In: Statistics and Probability
No Smoking: The general social survey conducted a poll of 685 adults in which the subjects were asked whether they agree that the government should prohibit smoking in public places. In addition, each was asked how many people lived in his or her household. The results are summarized in the following contingency table.
Household Size
1 2 3 4 5
Agree 66 109 37 28 30
no opinion 26 56 29 19 30
Disagree 51 79 46 48 31
(a) compute the expected frequencies.
(b) compute the value of the test statistics.
(c) how many degrees of freedom are there?
(d) Test the hypothesis of independence. Use the a=0.05 level of significance. What do you conclude?
Observed Frequencies | ||||||
S1 | S2 | S3 | S4 | S5 | Total | |
Agree | 66 | 109 | 37 | 28 | 30 | 270 |
No opinion | 26 | 56 | 29 | 19 | 30 | 160 |
Disagree | 51 | 79 | 46 | 48 | 31 | 255 |
Total | 143 | 244 | 112 | 95 | 91 | 685 |
Expected Frequencies | ||||||
S1 | S2 | S3 | S4 | S5 | Total | |
Agree |
143 * 270 / 685 = 56.365 |
244 * 270 / 685 = 96.1752 |
112 * 270 / 685 = 44.146 |
95 * 270 / 685 = 37.4453 |
91 * 270 / 685 = 35.8686 |
270 |
No opinion |
143 * 160 / 685 = 33.4015 |
244 * 160 / 685 = 56.9927 |
112 * 160 / 685 = 26.1606 |
95 * 160 / 685 = 22.1898 |
91 * 160 / 685 = 21.2555 |
160 |
Disagree |
143 * 255 / 685 = 53.2336 |
244 * 255 / 685 = 90.8321 |
112 * 255 / 685 = 41.6934 |
95 * 255 / 685 = 35.365 |
91 * 255 / 685 = 33.8759 |
255 |
Total | 143 | 244 | 112 | 95 | 91 | 685 |
(fo-fe)²/fe | ||||||
Agree |
(66 - 56.365)²/56.365 = 1.647 |
(109 - 96.1752)²/96.1752 = 1.7102 |
(37 - 44.146)²/44.146 = 1.1567 |
(28 - 37.4453)²/37.4453 = 2.3825 |
(30 - 35.8686)²/35.8686 = 0.9602 |
|
No opinion |
(26 - 33.4015)²/33.4015 = 1.6401 |
(56 - 56.9927)²/56.9927 = 0.0173 |
(29 - 26.1606)²/26.1606 = 0.3082 |
(19 - 22.1898)²/22.1898 = 0.4585 |
(30 - 21.2555)²/21.2555 = 3.5975 |
|
Disagree |
(51 - 53.2336)²/53.2336 = 0.0937 |
(79 - 90.8321)²/90.8321 = 1.5413 |
(46 - 41.6934)²/41.6934 = 0.4448 |
(48 - 35.365)²/35.365 = 4.5142 |
(31 - 33.8759)²/33.8759 = 0.2442 |
b) Test statistic:
χ² = ∑ ((fo-fe)²/fe) = 20.7164
c) df = (r-1)(c-1) = 8
d) Critical value:
χ²α = CHISQ.INV.RT(0.05, 8) = 15.5073
p-value:
p-value = CHISQ.DIST.RT(20.7164, 8) = 0.0079
Decision:
p-value < α, Reject the null hypothesis.
There is enough evidence to conclude that there is an association between factors at 0.05 significance level.