In a regression context, the dichotomization of a continuousoutcome variable is often motivated by the need to express results in terms ofthe odds ratio, as a measure of association between the response and one ormore risk factors. Starting from the recent work of Moser and Coombs (Oddsratios for a continuous outcome variable without dichotomizing, Statistics inMedicine, 2004, 23, 1843-1860), in this article we explore in a mixed modelframework the possibility of obtaining odds ratio estimates from a regressionlinear model without the need of dichotomizing the response variable. It isshown that the odds ratio estimators derived from a linear mixed modeloutperform those from a binomial generalized linear mixed model, especiallywhen the data exhibit high levels of heterogeneity.
|Numero di pagine||12|
|Rivista||STATISTICAL METHODS & APPLICATIONS|
|Stato di pubblicazione||Published - 2008|
All Science Journal Classification (ASJC) codes