This preview shows pages 1–5. Sign up to view the full content.
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: Twoway tables : Outline • Quick correcton for onesample tests of p • Twoway tables for testing the hypothesis of no association • The χ 2 distribution T. Parker () 1 / 16 Note: onesample test for p • Before, I said that the test statistic for testing H : p = p against some alternative was z = ˆ p p q ˆ p ( 1 ˆ p ) n . • That is not right. We should impose H on the test statistic (assume H is true): z = ˆ p p q p ( 1 p ) n . • There is no hypothesis imposed on a confidence interval (because it is an interval estimate) so the formula for the confidence interval is ˆ p ± z α/ 2 r ˆ p ( 1 ˆ p ) n T. Parker () 2 / 16 A summary table • Suppose the managers of a department store have data on whether cash or credit cards are used, and how much the purchase was worth. • They call a “small purchase” one that is for under $20. Population n credit? Small purchase 36 27 Large purchase 78 63 Total 114 90 T. Parker () 3 / 16 Step 1: rewrite the table • We rewrite the table so that it is a twoway table. < $20 ≥ $20 Total Credit 27 63 90 Cash 9 15 24 Total 36 78 114 • Added: a row for Cash ( = not credit) and a column of row totals. • It seems like people use credit cards more for large purchases....
View
Full
Document
This note was uploaded on 09/26/2011 for the course 06E 071 taught by Professor Stuff during the Spring '11 term at University of Iowa.
 Spring '11
 stuff

Click to edit the document details