Every procedure in the course has been taught. The remaining skill is choosing among them.
Step 1: Let's Learn
Read it, or press Listen and follow the words.
First question: what kind of data
Counts or proportions of a category point to proportion procedures or chi-square. Measurements point to means.
Second question: how many groups
One group, two groups, or more than two. More than two categorical groups usually means chi-square.
Third question: estimate or decide
A question asking "how much" wants a confidence interval. One asking "is there evidence" wants a significance test.
Watch for pairing
Two columns of measurements on the same subjects is a paired design, analysed as one sample of differences.
Conditions still apply
Choosing correctly is only the first mark. The conditions for that procedure must still be checked and stated.
Three questions identify the procedure
What type of variable — categorical or quantitative? How many samples or groups? Estimating or testing? Answering those three narrows the choice to one procedure almost every time.
Where the choice goes wrong
Treating paired data as two independent samples, using a z procedure when sigma is unknown, and applying a proportion procedure to quantitative data. Each is a specific and frequent misidentification.
Conditions are part of the choice
A procedure whose conditions fail is the wrong procedure, however well it matches the question. Checking conditions is therefore not a formality appended afterwards but part of selecting the method.
Name the procedure explicitly
"A two-sample t test for a difference in means" earns credit that unnamed working does not. The exam wants the identification stated, not merely implied by the formula used.
Step 2: Try It Yourself
Tap and try it out.
Category A
Category B
Category A has the most. It has 6 more than Category B.
Step 3: Watch an Example
One step at a time.
Watch Rosa Choose a Procedure
Rosa is asked whether the mean commute differs between two independent towns.
- Step 1
Commute time is a measurement, so this is a mean procedure rather than a proportion one.
Step 4: Your Turn
Practice makes it stick.
The Choice
Problem 1 of 2
Comparing mean commute in two unrelated towns. 1 two-sample t-test, 2 chi-square.
The Estimate
Problem 2 of 2
A question asks how much support a candidate has. 1 confidence interval, 2 significance test.
Which Procedure
1 of 8
Before and after measurements on the same subjects. 1 paired t, 2 two-sample t.
2 of 8
Testing whether two categorical variables are related. 1 chi-square, 2 t-test.
3 of 8
Estimating one population proportion. 1 one-proportion interval, 2 chi-square.
4 of 8
"Is there evidence" asks for which? 1 interval, 2 test.
5 of 8
A 3 by 3 table of counts. Degrees of freedom for the chi-square test?
6 of 8
Is choosing the procedure enough, without checking conditions? 1 yes, 0 no.
7 of 8
Sort each situation by the family of procedure it needs.
Tap something to move it.
- Empty
- Empty
8 of 8
Comparing the distribution of one variable across three schools. 1 chi-square, 2 paired t.
Step 5: Quick Check
Show what you know.
Question 1 of 2
Before and after readings on the same people. Which procedure? 1 paired t, 2 two-sample t.
Question 2 of 2
What is the first question to ask when choosing a procedure?
What You Learned
- Ask what kind of data, how many groups, and whether to estimate or decide.
- Watch for pairing, which turns two columns into one sample of differences.
- Choosing correctly is only the first step; conditions must still be checked.