Mixed item type in one IRT model

[Erin707]

Hi,

I am trying to learn IRT using your package TAM. It has been very helpful.
I got puzzled by the different item types in one IRT model.
There are 20 items in total. The first 16 are dichotomous, last 6 are polytomous.
For example, I first ran a 2PL model as such:

mod1<- tam.mml.2pl(data, irtmodel="2PL")
summary(mod1)

Then, I tried the tamaanify method and specified the item type for the first 14 items to be 2PL, and the last 6 to be GPCM.

tammodel1 <- "
LAVAAN MODEL:
  HOME =~ v1__v20
  HOME~~HOME

ITEM TYPE:
  v1__v14 (2PL) ;
  v15__v20 (GPCM);"

Estimate the model:
mod2 <- tamaan(tanmodel1, data)
Request the result:
summary(mod2)

mod1 and mod2 yielded the exact same results. I was wondering if I did something wrong here?
To be specific, I was trying to mimic the mirt package:

mirt(data, model=1, 
             itemtype = c(rep("2PL",14), rep("gpcm",6))

Could you please help me with this?

Thank you!
Erin

[thkiefer]

I can replicate that behaviour; seems to be a bug. For now, you can avoid it by passing tam.mml.3pl as method for the argument tam.method, i.e.

mod2 <- tamaan(tammodel1, data, tam.method = "tam.mml.3pl")

(@alexanderrobitzsch) The reason seems to be that in case of tam.mml.2pl being the selected method in tamaan.R the item type is extracted from the tamaan model but not passed to the call of tam.mml.2pl in line 64 to 70 of tamaan.R