Attachment | Size |
---|---|

diabetes analysis for metSEM forum.R [6] | 2.06 KB |

diabetes4.csv [7] | 1.15 KB |

Hi metaSEM users,

As I have posted previously in this forum, I am using metaSEM to conduct a network meta-analysis. I have been attempting to fit a metaSEM model using both ML and REML estimation using some sample data from a meta-analysis of the effectiveness of diabetes interventions. I was able to successfully conduct the analysis using ML estimation, but for some reason when I attempt to obtain a fit using the reml() function, I obtain the following error:

Running Variance component with REML

Error in running the mxModel:

Error in reml(trt, S, RE.constraints = con) :

trying to get slot "runstate" from an object (class "simpleError") that is not an S4 object

In addition: Warning message:

The job for model 'Variance component with REML' exited abnormally with the error message: Objective function returned an infinite value at iteration 0.1.

What's particularly strange is that I have been able to successfully obtain ML and REML model fits using these same data using the mvmeta routine in STATA. Moreover, I am able to exactly replicate the results from the ML fit using the meta() function in metaSEM in STATA. So, I don't think that the error I observed is due to some weirdness in the data.

I have attached the code and data that generated this error. Is it possible that there is a bug in reml() that is responsible for this error?

Thanks in advance for your help.

Kind regards,

- Patrick