Sleuth3 (version 0.1-8)

case1803: Smoking and Lung Cancer

Description

In a retrospective case-control study, researchers identified 86 lung cancer patients and 86 controls (without lung cancer), and categorized them according to whether they were smokers or non-smokers. The goal is to see whether the odds of lung cancer are greater for smokers than for non-smokers.

Usage

case1803

source

Ramsey, F.L. and Schafer, D.W. (2013). The Statistical Sleuth: A Course in Methods of Data Analysis (3rd ed), Cenage Learning.

References

Anderson, T.W., Reid, D.B.W. and Beaton, G. H. (1972). Vitamin C and the Common Cold, Canadian Medial Association Journal 107: 503--508.

Examples

str(case1803)
attach(case1803)
   
## INFERENCE
myTable   <- cbind(Cancer,Control)   # Make a 2-by-2 table of counts 
row.names(myTable)  <- Smoking   # Assign the levels of Smoking as row names  
myTable   

fisher.test(myTable,  alternative="greater")  # Alternative: that odds of Cancer 
  # in first row are greater.
fisher.test(myTable) # 2-sided alternative to get CI for odds ratio
myGlm1  <- glm(myTable ~ Smoking, family=binomial) # logistic reg (Ch 21)
summary(myGlm1)
exp(myGlm1$coef[2]) # 5.37963 : Estimated odds ratio
exp(confint(myGlm1)[2,]) #  1.675169 24.009510:  Approximate confidence interval
# Interpretation: The odds of cancer ar 5.4 times as large for smokers as for 
# non-smokers (95% confidence interval: 1.7 to 24.0 times as large).

detach(case1803)