In a study, researchers designed a program to prevent college students from smoking. The program ran for a year at a college on randomly selected 200

students. Before the program was administrated, there were 62 students smoked almost everyday and 138 students did not smoke. After a year, the same group of students was surveyed again. There were 32 students who smoked a year ago decided not to smoke any more. However, 18 students who did not smoke a year ago start smoking now. The survey result is recorded in the following table.

	Before the program
After the program	Do not smoke	Smoke
Do not smoke	120	32
Smoke	18	30

Which of the following statement(s) is(are) correct?

		The Mantel-Haenszel Method should be used for this data.
		McNemar s Test should be used for this data.
		At 5% level of significance, the evidence is statistically significant that there is reduction in the rate of smoking among students after the program.
		At 5% level of significance, the evidence is statistically insignificant that there is reduction in the rate of smoking among students after the program.

Sometimes the results from a restricted sample may not be the same as the results from a larger population. For example, the results from the analysis on patients in a hosptial may be significantly different from the results from a sample taken from the general population. Therefore, the results from analyzing a sample from the patients in a hosptial may be signficantly different from the results from a sample taken from the general population. This phenomenom is called

Simpson's Paradox

Berkson's Fallacy

Pearson's Correlation

Sampling Error