Nonparametric Evaluations of Familial Aggregation

R. Chakraborty

Research output: Contribution to journalArticlepeer-review


Limitations of detection of familial aggregation, examined by nonparametric aggregate measures (e.g., SMR), or by parametric methods (e.g., segregation analysis) may involve: (1) heterogeneity of risk patterns in different kindreds, or (2) restricted choice of parametric models. Nonparametric approaches dealing with individual kindreds, on the contrary, involve sparce data for which large sample theory may not apply. When the expected risk schedules of individuals are known, some optimal test procedures may be developed. Nevertheless, the optimality criteria are shown to be satisfied for only restricted choice of alternatives, and large sample approximation are not warranted in analysis of individual kindreds. Truncations of points of entry and exit of individuals in the study are needed to avoid complications due to loss of followup, in presence of which expected risk calculations are not reliable. Covariances of components of a goodness‐of‐fit test criterion detect specific modes of aggregation. It is illustrated that this test statistic is not always less powerful than the aggregate measures in detecting aggregation, and the covariances of its components partitioned by relative classes are in conformity with the results of segregation analysis.

Original languageEnglish
Pages (from-to)483-494
Number of pages12
JournalBiometrical Journal
Issue number4
StatePublished - 1988


  • Genetic epidemiology
  • Locally optimal test criterion
  • Standardized mortality (incidence) ratio


Dive into the research topics of 'Nonparametric Evaluations of Familial Aggregation'. Together they form a unique fingerprint.

Cite this