130 likes | 311 Views
Who Has the Power? Proc Power. Kelly Guyton SAS Final Presentation. Introduction. Determining power, type I error rate, and sample size are imperative to designing a successful epidemiologic study or clinical trial.
E N D
Who Has the Power?Proc Power Kelly Guyton SAS Final Presentation
Introduction Determining power, type I error rate, and sample size are imperative to designing a successful epidemiologic study or clinical trial. Calculations are often complex and scientists must purchase expensive software.
Proc Power • Were added to Analyst window for SAS V7.0. • SAS V9.1 introduced Proc power and Proc GLMpower as an experimental component. • Proc Power is full featured, easy to use, and tables can be added directly into other documents.
Statistical Tests Continuous methods: T-tests (1 and 2 sample, paired) ANOVA (repeated measure, and 2-way) Linear regression (simple, multiple) Wilcoxon (sign-rank, Man-Whitney) Equivalence Tests Confidence intervals
Statistical Tests Categorical Methods: Binomial Chi-square Likelyhood ratio Fisher’s exact Mc Nemar’s Chochran Survival Tests: Log Rank Gehan Rank test
SAS Code-MeansSample Size ProcPower; Twosamplemeans meandiff=50 to 100 by 10 stddev=15 to 30 by 5 Groupweights=(1,3) power=0.8 ntotal=.; Plot y=power min=0.5 max=0.99; run;
SAS Code-FrequencySample Size ProcPower; Twosamplefreq test=pchi proportiondiff=0.10 to 0.80 by 0.1 refproportion=0.26 power=0.9 alpha=0.05 to 0.25 by 0.05 ntotal=.; run;
SAS Code-FrequencyPower calculation ProcPower; Twosamplefreq test=fisher proportiondiff=0.10 to 0.80 by 0.1 refproportion=0.26 npergroup=250 power=.; run;
Who Knew? SAS Genetics Proc Allele Proc Casecontrol Proc Family