imi.t.test.conv: imi.t.test.conv

In vahidnassiri/imi: Implemenets Iterative Multiple Imputation

Description

This function uses iterative multiple imputation idea to compute the distance between p-values of t-tests from imputed datasets to see whether the already generated sets of imputed data are sufficient for such analysis or not.

Usage

imi.glm.conv(data.imp,epsilon,x, y = NULL, alternative='two.sided',
              mu,paired = FALSE, var.equal = FALSE, conf.level = 0.95,
              conv.plot=TRUE,successive.valid=3)

Arguments

data.imp	A list imputed sets of data as its components.
epsilon	The threshold for difference between two iterations.
x	Name of the data column which the t-test should be performed on it.
y	In case of two sample t-test this input variable specifiess the name of the second column for paired-test, or an indicator variable showing different populations for indepdendent two sample t-test.
alternative	A character string specifying the alternative hypothesis, it takes the values "two.sided" (default), "greater" or "less".
mu	A real number specifying the test value.
paired	A logical value specifying whether a paired t-test should be performed or not.
var.equal	A logical value specifying whether variance equality should be assumed or not.
conf.level	Confidence level of the test.
conv.plot	A logitical value, if TRUE then a convergence plot will be generated, if FALSE no plot will be provided.
successive.valid	An integer with minimum 1 which specifies the number of successive steps the stopping rule should be validated so the procedure could terminate.

Value

dis.steps	A vector with computed distance between iterations.
sufficient.M	An integer indicating the minumum number of sufficent imputed datasets, in case of insufficiency it will take the value 'Not sufficient!'

See Also

imi.t.test

Examples

# In th setting of the example of imi.t.test
imi.t.test.conv (chol.ttest$data.imp,epsilon=0.05/10,x=names(cholesterol)[3], 
+     y = NULL, alternative='two.sided',mu,paired = FALSE, var.equal = FALSE,
+                        conf.level = 0.95,conv.plot=TRUE,successive.valid=3)
$dis.steps
 [1] 0.3117110189 0.1578016324 0.0295781152 0.0487635084 0.0288667328 0.0318036141 0.0173372375 
 [8] 0.0090263689 0.0575828782
[10] 0.0007220647 0.0145497285 0.0417895357 0.0044211378 0.0124385165 0.0203312525 
[16] 0.0050927153 0.0058802488 0.0117815975
[19] 0.0022407713 0.0026592932 0.0007328110

$sufficient.M
[1] 22

>
> imi.t.test.conv (chol.ttest$data.imp,epsilon=0.05/100,x=names(cholesterol)[3], 
+        y = NULL, alternative='two.sided',mu,paired = FALSE, var.equal = FALSE,
+                  conf.level = 0.95,conv.plot=TRUE,successive.valid=3)
$dis.steps
 [1] 0.3117110189 0.1578016324 0.0295781152 0.0487635084 0.0288667328 0.0318036141 0.0173372375
 [8] 0.0090263689 0.0575828782
[10] 0.0007220647 0.0145497285 0.0417895357 0.0044211378 0.0124385165 0.0203312525 0.0050927153 
[17] 0.0058802488 0.0117815975
[19] 0.0022407713 0.0026592932 0.0007328110

$sufficient.M
[1] "Not sufficient!"

vahidnassiri/imi documentation built on June 25, 2019, 5:50 a.m.