Chapter 11--Regression and Correlation Methods

Let z0 be the fishers transform of 0 the test

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

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: e Fisher’s transform by zxy = 1 log 2 1 + rxy 1 − rxy . Let z0 be the Fisher’s transform of ρ0 . The test statistics is √ H0 λ = (zxy − z0 ) n − 3 ∼ N (0, 1) We reject H0 if |λ| > z1−α/2 and p − value = 2 × P (Z > λcomputed ). This test is approximate. For the birth weight data test H0 : ρ = 0.85 vs Ha : ρ = 0.85 Chapter 11: Regression and Correlation Methods Stat 491: Biostatistics Introduction Least Square Estimates of the Parameters Inference about the Parameters Prediction Assessing Adequacy of Fit Correlation Multiple Regression R-codes for testing H0 : ρ = ρ0 when ρ0 = 0 r<-cor(bw,estriol) n<-length(bw) alpha<-.05 rho_0=0.7 z<-(1/2)*log((1+r)/(1-r)) z_rho_0 <-(1/2)*log((1+rho_0)/(1-rho_0)) lambda=(z-z_rho_0)*sqrt(n-3) p_value <- 2*(1-pnorm(abs(lambda))) Chapter 11: Regression and Correlation Methods Stat 491: Biostatistics Introduction Least Square Estimates of the Parameters Inference about the Parameters Prediction Assessing Adequacy of Fit Correlation Multiple Regression A 100(1 − α)% Confidence Interval for ρxy Based on Fisher’s transform, construct confidence interval for zρ as 1 = (a, b ), say. zxy ± z1−α/2 √ n−3 Then re-transforming the end points e 2a − 1 e 2b − 1 , e 2a + 1 e 2b + 1 yields an approximate 100(1 − α)% confidence interval for ρxy . Chapter 11: Regression and Correlation Methods Stat 491: Biostatistics Introduction Least Square Estimates of the Parameters Inference about the Parameters Prediction Assessing Adequacy of Fit Correlation Multiple Regression A 100(1 − α)% Confidence Interval for ρxy : R-code and Example For the birth weight-estriol example, construct 95% Confidence interval for ρ. R-codes r<-cor(bw,estriol) n<-length(bw) alpha<-.05 z<-(1/2)*log((1+r)/(1-r)) conf_z<- z + c(-1,1)*qnorm(1-alpha/2)/sqrt(n-3) conf_rho<- (exp(2*conf_z)-1)/(exp(2*conf_z)+1) Chapter 11: Regression and Correlation Methods Stat 491: Biostatistics Introduction Least Square Estimates of the Parameters Inference about the Parameters Prediction Assessing Adequacy of Fit Correlation Multiple Regression Introduction In...
View Full Document

This note was uploaded on 02/03/2014 for the course STAT 491 taught by Professor Solomonharrar during the Fall '12 term at Montana.

Ask a homework question - tutors are online