|
Message/Author |
|
Anonymous posted on Thursday, July 21, 2005 - 6:27 pm
|
|
|
I have 10 4-category ordinal items which are both part of a measurement model and also regressed onto covariates. On 3-4 items the last category is hardly used at all (< 1% in samples of 1000-1500). Is there any harm in leaving these as is or should one collapse the last two categories? |
|
|
If the sparse cells don't cause you any convergence problems, leave them as is. Otherwise collapse. |
|
Back to top |
|
|