Question

What statistical test(s) need to be done to determine if there is an association between the new drug treatment and the occurance of angina-free days? Please explain your reasoning for choosing this test(s) and show your work. Thanks!

A new antianginal drug is evaluated in its ability to reduce the occurance of chest pain in patients. A group of 267 patients suffering from angina were treated with placebo and the new drug in a double blind cross over study (l.e., each patient received both placebo and drug, with the order randomized and with the patient and observer blinded to treatment). The frequency of reduced symptoms were assessed. Is there an association between the new drug treatment and the occurance of angina-free days? Reduced Symptoms 67 Placebo No Reduced Symptoms 103 Drug Reduced Symptoms No Reduced Symptoms 32 65

Answer 1

A Drug Solution - from the gives information, we define two attributes A & B such that A: (with level B: Placebo (with a levels) Here 2x2 contingency table of counts observed frequencie) is gives. The appropnate test to check association beth drug & occurence of angina free days is Chisquore test for independente of two attributes. The problem is to test Hot The drug 4 occurrence of symptoms are independent (there is no association) Vs H: The doug & occurence of symptoms ore dependent (there is association) Test statistic- x² = N(ad-bc72 X² (a+b)(c+d)(at)6+). To find this value consider 3 A 267 BI B Total x = 267 (62)(6s– (lo3 (32)]^ a=67 b= 103 170 (170) (97) (99) (168) c=32 d=65 9.7 Xcan 1. og 18 Total gg 168 Critical value At Stol-ors. ² = 3.841 Decision : Xcal < x², we accept He at s y los Conclusion - The drug & occurence of symptoms are independest ie there is no association