Features
Analysis of Variance
-
ANOVA for balanced data
-
general linear models
-
unbalanced data
-
analysis of covariance, response-surface models, weighted regression, polynomial regression, MANOVA, repeated measurements analysis
-
least squares means
-
random effects
-
estimate linear functions of the parameters
-
test linear functions of the parameters
-
multiple comparison of means
-
homogeneity of variance testing
-
-
mixed linear models
-
fixed and random effects
-
REML, maximum likelihood, and MIVQUE0 estimation methods
-
least-squares means and differences
-
sampling-based Bayesian analysis
-
many covariance structures, some of which are compound symmetry, unstructured, AR(1), Toeplitz, heterogeneous AR(1), and Huynh-Feldt
-
multiple comparison of least-squares means
-
repeated measurements analysis
-
-
nonlinear mixed models
-
variance components
-
nested models
-
lattice designs
Regression Analysis
-
ridge regression
-
linear regression
-
nine model-selection techniques including backwards, forwards, stepwise, and those based on R-squared
-
diagnostics
-
hypothesis tests
-
partial regression leverage plots
-
outputs predicted values and residuals
-
graphics device plots
-
-
response surface regression
-
nonlinear regression
-
derivative-free
-
steepest-descent, Newton, modified Gauss-Newton, Marquardt and DUD methods
-
-
linear models with optimal nonlinear transformation
-
partial least squares
Categorical Data Analysis
-
weighted least squares analysis
-
loglinear models
-
generalized estimating equations
-
Mantel-Haenszel Methods
-
Fisher's exact test
-
r x c exact tests
-
probit analysis
-
logistic regression
-
various model-selection methods
-
proportional odds model for ordinal response
-
regression diagnostics
-
conditional logistic model
-
roc curves
-
-
discrete choice models
-
multinomial logit models
-
generalized linear model
-
probability distributions include normal, binomial, Poisson, negative binomial, gamma, and inverse Gaussian
-
link functions include logit, probit, identity, complementary log-log, log, and power with lamda=value
-
profile likelihood-based confidence intervals
-
likelihood ratio statistics for contrasts
-
user-defined link functions and probability distributions
-
Multivariate Analysis
-
principal components
-
canonical correlation
-
discriminant analysis
-
structural equation modeling and path analysis
-
general COSAN model
-
optimization methods include Levenberg-Marquart algorithm, ridge-stabilized Newton-Raphson, quasi-Newton, and conjugate gradient algorithms
-
estimation methods include maximum likelihood, least squares, generalized least squares, weighted least squares, and diagonally weighted least squares
-
equality and inequality constraints
-
-
multiplicity-adjusted p-values
-
multivariate one-way ANOVA model, discrete or continuous variables
-
linear contrasts to compare proportions/means
-
adjustments include bootstrap and permutation resampling
-
-
multidimensional scaling models
-
simple Euclidean and weighted Euclidean models
-
ordinal, interval, ratio, or absolute levels of measurement
-
fits distances, squared distances, log distances, or distances raised to any power
-
Survival Analysis
-
parametric models for failure-time data
-
left, right, interval-censored
-
model random disturbance with extreme value, normal, logistic, exponential, Weibull, log-normal, log-logistic, gamma distributions
-
-
nonparametric methods
-
Kaplan-Meier and lifetable estimates
-
tests for homogeneity of survival distributions
-
rank tests for association of response with covariates
-
-
Cox regression (semi-parametric proportional hazards model)
-
time-dependent covariates
-
stratified analyses
-
counting process formulation
-
various model-selection methods
-
four methods of handling ties
-
Psychometric Analysis
-
factor analysis
-
simple and multiple correspondence analysis
-
multidimensional scaling
-
conjoint analysis
-
principal components
-
multidimensional preference analyses
-
preference mapping
Cluster Analysis
-
hierarchically cluster data
-
some methods include average linkage, centroid method, complete linkage, density linkage, flexible-beta method, median method, single linkage, and Ward's minimum-variance method
-
-
disjoint clustering of very large data sets
-
approximate covariance estimation for clustering
-
disjoint or hierarchical clustering based on correlation or covariance matrix
-
clustering based on nonparametric density estimates
-
numeric coordinates or distance data
-
approximate significance tests for number of clusters
-
hierarchical joins of nonsignificant clusters
-
Nonparametric Analysis
-
simple linear rank statistics based on Wilcoxon, median, Savage, and Van der Waerden scores
-
exact p-values for simple linear rank statistics
-
tests for scale differences include Siegel-Tukey, Ansari-Bradley, Klotz, and Mood
-
Kolmogorov-Smirnov statistic
-
kernel density estimation
-
loess regression
-
thin-plate smoothing splines
Statistics and Operations Research Home Page | SAS/STAT Software