This preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: STAT 512: Applied Regression Analysis Topic 6 Spring 2008 Topic Overview: • Oneway Analysis of Variance (ANOVA) Chapter 16: OneWay ANOVA • Also called single factor ANOVA . • The response variable Y is continuous (same as in regression). • There are two key di erences regarding the explanatory variable X . It is a qualitative variable (e.g. gender, location, etc). Instead of calling it an explanatory variable, we now refer to it as a factor. No assumption (i.e. linear relationship) is made about the nature of the relation ship between X and Y. Rather we attempt to determine whether the response di er signi cantly at di erent levels of X. This is a generalization of the two independentsample ttest . • We will have several di erent ways of parameterizing the model: the cell means model the factor e ects model: two di erent possible constraint systems for the factor e ects model 1 Notation for OneWay ANOVA X (or A ) is the qualitative factor • r (or a ) is the number of levels • we often refer to these as groups or treatments Y is the continuous response variable • Y ij is the jth observation in the ith group • i = 1 , 2 , . . . , r levels of the factor X • j = 1 , 2 , . . . , n i observations at factor level i . 2 KNNL Example (p 685) • See the le nknw677.sas for the SAS code • Y is the number of cases of cereal sold (CASES) • X is the design of the cereal package (PKGDES) • There are 4 levels for X representing 4 di erent package designs: i = 1 to 4 levels • Cereal is sold in 19 stores, one design per store. (There were originally 20 stores but one had a re.) • j = 1 , 2 , . . . , n i stores using design i . Here n i = 5 , 5 , 4 , 5 . We simply use n if all of the n i are the same. The total number of observations is n T = ∑ r i =1 n i = 19 . /* File: nknw667.sas */ *NKNW677.sas, one way anova using data in Table 16.1; data cereal; infile 'ch16ta01.dat'; input cases pkgdes store; proc print data=cereal; run; Note that the store variable is just j ; here it does not label a particular store, and we do not use it (only one design per store). Model (Cell Means Model) Model Assumptions • Response variable is normally distributed • Mean may depend on the level of the factor • Variance is constant • All observations are independent 3 Cell Means Model Y ij = μ i + ij • μ i is the theoretical mean of all observations at level i • ij iid ∼ N (0 , σ 2 ) and hence Y ij iid ∼ N ( μ i , σ 2 ) • Note there is no intercept term and we just have a potentially di erent mean for each level of X . In this model, the mean does not depend numerically on the actual value of X (unlike the linear regression model)....
View
Full Document
 Fall '10
 Yen
 Statistics, Regression Analysis, Variance, PROC GLM

Click to edit the document details