Message/Author 

Ben Chapman posted on Saturday, March 24, 2007  12:43 pm



HiI have just purchased Mplus and fit a few path models in which observed outcomes are variously continuous, ordered categorical, and poisson. Estimator is MLR. While there is an Rsquare for the ordered categorical outcome (the source of which I see referenced on the board here), none is generated for the poisson outcome. Is it possible to obtain some sort of (pseduo?)Rsquare estimate for the poisson outcome? Thanks. 


I haven't seen this done and I cannot think of a natural way to do this. Other readers? 


Thanks for the quick reply. I was just curious, having seen seen deviancebased Rsquare measures for poisson models, i.e. Waldhör et al 1998, but maybe it's not applicable here or wouldn't be a good idea. Thanks again. 


That's a possible approach. 

Jon Elhai posted on Thursday, March 29, 2007  12:42 am



In answer to Ben Chapman's question about generating pseudo Rsquared values for poisson regression... If you have access to a general purpose stats software program like Stata, for estimating regression (albeit, not for a path model), you can generate pseudo rsquared values for poisson, as well as negative binomial regression, and zeroinflated and zerotruncated versions of these models. If you have access to Stata, this is accomplished by using the "fitstat" ado userwritten supplemental file. Again, this is for regression  not for path models. 


Dear Prof. Muthen, I am running a zeroinflated negative binomial model. The model fits very well. Reviewers are asking for pseudoRsqr to be reported. May I kindly request you to guide me to the formula which I can use to calculate pseudoRsqr from the output values produced by MPlus please. As I am not familiar with other software like STATA it would be great if you would advice to calculate this value within the MPlus environment itself. Thanking you so very much in advance. Respectfully, Arun. 


The negbin book by Hilbe (2011, 2nd ed.) gives the pseudoR2 statistic on page 65. This is easily obtained by Mplus by doing two runs (see also Hilbe, pp. 6566), one without the covariates (interceptonly model) and one with the covariates. Then R2 = 1  LLF/LLI, where LLF is the loglikelihood value for the model with covariates and LLI is the loglikelihood value for the interceptonly model. This type of R2 is also used with binary outcomes and called the McFadden R2. 


Respected Prof Muthen, Thank you so very much for your quick reply. My sincere gratitude to you. Respectfully, Arun. 


Professor Muthen, In attempting to calculate the McFadden R2 for negbin how do I request an intercept only model? I thought this would be done by simply running without any covariates but my loglilihood for this model with only the count outcome is larger than my full model. This seems incorrect to me and thus I suspect I am not correctly running the intercept only model. I greatly appreciate any clarification you can offer. 


Run the model with all the covariates, but fix their slopes at zero. 


Thank you. 

Back to top 