatedness among different relative pairs over 20 families with respect to the longitudinal blood pressure measurements. Familial correlations are estimated without any distributional assumptions.