Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
NCTM or The Math Forum.



URGENT Help needed with LCA model selction
Posted:
Jun 29, 2006 7:31 PM


Hello. I am working on my MS thesis, and a committee member and I are disagreeing about how I have selected the best model. Based on my readings and the LatentGold LCA manual, I think this is reasonable. I would appreciate anyone agreeing or disagreeing with this method and letting me know why (time is of the essence for me to complete my thesis). Thanks! Danielle
Quick overview of project: I sought to determine if the symptoms of individuals experiencing a major depressive episode (MDE), with a diagnosis of recurrent unipolar or bipolar I disorder, would cluster into meaningful and clinically interpretable subgroups of depression. Outpatients (N = 293; Unipolar n = 195, Bipolar n = 98) who met DSMIV diagnostic criteria for a MDE were assessed at intake to treatment with
the 25item Hamilton Rating Scale for Depression 25item (HRSD25). I
applied latent class analysis (LCA) to the 25 HRSD items in the unipolar depressed sample and then the unipolar sample combined with the bipolar I depressed sample.
Data analysis plan: I entered the HRSD25 items as coded, ordinal variables (rather than dichotomized absent/present). I applied LCA to a
195row x 25column data matrix and 293row x 25column data matrix with Latent Gold 4.0. For each model, gender was entered as an active covariate.
I first fit a oneclass solution followed by a twoclass solution, and so on, until I reached a sevenclass solution, or the "best" solution. I stopped the analyses at seven classes because I felt that more than seven classes would provide little clinical or practical significance. I defined the best solution by the following criteria: (1) the solution fit the data significantly better than the previous solution and (2) the estimated parsimony of the model was associated with a significant pvalue.
First, I estimated the parsimony of each possible model solution, using
the bootstrap pvalue test statistic. Since I have a small sample size
and the chisquared estimation may be problematic, I used the bootstrap
pvalue. Then, I sought to determine if each level of a restricted model (a solution with a greater number of classes) was a significant improvement of the less restricted model (a solution with a smaller number of classes). I thought that if the bootstrap 2LL Diff test statistic has a significant pvalue, then the less restricted model is more parsimonious than the more restricted model. For instance, if the
bootstrap 2LL Diff statistic is not significant, then a 3class solution may be accepted as a good fit and as a more parsimonious model
than a 2class solution. Conversely, if the bootstrap 2LL Diff pvalue is significant, then a 2class solution should be accepted as the more parsimonious model. Next, I evaluated how well the selected model permitted predictions of class membership based on the observed indicator variables with entropy Rsquared.
Results: Based on the Bootstrap 2LL Diff statistic, a sixclass solution was the most parsimonious solution and was able to adequately predict class membership (entropy R2 = 0.85). The Bootstrap 2LL Diff statistic between a 6 and 7class solution was significant. This
suggests that the less restricted 6class model was preferrable to the more restricted 8class model. In the combined unipolarbipolar sample, a 5class solution was the most parsimonious solution and was able to adequately predict class membership based on the HRSD25 items
(entropy Rsquared = 0.87).
UNIPOLAR MODEL L2 Bootstrap p Entropy R2 BIC (L2) Bootstrap 2LL Diff 1 3457.31 .33 1.0 2592.54 2 3262.55 .25 .91 2534.87 3 3174.07 .20 .83 2583.49 4 3098.52 .17 .84 2645.04 5 3027.51 .07 .86 2711.13 6 2978.08 .03 .85 2798.80 6v5, p = .10 7 2930.77 .01 .87 2888.59 7v6, p = .04
BIPOLAR MODEL 1 5310.86 .32 1.0 3834.01 2 4976.12 .27 .93 3681.04 3 4835.15 .18 .86 3721.83 4 4742.17 .10 .84 3810.62 5 4635.77 .05 .89 3885.99 6 4568.08 .04 .87 4000.07 6v5, p = .01 7 4507.73 .01 .86 4121.48 7v6, p = .00



