Have an idea?

Visit Sawtooth Software Feedback to share your ideas on how we can improve our products.

Calculating df for chi square in LC analysis

I wondered how you are calculating the degrees of freedom in latent class scenarios. Is it:
[total number of levels] - [number of attributes] x [number of groups] + [number of groups - 1]?
asked Jun 28, 2013 by strix (140 points)
retagged Jun 28, 2013 by Walter Williams

1 Answer

+3 votes
For the Chi-square value we use: [log likelihood] - [null log likelihood]  * 2.

For other values such as BIC, we calculate the number of free model parameters using [number of groups] * [number of variables] + [number of groups] - 1.  The number of variables depends on the attribute codings.
answered Jun 28, 2013 by Walter Williams Gold Sawtooth Software, Inc. (20,005 points)