BiostatisticsAnalysis.ppt

Chapter 8 Analysis of Variance （ANOVA 方差分析 Analysis of Variance ANOVA is a statistical test used to determine if more than two population means are equal. The test uses the F-distribution. ANOVA is also called F test . It was named to honor Sir Ronald Fisher, one of the founders of modern-day statistics Computing ANOVA F statistic data d; do id 1 to 5; do a 1 to 3; input y @@; output; end; end; cards; 3.1 7.6 6.8 2.9 6.4 6.8 3.3 7.5 7.4 3.6 6.9 7.0 3.5 6.3 6.5 run; proc sort data d; by a; run; proc univariate data d normal; var y; by a; run; proc anova data d; class a; model y a; means a / snk hovtest; run; SAS program for normality homogeneity test Univariate Procedure Variable Y A 1: W:Normal 0.962048 Pr W 0.8212 A 2: W:Normal 0.87849 Pr W 0.3000 A 3: W:Normal 0.94962 Pr F A 2 0.1533 0.0767 5.0515 0.0256 Error 12 0.1821 0.0152 Variances: A1 0.082 A2 0.363 A3 0.110 data d; do id 1 to 5; do a 1 to 3; input y @@; output; end; end; cards; 3.1 7.6 6.8 2.9 6.4 6.8 3.3 7.5 7.4 3.6 6.9 7.0 3.5 6.3 6.5 run; data c; set d; z log y ; run; proc sort; by a; run; proc univariate normal; var z; by a; run; proc anova; class a ; model z a ; means a/snk hovtest; run; SAS program for data transformation: Univariate Procedure Variable z A 1: W:Normal 0.958599 Pr W 0.7999 A 2: W:Normal 0.879645 Pr W 0.3052 A 3: W:Normal 0.955174 Pr F A 2 0.000062 0.000031 1.8674 0.1967 Error 12 0.0002 0.000017 Variances: A1 0.0078 A2 0.0075 A3 0.0023 Two way ANOVA Two-way ANOVA 双方向方差分析） Two-factor ANOVA 双因素方差分析） Without interaction With interaction §5 Method of two-way ANOVA without interaction # of repeated measures r 1 1） The data ： Data: response y, factors A,B level a,b, r 1 , n ab 2） The principle: SS: SST SSA + SSB + SSE df: n -1 a -1 + b-1 + a-1 b-1 Partitioning the sum of square of deviation: Is big enough, then A effect y Is big enough, then B effect y 3 The