sorted order

2 replies [Last post]
userzht's picture
Offline
Joined: 02/19/2011

Hi all,

I was fitting an bivariate ordinal model, after the use of mxFactor(), I made the variables into ordered factors, like this

maxthresh <- 2
mzData$smoking1=mxFactor(mzData$smoking1, levels= c(0:maxthresh))

. After I ran the saturated (or ACE) model, the error appeared. It said that

In model 'Saturated_Fit' 'the thresholds in column 'smoking1' are not in sorted order.

Can anyone tell me what the reason is? Thank you in advance.

tbrick's picture
Offline
Joined: 07/31/2009
Threshold values must be strictly increasing down a column

Check your starting values. You want to make sure that the threshold values you've specified for this column are strictly increasing.

The thresholds specified in the objective function for each ordinal variable must be strictly increasing; the value at (z,y) must be greater than the value at (x,y) for all z > x.

If any two thresholds within the data column do not strictly increase row by row (or if two are identical), you'll see this error.

userzht's picture
Offline
Joined: 02/19/2011
got it.

Thank you very much.