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: Advantages of 2 χ statistic over z statistic : A statistic can only be used to compare a estimate to its hypotheses value i.e for hypotheses like : H : or H : This is only possible in tables as we saw above. For contingency tables with more than rows and/or columns, we need more than hypotheses to describe the association. Eg : Suppose we want to compare the effectiveness of Advil, Tylenol and Excedrin in reducing headaches. Our contingency table would then look like : Medicine Headache ? Y N Advil p 1 1 p 1 Tylenol p 2 1 p 2 Excedrin p 3 1 p 3 Here is the population proportion of patients taking Advil whose headache was reduced. Similarly for p 2 and p 3. Here the explanatory variable ( ) has categories corresponding to the three medicines used. So, the null hypotheses for independence will be H : We can rewrite this hypotheses as H 01 : and H 02 : Now, in order to test H 01 and H 02 we have to use z statistics (one for each). On the other hand, only 2 χ statistic (with 2 df) will be enough to test this . So, crudely speaking, statistics are easier to use when we have large contingency tables. Limitations of the Chisquare test : We have seen that a p – value corresponding to a...
View
Full
Document
This note was uploaded on 04/18/2008 for the course EXP 3116 taught by Professor Fasig during the Spring '08 term at University of Florida.
 Spring '08
 Fasig

Click to edit the document details