CompareMetricBySplit

compares one metric variable by a split variable with t.test-oneway.test and wilcox.test-kruskal.test. For two groups confidence intervals for mean difference, difference in pseudomedians (Hodge-Lehmann), and proversion probabilities are provided.

CompareMetricBySplit(DFR, METRIC, SPLIT, DIGITS = 3, CONF.LEVEL = 0.95)

Arguments

DFR,	is a dataframe
METRIC,	name of metric variable in DFR
SPLIT,	name of split factor (grouping variable) in DFR
DIGITS	= 3, Number of digits for estimates
CONF.LEVEL	= 0.95, Confidence Level for confidence interval

Value

a data.frame with results

Details

CompareMetricBySplit

Examples

x<-c(rnorm(100,2,1),rnorm(200,2.4,1))
x[sample(1:300,13)]<-NA
x<-c(rnorm(100,2,1),rnorm(200,2.5,1))
gr2<-factor(c(rep("Gr1",100),rep("Gr2",200)))
gr3<-factor(c(rep("Gr1",100),rep("Gr2",100),rep("Gr3",100)))
dfr<-data.frame(X=x,GR2=gr2,GR3=gr3)# 
CompareMetricBySplit(dfr,METRIC="X",SPLIT="GR2")
#>                       X: (Gr1-Gr2)   Diff lower95 upper95 p_test         Method
#> 1                  Mean Difference -0.708  -0.931  -0.485 <1e-04         t.test
#> 2 HodgeLehmann-Diff. PseudoMedians -0.718  -0.957  -0.492 <1e-04    wilcox.test
#> 3        ProversionProbability [%] 29.255  23.554  35.853        ProversionProb
CompareMetricBySplit(dfr,METRIC="X",SPLIT="GR3")
#>                      p.value
#> oneway.test  X ~ GR3  <1e-05
#> kruskal.test X ~ GR3  <1e-05