Latent GOLD 統計分析軟體

New and improved features in Latent GOLD include:

 1. Improved and extended algorithms  

Latent GOLD® , already the fastest program for latent class modeling, is now even faster under our new enhanced EM algorithm implemented in Latent GOLD® 3.0!. Regression module now up to 3 times faster.

 2. Inactive covariate scale type  

This option allows you to see how the latent classes differ from each other on many exogenous variables (covariates) without allowing these covariates to influence the solution. By including many covariates in the model and specifying them as "inactive" (the new scale type), estimation takes only slightly longer than when omitting the covariates completely.
Benefits.

covariates appear in Profile output and in new Probmeans output of regression

output on classification based on covariates

 3. New covariate-based prediction/classification output  

Latent GOLD® parameters output includes gammas which are the parameters in a multinomial logit model that show how the (active) covariates included in the model can be used to predict class membership. This option allows you to output onto your data file predicted probabilities for each class plus the modal class assignment.

 4. Extensions in cluster/factor modules  

Bootstrap p-value
Latent GOLD® provides the option to compute the p-value using a bootstrap. This is important for sparse data where the usual p-value data is not valid.

direct effects can now be class-dependent

full multiple group analysis

class-specific local dependencies

new marginal and joint Profile output

communality/R-square for each indicator

 5. Improved technical options  

starting values: number of iterations per set can be increased

missing data allowed for covariates and predictors

option to suppress computation SEs and Wald statistics for large problems

 6. Extended interactive graphs  

new profile graphs for all scale types and covariates

new probmeans graphs for covariates in LC regression

7. Link to SPSS/SAS  

Latent GOLD® 3.0 is a stand-alone program that reads data from ASCII file formats as well as SPSS files. An additional add-on is available that links to 80 other file formats, including SAS and Excel.

 8. Extended save options  

all output can easily be saved to HTML and ASCII files

best set of starting values saved for easier recovery of previous results

 9. Extended Regression module  

new restrictions on coefficients: equalities between classes, fixed zeroes, and order constraints

replication weights in addition to case weights (frequencies) for repeated measures

multilevel type output: means, variances and individual coefficients

prediction statistics: overall and class-specific R-squared measures, output predicted values to a file

 10. Enhanced speed  

Latent GOLD® 3.0 is by far the fastest Latent Class product on the market.