These are footnotes to accompany the Introduction to LatentGOLD movies. We are very grateful to Jay Magidson of Statistical Innovations for providing us these very useful comments, as well as being so generous with his time to assist us in making these movies.

Footnotes for Tutorial 1

1. The first speaker mentioned that 'Ordinal-Fixed is not what we want for these data'. and that 'good vs. not good' should be specified as 'Nominal'. Actually, for dichotomous indicators Nominal and Ordinal 'scale type' settings give exactly the same model -- it's just that the labels for both categories will show up in the Parameters output with the Nominal setting, as opposed to just showing a single label for the variable with the Ordinal setting. For the 2 trichotomous variables, tutorial #1 illustrates the traditional latent class model where all variables are Nominal. 'Ordinal' also is appropriate for these trichotomous variables since the 2nd category does fall in-between the other 2 in both cases. We illustrate the ordinal setting for these on page 5 of Tutorial #2.
2. When examining the Advanced tab, this tab is present only in the Advanced version. Since only the sampling options are described, the impression is given that it contains only sampling options. Below the Sampling options section, it also contains options for adding various continuous random effects 'CFactors' to models, followed by options for setting up multilevel latent class models. Regarding the Model tab, it is mentioned that 'Cfactor Effects' are class independent. There are no CFactor effects in the models for tutorial 1. The section labeled 'Cluster independent' contain default options for more advanced models ('Error Variances and Error Covariances apply only to Cluster models containing continuous indicators). The checkbox for 'Order-restricted' could be checked if the resulting classes are to be restricted to be ordered. usually, this type of restriction is not desired.
3. When illustrating the FILE OPEN for a previously saved .lgf definition file -- it is not necessary to re-open the data file.