Error: "if(any(longpars < LBOUND...missing value where TRUE/FALSE needed"

[marerikri]

I'm getting an error I'm a bit at of a loss about.

I ran this command:

mirt.2f <- mixedmirt(data = work.dat, model = model.2f, covdata = work.id, fixed =~ 0 + items, random = ~1|person, itemtype = "3PLu") where work.dat is a 2037395x27 binary (0 or 1, no missing) data frame; work.id is a 2037395x1 dataframe.

It runs for a few days and then at this point:

Stage 3 = 124, LL = -19181231.5, AR(1.30; 2.10) = [0.21; 0.30], gam = 0.0048, Max-Change = 8.2974

I get this message:

"Error in if (any(longpars < LBOUND)) longpars[longpars < LBOUND] <- (longpars0[longpars <  :
  missing value where TRUE/FALSE needed"

I can find the relevant code in the repository, but from a cursory look am not sure what is going on or how to fix it. Any help would be greatly appreciated; thanks.

[philchalmers]

The Max-Change above seems to indicate that your model is really unstable, so no doubt this is a hard crash. It's not the most elegant, and could probably be detected better. Can you provide a reproducible example?

[marerikri]

Unfortunately, probably not. The model takes a long time to run, and this error occurred after running for a couple of days at least.

The model (which I should have posted) is:

model.2f <- mirt.model( 
            "ne = 1,2,4,7,11,13,15,17,19,21,22,23,24,26,27
             pe = 3,5,6,8,9,10,12,14,16,18,20,25
             COV = ne*pe"
            )

The datafile consists of these 27 binary items observed on a number of individuals; there is quite a bit of heterogeneity in how many times each person responded to the 27 items. There are no missing data points, all coded 0-1.

The model above is a CFA; there is a similar 2-factor EFA that has been running for longer without error, with a much smaller max-change at this point, although it hasn't stopped.

[philchalmers]

You are more than welcome to send me some data directly as well as some source code to reproduce the issue. This code need not be presented publicly. Otherwise, if I have no way of reproducing the issue I cannot provide a patch.

[philchalmers]

I'm going to close this for now until a reproducible example can be provided.