setting starting values

61 views
Skip to first unread message

pda...@ucdavis.edu

unread,
Oct 27, 2016, 6:19:15 PM10/27/16
to mirt-package
Hi Phil, 

I am trying to set starting values for the a1 parameter in a multiple group model. I should qualify this by saying I am fairly new to the mirt package, so sorry if this questions is redundant or simple. In the simplest case, I want the starting values to be the same for both groups, but different for each item. I started by just trying to fix all loadings to the same value. Which brings me to my next question...the value I provided was a standardized factor loading. On what scale are the values provided to the START command? Anyway, I'm not even quite sure if I set this up correctly. But it did not work. I received this error: "Error in if (regexpr("-", tmp[j]) > 1L) { : missing value where TRUE/FALSE needed". I would eventually like to be able to provide different starting values for each item, once I can figure out how to get the code to work. Any help is much appreciated! The code is provided below:

newmodel <- 'F = 1-8
                      START = (1:8, a1, .9)'
mod <- mirt.model(mAAvW[ ,1:8], newmodel)
refmodel <- multipleGroup(data = mAAvW[ ,1:8], model = mod, group = mAAvW$group, SE = TRUE, 
                          invariance = c('free_means', 'free_var', itemnames))

Thanks,
Pega

Phil Chalmers

unread,
Oct 27, 2016, 11:06:04 PM10/27/16
to Pega Davoudzadeh, mirt-package
On Thu, Oct 27, 2016 at 6:19 PM, <pda...@ucdavis.edu> wrote:
Hi Phil, 

I am trying to set starting values for the a1 parameter in a multiple group model. I should qualify this by saying I am fairly new to the mirt package, so sorry if this questions is redundant or simple. In the simplest case, I want the starting values to be the same for both groups, but different for each item. I started by just trying to fix all loadings to the same value. Which brings me to my next question...the value I provided was a standardized factor loading. On what scale are the values provided to the START command? Anyway, I'm not even quite sure if I set this up correctly. But it did not work. I received this error: "Error in if (regexpr("-", tmp[j]) > 1L) { : missing value where TRUE/FALSE needed". I would eventually like to be able to provide different starting values for each item, once I can figure out how to get the code to work. Any help is much appreciated! The code is provided below:

newmodel <- 'F = 1-8
                      START = (1:8, a1, .9)'

See the ?mirt.model documentation here. The : operator is not appropriate, you probably want the dash - . 

  
mod <- mirt.model(mAAvW[ ,1:8], newmodel)
refmodel <- multipleGroup(data = mAAvW[ ,1:8], model = mod, group = mAAvW$group, SE = TRUE, 
                          invariance = c('free_means', 'free_var', itemnames))

Thanks,
Pega

--
You received this message because you are subscribed to the Google Groups "mirt-package" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mirt-package+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply all
Reply to author
Forward
0 new messages