Modelación Bayesiana de datos multinivel en enfermedad periodontal.

Abstract

Hierarchical data is a data type with a hierarchical or nested structure. One level ordinary regression models assume independence of the observations that make up a group, however this is almost always not the case. That is, it is expected that the observations of the same group are similar. Using an ordinary (one-level) regression model on hierarchical data can lead to abnormal statistical inferences caused by an increase in the type 1 error rate. The increase in the type 1 error rate arises from not considering the units of highest level and only consider the measurements of the lowest level where the response variable is measured (there is an underestimation of the standard errors that makes the parameters significant when in reality they are not).