In: Statistics and Probability
In reference to the year 2000 national census, the distribution of United States households according to type of living arrangement is given in the table below. An SRS of 500 households gives the observed frequencies, by household type, for the city of Aurora, Illinois.
Ethnic Origin Census Percent Observed frequency from sample
Married, with children 25% 95
Married, with no children 20% 110
Single parent 5% 35
Single individual 15% 65
One person 20% 120
Other (roommate, etc.) 15% 75
Using a significance level of 0.05, test the claim that the Aurora distribution fits the distribution obtained by the census personnel.
null hypothesis:Ho: Aurora distribution fits the distribution obtained by the census personnel.
alternate hypothesis:Ha: Aurora distribution is different thn the distribution obtained by the census personnel.
degree of freedom =categories-1=6-1=5
for 5 df and 0.05 level rejection region >11.070
applying chi square goodness of fit test:
observed | Expected | Chi square | |||
category | Probability(p) | Oi | Ei=total*p | R2i=(Oi-Ei)2/Ei | |
married w children | 0.250 | 95.000 | 125.00 | 7.200 | |
married w no children | 0.200 | 110.000 | 100.00 | 1.000 | |
single parent | 0.050 | 35.000 | 25.00 | 4.000 | |
single individual | 0.150 | 65.000 | 75.00 | 1.33 | |
one person | 0.200 | 120.000 | 100.00 | 4.00 | |
other | 0.150 | 75.000 | 75.00 | 0.00 | |
total | 1.000 | 500 | 500 | 17.53 |
as test statistic 17.53 is higher than critical value ; we reject null hypothesis
we have sufficient evidence to conclude that Aurora distribution is different from the distribution obtained by the census personnel.