bootstrap_hypo

bootstrap_hypo - x * , calculate the mean of the remaining...

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
Bootstrap Hypothesis Testing To illustrate the technique, consider the case of two independent samples: Observed Sample 1 of size n : { x obs, 1 ,x obs, 2 ,...,x obs,n } ⇒ x obs Observed Sample 2 of size m : { y obs, 1 ,y obs, 2 ,...,y obs,m } ⇒ y obs Observed sample mean difference: t * obs = x * obs - y * obs Hypotheses and Alpha Level H 0 : Both samples are from the same population H 1 : Both samples are NOT from the same population and μ x > μ y α = 0 . 05 Bootstrap Procedure Step 0 : Merge the two observed samples into one sample of ( n + m ) observations Step 1 : Draw a bootstrap sample of ( n + m ) observations with replacement from the merged sample. Step 2 : Calculate the mean of the first n observations and call it
Background image of page 1
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: x * , calculate the mean of the remaining m observations and call it y * , and nally, evaluate the test statistic: t * = x *-y * (1) Step 3 : Repeat Step 1 and Step 2 for B (e.g., 3000) times and obtain B values of the test statistic. Step 4 : The desired p-value is then estimated as p-value = number of times { t * > t * obs } B (2) Reject H if p-value < and retain H otherwise. Points to Note No assumption of normality As such, the sampling distribution (i.e., distribution of all possible t * s) does not gen-erally follow a t-distribution. 1...
View Full Document

This note was uploaded on 07/17/2008 for the course PSYCH 826&N taught by Professor Myung during the Fall '04 term at Ohio State.

Ask a homework question - tutors are online