1/27/2009

Inferences in Regression and Correlation Analysis (continued)

• Confidence band for the entire regression line
• Example: Airfreight,
• 
• 
• 
• Using SAS: See modification to airfreight.sas

• ANOVA approach to regression
• General Linear Test Approach
• Full model
• Reduced model
• Test stat:

• Coefficient of Determination, R²
• Normal Correlation Model and Bivariate Normal dist
• Correlation model
• Density function
• Marginals
• Conditionals
• Equivalence to normal error regression model
• Correlation:
• Pearson
• Formula
• Inferences about rho

1. Testing that rho=0

2. Interval estimation using the Fisher transformation

• Using SAS: Proc Corr
• Spearman
• Nonparametric correlation
• Procedure

1. rank within Y1 and Y2

2. Pearson correlation on ranks

• Inference
• Using SAS

Prepare the following problems for a Problem Session on Thursday:

2.4, 2.13, 2.23, 2.45(a), 2.51, 2.52
For #2.45, also answer this question: Does it make sense to use a correlation model here rather than a normal error regression model?