Selecting an Inference Procedure

Question 1

How do you tell if a problem is an interval (panic) or a significance test (phantom)?

Answer 1

Intervals: Look for the words “estimate” or “interval”. Should your answer be a number?
Tests: Look for a yes or no question: “Do we have…?” or “Is/are there…?”. key word: “evidence”

Question 2

How do you tell if it’s a mean or proportion problem?

Answer 2

Mean: key word “average” or something that could be averaged. If you are given an x-bar or a standard deviation.
Proportion: percent given or asked for. The problem just lists counts…how many said yes?

Question 3

What 3 (and a half) things do I need to look for?

Answer 3

Is it a mean or proportion (t or z)? Is it a test or interval (phantom or panic)? Is it 1 or 2 sample (or possibly paired)?

Question 4

Do I use z or t?

Answer 4

zap tax. Figure out if it’s means or proportions. Then, proportions are always z. Means could be z but in AP stats we always use t instead.

Question 5

How do you tell if it’s 1 or 2 samples?

Answer 5

Generally, if they only give you one number or set of numbers, it’s one sample. Beware of paired data.

Question 6

How do you tell if it’s paired?

Answer 6

There’s just 1 sample that gets two different treatments or would have a reason to be specifically paired (like a husband/wife or parent/child), then it’s paired. If there is no logical reason to pair the data and subtract individuals first, it’s 2 sample.