search

paul-buerkner / thurstonianIRT

Fit Thurstonian IRT models in R using Stan, lavaan, or Mplus
GNU General Public License v3.0
31 stars 12 forks source link

How to omit blocks when analyzing a dataframe? #21

Closed leonardomose closed 4 years ago

leonardomose commented 4 years ago

I have a database with 22 blocks, three items each. However, when analysing the data I do not want to include 9 of this 22 blocks, as some of them have poor psychometric properties.

When I use the set_blocks command there is no argument to ignore a particular triplet that I don't want to include when analyzing the model.

Any idea how to solve this?

paul-buerkner commented 4 years ago

When adding the blocks together you could choose to simply not include the 9 unwanted blocks. Nobody forces you to add all 22 blocks just because you have them.

leonardomose commented 4 years ago

When adding the blocks together you could choose to simply not include the 9 unwanted blocks. Nobody forces you to add all 22 blocks just because you have them.

I know, but for example, when using the make_mplus_code function, the desired and unwanted triplets are not included in the NAMES section, not even with the exclamation point "!". This can cause a problem, because my dataframe includes 22 triplets, but the generated code only contains 13 triplets of them. In the "MODEL" section, variables start from i1 and go up to i39, which can also cause an error, since the bank goes up to i66.

paul-buerkner commented 4 years ago

I don't think this is a problem since thurstonianIRT takes care of all of this from the R side, that is passing only the right data etc. Unless there is something not actually working in which case I need a minimal reproducible example.

paul-buerkner commented 4 years ago

A small addition to better explain things. While Mplus requires all variables in the data to be used or at least named, thurstonianIRT does not. So it simply selects only the used variables and sends only them to mplus.