142 CHAPTER 8 Contingency Tables
Table 8.15
Relationship Between Reported Health
Status and Smoker vs. Non-smoker
Categories
Reported
health status
Current
smoker
Quit
smoking
Never
smoked
Total
Excellent 20 42 40 102
Good 250 300 285 835
Poor 117 72 70 259
Very poor 29 18 12 59
Total 416 432 407 1255
Table 8.14
Relationship Between Reported Health
Status and Smoking Usage
Reported
health status
Smoked 250 or
more cigarettes
Smoked 0 to
249 cigarettes
Total
Excellent 30 72 102
Good 350 485 835
Poor 121 138 259
Very poor 39 20 59
Total 540 715 1255
Why is the chi - square test appropriate here?
In the same survey, the subjects also were asked to classify themselves
according race, with the choices African American, Asian, Hispanic, Native
Americans, or European American. Twelve subjects failed to respond.
Table 8.16 shows race versus smoking status.
Is race related to smoking status? Test at 5%. Should you do the test
one - sided or two - sided? Is the chi - square an appropriate test? Does it matter
that 12 subjects did not respond? If a much higher percentage of the sub-
jects did not respond what might invalidate the analysis?
Again using the same survey data, suppose we have statistics at baseline
regarding the subjects drinking status, as well as their smoking status.
Given the results in the following table, what would you conclude about
the relationship between smoking status and drinking status? Test at the 5%
level (Table 8.17 ).