Speed Accuracy Hierarchical Model PreviousNext
Mplus Discussion > Multilevel Data/Complex Sample >
Message/Author
 Maya Greene posted on Monday, March 09, 2015 - 9:32 am
I am intersted in fitting Fox et al (2007) model of speed and accuracy. Accuracy is modeled using a 2 parameter IRT model and speed is modeled using a lognormal model. The two models are combined in a hierarchical framework. http://www.jstatsoft.org/v20/i07/paper

What is the script for doing this in Mplus?

Thanks!
 Tihomir Asparouhov posted on Tuesday, March 10, 2015 - 1:21 pm
Here is a sample code with 3 items.

variable:
names=U1-U3 T1-T3;
categorical=U1-U3;
usevar=U1-U3 logT1-logT3;

define:
logT1=log(T1);
logT2=log(T2);
logT3=log(T3);

model:
theta by U1-U3; ! accuracy
speed by logT1-logT3; ! speed
 Maya Greene posted on Wednesday, March 11, 2015 - 3:09 pm
Hi Tihomir, thanks for your answer. I am confused about how the multilevel structure of the data (trials nested within person) would be modeled here.
 Tihomir Asparouhov posted on Thursday, March 12, 2015 - 10:20 am
You can add multilevel model but the paper you referred to doesn't have a multilevel model. Items are nested within person. I don't see multiple trials by the same person on the same item. The hierarchical discussion in that paper seems to be regarding priors. In Mplus you can do the above model using ML, Bayes or WLSMV estimators.

Take a look at Mplus user's guide examples 9.6 and 9.7 for possible multilevel modeling.
 Maya Greene posted on Thursday, March 12, 2015 - 6:25 pm
Thank you for the reference. I looked at the examples and I believe I'll be able to work something out with them.
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: