What’s New in SAS/STAT 9.3

The experimental FMM procedure fits statistical models to data where the distribution of the response is a finite mixture of univariate distributions. These models are useful for applications such as estimating multimodal or heavy-tailed densities, fitting zero-inflated or hurdle models to count data with excess zeros, modeling overdispersed data, and fitting regression models with complex error distributions.

PROC FMM fits finite mixtures of regression models or finite mixtures of generalized linear models in which the regression structure and the covariates can be the same across components or different. Maximum likelihood and Bayesian methods are available with the FMM procedure.

More information about the changes and enhancements follows. Details can be found in the documentation for the individual procedures in the SAS/STAT 9.3 User’s Guide.

Producing graphs with ODS Graphics no longer requires a SAS/GRAPH® license. In addition, the family of statistical graphics procedures (SGPANEL, SGPLOT, SGRENDER, and SGSCATTER) has moved from SAS/GRAPH to Base SAS® license.

The MAXPOINTS= option has been added to the ANOVA, CLUSTER, GLM, LOGISTIC, MIXED, QUANTREG, and VARCLUS procedures. This option specifies a limit for the number of points that can be displayed on certain plots, and these plots are not created when this limit is exceeded. Note that the REG procedure already provided this option.

The frequency plots and cumulative frequency plots of PROC FREQ and the weighted frequency plot of PROC SURVEYFREQ are no longer produced automatically when ODS Graphics is enabled. You can request these graphs with the PLOTS= option.

In SAS 9.3, the default destination in the SAS windowing environment is HTML; in addition, ODS Graphics is enabled by default in the SAS windowing environment. These new defaults have several advantages. Graphs are integrated with tables, and all output is displayed in the same HTML file using a new style. This new style, HTMLBLUE, is an all-color style, which is designed to integrate tables and modern statistical graphics. You can view and modify the default settings by selecting Tools Options Preference from the menu at the top of the main SAS window. Then click the Results tab.

The following features are now production:

In addition, PROC CALIS now provides detailed analysis of the missing patterns with the FIML estimation method. With the COVPATTERN= and MEANPATTERN= options, you can specify various standard mean and covariance patterns by using keywords. PROC CALIS then generates the required covariance and mean structures automatically.

The CLUSTER procedure now produces a dendrogram by default when ODS Graphics is enabled. The MAXCLUS= option enables you to right-truncate the CCC, PSF, and PST2 plots to improve readability. The MAXPOINTS= option enables you to suppress the dendrogram when there is a large number of clusters.

The EFFECT statement is now production. This statement is available in the HPMIXED, GLIMMIX, GLMSELECT, LOGISTIC, ORTHOREG, PHREG, PLS, QUANTREG, ROBUSTREG, SURVEYLOGISTIC, and SURVEYREG procedures.

The NATURALCUBIC option specifies a natural cubic spline basis for the spline expansion.

The CLUSTER option modifies the box plot display by displaying a plot for each level of the SLICEBY= classification variable.

The FREQ procedure now produces agreement plots when the AGREE option is specified and ODS Graphics is enabled. It also offers a number of alternative confidence limits for the proportion difference, and it provides exact unconditional confidence limits for the proportion difference that are based on the Farrington-Manning score statistic.

The EXACTMAX option in the MODEL statement limits the number of response values for exact Poisson regression.

The EFFECT statement is now production.

The GLMPOWER procedure now produces its graphs with ODS Graphics.

The GLMSELECT procedure now provides a STORE statement which enables you to save the context and results of the statistical analysis for further processing with the PLM procedure.

The MODELAVERAGE statement, which specifies model selection on resampled subsets of the input data, is now production.

The EFFECT statement is now production.

The HPMIXED procedure now provides the REPEATED statement, which defines the repeated effect and the residual covariance structure in the mixed model. The AR(1), CS, CSH, UC, UCH, and UN covariance structures are now available with the TYPE= option in the RANDOM statement.

The EFFECT statement is now production.

The X axis tick marks are now aligned with the at-risk values in the survival plot.

You can now request that standardized residuals be saved in the output data set. In addition, the STDRES suboption of the INFLUENCE option in the MODEL statement includes standardized residuals and likelihood residuals in the resulting display. The FITSTAT option in the SCORE statement produces the AIC, SBC, RSq, AUC, and Brier score fit statistics. Additionally, the ODDSRATIO statement and the CLDISPLAY= suboption of the CLODDS option control the appearance of the confidence limit error bars.

The EFFECT statement is now production.

The new RANDOM statement simplifies the construction of hierarchical random-effects models and significantly reduces simulation time while improving convergence, especially in models with a large number of subjects or clusters. This statement defines random effects that can enter the model in a linear or nonlinear fashion and supports univariate and multivariate prior distributions.

In addition to the default Metropolis-based algorithms, PROC MCMC now takes advantages of certain forms of conjugacy in the model in order to sample directly from the target conditional distributions. In many situations, the conjugate sampler increases sampling efficiency and provides a substantial reduction in computing time.

The MCMC procedure now supports multivariate distributions including the Dirichlet, inverse Wishart, multivariate normal, and multinomial distributions.

The experimental FCS statement specifies a multivariate imputation by fully conditional specification (FCS) methods. For data with an arbitrary missing data pattern, these methods enable you to impute missing values for all variables, assuming that a joint distribution for these variables exists. The FCS method requires fewer iterations than the MCMC method.

The STOUFFER option in the PROC statement produces adjusted p-values by using the Stouffer-Liptak combination method.

The NLIN procedure provides several experimental features for diagnosing your nonlinear model fit, including the PLOTS, NLINMEASURES, and BIAS options in the PROC NLIN statement, in addition to producing observation-wise statistics in the OUTPUT data set. The PLOTS option enables you to plot the fitted model, fit diagnostics, tangential and Jacobian leverage, and local influence. The NLINMEASURES displays global measures of nonlinearity, and the BIAS option computes Box’s bias statistics for the parameter estimates. Finally, you can add the leverage, local influence, and residual diagnostics in the output data set that is produced with the OUTPUT statement.

The EFFECT statement is now production.

The PHREG procedure now fits frailty models with the addition of the RANDOM statement. You often use frailty models when you analyze clustered data and want to account for the within-cluster correlation with random effects. In addition, the NLOPTIONS statement is available with PROC PHREG, and the Zellner g-prior is now available for the piecewise exponential model.

The EFFECT statement is production.

The EFFECT statement is now production.

Graphs are now produced with ODS Graphics.

The new QINTERACT option in the TEST statement enables you to test whether any difference exists among the coefficients across quantiles if several quantiles are specified in the MODEL statement.

The RANKSCORE option in the TEST statement now supports the tau score function, which is appropriate for non-iid error models.

The EFFECT statement is now production.

The new MCDINFO suboption of the LEVERAGE option in the MODEL statement displays detailed information about the MCD covariance estimate, including the low-dimensional structure, the breakdown value, the MCD center, and the MCD covariance.

The EFFECT statement is now production.

You can now produce Rao-Scott chi-square tests with second-order corrections.

Replication variance estimation is now available for domain analysis.

The EFFECT statement is now production.

Variance estimation based on replication methods is now available for quantiles.

The SURVEYPHREG procedure is now production. Also, the addition of programming statements enables you to include time-dependent covariates in the model.

The SURVEYREG procedure now provides replication variance estimation for domain analysis.

The EFFECT statement is now production.

Instead of specifying the total sample size to allocate among the strata, you can specify the desired margin of error for estimating the overall mean from the stratified sample.

The VARCLUS procedure now produces a dendrogram by default when ODS Graphics is enabled. The MAXPOINTS= option enables you to suppress the dendrogram when there is a large number of clusters.

Box plots, which are created with the MEANS statement or for one-way ANOVA when ODS Graphics is enabled, are not produced when the number of outlier points exceeds the limit, which is controlled by the MAXPOINTS= option.

The CLUSTER procedure now produces a dendrogram by default when ODS Graphics is enabled.

Frequency plots and cumulative frequency plots are no longer produced by default when ODS Graphics is enabled. You can request these plots with the PLOTS=FREQPLOT and PLOTS=CUMFREQPLOT options in the TABLES statement.

The fit plot, box plot, interaction plot, ANCOVA plot, and contour fit plot are not produced when the number of points exceeds the limit, which is contolled by the MAXPOINTS= option. This limit also applies to diagnostic plots and residual plots.

Plots associated with the INFLUENCE or IPLOTS= options in the MODEL statement are not produced when the number of points exceeds the limit, which is controlled by the MAXPOINTS= option.

If the ODDSRATIO statement or CLODDS= option is specified, the default "Odds Ratio" table is no longer produced, and only the requested results are displayed.

PROC MCMC no longer produces the tuning, burn-in, and sampling history tables by default. To produce this information, specify the MCHISTORY= option in the PROC MCMC statement.

The scaled inverse chi-square distribution is parameterized in terms of scale², as opposed to scale in the previous release.

Plots associated with the INFLUENCE, RESIDUAL, and VCIRY options are not produced when the number of points exceeds the limit, which is controlled by the MAXPOINTS= option.

The fit plot is not produced when the number of points exceeds the limit, which is controlled by the MAXPOINTS= option.

The rank score test has changed.

The weighted frequency plot is no longer produced by default when ODS Graphics is enabled. You can request this display with the PLOTS=WTFREQPLOT option in the TABLES statement.

The VARCLUS procedure now produces a dendrogram by default when ODS Graphics is enabled.