In: Statistics and Probability
Smokers are commonly thought of as nervous people whose
emotionality is at least partly caused by the stimulating effect
tobacco has on the nervous system. Nesbitt (1972) conducted a study
with 300 college students and concluded that smokers are less
emotional than nonsmokers, that smokers are better able to tolerate
the physiological effects of
anxiety, and that, over time, smokers become less emotional than
nonsmokers.
Subjects of both genders were drawn from three different colleges
and classified as smokers if they smoked any number of cigarettes
on a regular basis. In one aspect of the experiment, all subjects
were given the Activity Preference Questionnaire (APQ), a test
designed to measure the emotionality of the subjects. The APQ is
scored using an ordinal scale of 0–33, with lower scores indicating
less emotionality, that is, greater sociopathy. The mean overall
scores were 18.0 for smokers and 20.3 for nonsmokers.
Suppose this experiment is repeated using a group of 8 randomly chosen smokers and 10 randomly chosen nonsmokers. Do these data support the same conclusion concerning emotionality as Dr. Nesbitt’s data?
Smokers: 16, 18, 21, 14, 25, 24, 27, 12
Nonsmokers: 17, 15, 28, 31, 30, 26, 27, 20, 21, 19
*Nonparametric Statistical Inference, 5th edition, J. D. Gibbons and S. Chakraborti. (Chapter 8: Problem 8.10)
>
smoker=c(16,18,21,14,25,24,27,12)
> nonsmokers=c(17,15,28,31,30,26,27,20,21,19)
> wilcox.test(smoker,nonsmokers, paired=FALSE)
Wilcoxon rank sum test with continuity correction
data: smoker and nonsmokers
W = 24, p-value = 0.168
alternative hypothesis: true location shift is not equal to 0
Interpetation: Here pvalue is greater than 0.05 level of significance, we may fail to reject null hypothesis and conclude that smokers and non-smokers have same emotionality. so these data does not supported to Dr. Nesbitt’s data.
Assuming it follows assunptions of t-test,
> t.test(s,ns,var.equal=T)
Two Sample t-test
data: s and ns
t = -1.4233, df = 16, p-value = 0.1739
alternative hypothesis: true difference in means is not equal to
0
95 percent confidence interval:
-9.397793 1.847793
sample estimates:
mean of x mean of y
19.625 23.400
Interpetation: Here pvalue is greater than 0.05 level of significance we may fail to reject null hypothesis and conclude that smokers and non-smokers have same emotionality. so these data does not supported to Dr. Nesbitt’s data.