CISIZE computes exact target probabilities of interest (including power) for any scalar parameter in a General Linear Multivariate Model with Gaussian errors and fixed predictors. This software, consisting of a single SAS IML ® module, focuses on sample size computations for studies aimed at estimating a confidence interval. The theory and methods are detailed in article 74.
Reliability Confidence Intervals Version 1.0
This software, consisting of a set of SAS IML ® modules, computes two estimates of reliability: intraclass correlation and Cronbach’s alpha. Under the assumption of Gaussian, homogenous data with *any* covariance structure, the software gives excellent approximate confidence intervals. The exact results for the special case of compound symmetry are also available. The theory and methods are detailed in article 76.
Please click on the above link to the Journal of Statistical Science to find the most current version of the Powerlib software and related documentation. This software, consisting of a set of SAS IML ® modules, performs power analyses for the General Linear Multivariate Model (GLMM). It allows the creation of confidence intervals around power values that reflect the uncertainty due to using estimates of variances and/or means.
LINMOD (LINear MODels) performs a wide variety of General Linear Multivariate Model (GLMM) computations in SAS IML ® . This software is especially useful for repeated measures, complex designs, and teaching. See books 01 and 02.
MISSMOD generalizes LINMOD (above) by allowing missing data. For more information, see article 57.
Confluent Hypergeometric Function Version 1.0
This set of SAS ® modules computes the confluent hypergeometric function M(a,b,x) by five different methods. For details, see article 61.
Lookup Table of Type I Error Rates for Cluster Randomized Trials
The lookup table provides complete results from our simulation study of Type I error rates in cluster randomized designs. Type I error rates are displayed for several one-stage and two-stage analytic methods across 9090 experimental scenarios. Scientists may filter the results to find methods with good Type I error properties in experimental scenarios which best match their planned study design.
The Bias Correction Suite provides SAS macros to correct for bias in paired screening trials with Gaussian outcomes.
Expected Value of Missing Data Summary Statistics Module Version 1.0
The module calculates the expected value of N1 , N2 , and N9 from Table 1 of Catellier and Muller (2000).
Example Power Analysis for an Oral Cancer Biomarkers Study with Missing Data
The program approximates power for a hypothetical oral cancer biomarkers study with missing data and outputs a power curve.