A high school librarian wants to estimate the proportion of all students at the school who read a book for pleasure during the past month. She selects a random sample of 150 students, and 54 of them report that they read a book for pleasure during the past month. Which of the following is the most appropriate inference procedure for her goal?
- AA two-sample z-interval for p1 - p2, comparing the readers with the nonreaders in the sample
- BA one-sample t-interval for the mean number of books read by students at the school
- CA one-sample z-interval for p, the proportion of all students who read a book last monthCorrect
- DA one-sample z-test of H0: p = 0.36, where p is the proportion of all the students who read
Why C is right
The librarian wants to estimate one population parameter: the proportion p of all students at the school who read a book for pleasure last month. The data are categorical (read / did not read) from one random sample, so the right procedure is a one-sample z-interval for a population proportion. The sample proportion is p-hat = 54/150 = 0.36. The large-counts condition holds because n*p-hat = 54 and n*(1 - p-hat) = 96 are both at least 10.
Why the other 3 choices miss
- A
- Wrongly treats the two answer categories of one sample as two separate populations; readers and nonreaders together make up a single sample, and the comparison is meaningless because p1 - p2 has no population parameter to estimate here.
- B
- Mixes up categorical and quantitative data: each student gave a yes/no answer, not a count of books, so there is no mean to estimate and a t-procedure does not apply.
- D
- Confuses estimating with testing: the goal is to estimate p, which calls for a confidence interval, and using the sample's own value 0.36 as the null hypothesis is a further error, because hypotheses are about parameters and are set before the data are seen.