Tong & Zhang (2012)
Distribution diagnostics for robust growth curve models
Citation
Tong, X., and Zhang, Z. (2012). Diagnostics of Robust Growth Curve Modeling using Student's t Distribution. Multivariate Behavioral Research,47(4), 493518.
Software
Click the link below to use the online software:
http://webstats.psychstat.org/models/1gcm/robdiag/
Instructions
The main interface of the software looks like
The following steps can be followed for analysis.
Step 1
Click on the Browse or Choose File button depending on the browser you use. Upload the data file from your local computer. Make sure that your data file is in free format and is either a .txt or .dat file.
The data in the file should be organized as a matrix with data from the first occasion on the first column, data from the second occasion on the second column, and so on. For example, data for the first ten subjects which are measured at four occasions may look like this:
3.241 5.355 9.217 11.69 3.071 5.967 6.31 8.228 6.133 9.297 8.162 11.01 5.106 5.323 6.429 6.882 6.282 6.202 9.934 11.19 5.811 8.756 10.04 11.24 3.799 8.967 17.37 23.91 2.664 3.917 6.093 6.981 9.558 7.701 10.31 12.83
Step 2
Check the button before each method to select the method which is going to be applied. Notice that you can only use one method at a time.
For the second method (distribution comparison based on DIC), one can input a nonnegative number as the degrees of freedom. The default 0 requests the estimation of the degrees of freedoms.
For the second and the third method, The burnin period and the length of Markov chain can be controlled by supplying them in the respective fields.
Step 3
Click on the Submit button to submit your job.
Then, You may view the results later through a link on the screen like http://webstats.psychstat.org/models/1gcm/robdiag/results.php?url=33eca245c87ada431e67eae933ed5ab0 later. Or you can view the results by clicking on the button “I cannot wait! View the results now!”
An example

Sample output

distribution checking based on individual growth curve analysis

distribution comparison based on Deviance Information Criterion (DIC)

post hoc checking of degrees of freedom estimates for t distributions
