Contents
How are quasi likelihood estimators used in generalized estimating equations?
The quasi-likelihood estimators are estimates of quasi-likelihood equations which are called generalized estimating equations. A quasi-likelihood estimate of β arise from maximization of normality-based loglikelihood without assuming that the response is normally distributed.
How are Gee estimates of model parameters obtained?
In general, there are no closed-form solutions, so the GEE estimates are obtained by using an iterative algorithm, that is iterative quasi-scoring procedure. GEE estimates of model parameters are valid even if the covariance is mis-specified (because they depend on the first moment, e.g., mean).
What are the four structures of generalized estimating equations?
Here are four correlation structures: Independence – (correlation between time points is independent) ρ i j = c o r r ( Y i j, Y i k) for the i t h subject at times j and k. The quasi-likelihood estimators are estimates of quasi-likelihood equations which are called generalized estimating equations.
Is there a likelihood function in the Gee?
There is no likelihood function since the GEE does not specify completely the joint distribution; thus some do not consider it a model but just a method of estimation. Likelihood-based methods are NOT available for testing fit, comparing models, and conducting inferences about parameters.
How to use generalized estimating equations ( Gee )?
For model based output, we can still use overall goodness-of-fit statistics: Pearson chi-square statistic, X 2 , Deviance, G 2 , Likelihood ratio test, and statistic, Δ G 2 , Hosmer-Lemeshow test and statistic, and Residual analysis: Pearson, deviance, adjusted residuals, etc… Computationally more simple than MLE for categorical data
When to use generalized estimating equations ( Gee )?
Using the Generalized Estimating Equations is appropriate in this case. When you fit a model using GEE, you specify a correlational structure (such as AR (1)), and it can be quite reasonable that your data are independent conditional on both your covariates and the correlation matrix you specified.
When do Gee estimates of model parameters are valid?
GEE estimates of model parameters are valid even if the covariance is mis-specified (because they depend on the first moment, e.g., mean). However, if the correlation structure is mis-specified, the standard errors are not good, and some adjustments based on the data (empirical adjustment) are needed to get more appropriate standard errors.