The item response theory - which method to choose

[badibad]

Hi everybody,
I have a question about choosing the correct method from the item response theory framework.
In my dataset i have 10 variables as follows: var 1 has a score ranege 0-6 p; var 2, 3, and 4 have a score range 0.5p; var 5, 6, and 7 have the score rande 0-3 p; var 8,9, and 10 have the score range 0 - 1 p.

I want to shrink the number of variables and thinking using the item response theory (sample size 2,500 people). However, i can not decide which method is more appropriate.

Right now i am thinking about Graded response model. BUT the problem is that i can not find good tutorial for R.

Can you give some advise?

Thank you.

 

[jamesmartinn]

Hi,

The ltm package can fit the graded response model - I used this a long time ago for this very purpose. You might want to also check out the mirt package, which can do uni and multidimensional irt.

https://cran.r-project.org/web/packages/mirt/
https://cran.r-project.org/web/packages/ltm/index.html

Cheers,

 

[badibad]

Thank you very much!