The TCALIS Procedure |
FITINDEX Statement |
You can use the FITINDEX statement to set the options for computing and displaying the fit indices, or to output the fit indices. All but the OFF= and ON= options of the FITINDEX statement are also available in the PROC TCALIS statement. The options set in the FITINDEX statement will overwrite those set in the PROC TCALIS statement.
For the listing of fit indices and their definitions, see the section Overall Model Fit Indices. Note that not all fit indices are available with all estimation methods, which is specified by the METHOD= option of the PROC TCALIS statement. See the section Fit Indices and Estimation Methods for more details.
The options of the FITINDEX statement are as follows:
specifies a confidence interval () for the Browne and Cudeck (1993) expected cross validation index (ECVI). See the ALPHAECV= option of the PROC TCALIS statement.
specifies a confidence interval () for the Steiger and Lind (1980) root mean square error of approximation (RMSEA) coefficient. See the ALPHARMS= option of the PROC TCALIS statement.
specifies a correction factor for the chi-square statistics for model fit. See the CHICORRECT= option of the PROC TCALIS statement.
defines the criterion value p for indicating a close fit. See the CLOSEFIT= option of the PROC TCALIS statement.
reduces the degrees of freedom of the test by . See the DFREDUCE= option of the PROC TCALIS statement.
turns off the automatic adjustment of degrees of freedom when there are active constraints in the analysis. See the NOADJDF option of the PROC TCALIS statement.
disables the display of index types in the fit summary table. See the NOINDEXTYPE option of the PROC TCALIS statement.
turns off the printing of one or more fit indices or modeling information as indicated by names, where a name represents a fit index, a group of fit indices, or modeling information. Names must be specified inside a pair of parentheses and separated by spaces. By default, all fit indices are printed. See the ON= option for the value of names.
turns on the printing of one or more fit indices or modeling information as indicated by names, where a name represents a fit index, a group of fit indices, or modeling information. Names must be specified inside a pair of parentheses and separated by spaces. Because all fit indices and modeling information are printed by default, using an ON= list alone is redundant. When both ON= and OFF= lists are specified, the ON= list will override the OFF= list for those fit indices or modeling information that appear on both lists. If an ON(ONLY)= list is used, only those fit indices or modeling information specified in the list will be printed. Effectively, an ON(ONLY)= list is the same as the specification with an ON= list with the same selections and an OFF=ALL list in the FITINDEX statement.
Output Control of Fit Index Groups and Modeling Information Group
You can use the following names to refer to the groups of fit indices or modeling information available in PROC TCALIS:
Absolute or stand-alone fit indices that measures the model fit without using a baseline model.
All fit indices available in PROC TCALIS.
Incremental fit indices that measure model fit by comparing with a baseline model.
General modeling information including sample size, number of variables, number of variables, and so on.
Output Control of Modeling Information
You can use the following names to refer to the individual modeling information available in PROC TCALIS:
Degrees of freedom of the chi-square statistic for the independence model.
Number of elements in the moment matrices being modeled.
Number of observations assumed in the analysis.
Number of independent parameters.
Number of variables.
Output Control of Absolute Fit Indices
You can use the following names to refer to the individual absolute fit indices available in PROC TCALIS:
Chi-square statistic for model fit.
Hoelter’s critical N.
Percentage contribution to the chi-square value for multiple-group analyses.
Degrees of freedom for the chi-square test for model fit.
Elliptical chi-square statistic for ML and GLS methods in single-group analyses without mean structures.
Optimized function value.
Goodness-of-fit index by Jöreskog and Sörbom.
P-value of the chi-square statistic for model fit.
Root mean square residual.
Standardized root mean square residual.
Z-test of Wilson and Hilferty.
Output Control of Parsimonious Fit Indices
You can use the following names to refer to the individual parsimonious fit indices available in PROC TCALIS:
Adjusted GFI.
Akaike information criterion.
Bozdogan corrected AIC.
McDonald centrality measure.
Expected cross-validation index.
Lower confidence limit for ECVI.
Lower confidence limit for RMSEA.
Parsimonious GFI.
Probability of close fit.
Root mean squares of error approximation.
Schwarz Bayesian criterion.
Upper confidence limit for ECVI .
Upper confidence limit for RMSEA.
Output Control of Incremental Fit Indices
You can use the following names to refer to the individual incremental fit indices available in PROC TCALIS:
Bentler comparative fit index.
Bentler-Bonett normed fit index.
Bentler-Bonett non-normed fit index.
Bollen normed fit index (Rho1).
Bollen non-normed fit index (Delta2).
James et al. parsimonious normed fit index.
creates an output data set containing the values of the fit indices. This is the same as the OUTFIT= option of the PROC TCALIS statement. See the section OUTFIT= SAS-data-set for details.
Note: This procedure is experimental.
Copyright © 2009 by SAS Institute Inc., Cary, NC, USA. All rights reserved.