Click to edit Master title style Yesterday – looking ahead… Two types of wrong actions.

Yesterday – looking ahead…

Two types of wrong actions.

Minitab

Type ‘Minitab’ in the program search box (click start button).

Descriptive Statistics Menu TreeStat

Basic StatisticsDisplay Descriptive Statistics

Enter column name (e.g., C1) in the “Variables” box, then hit “OK”.

Example

• 12% of Calpoly alumni make donations to Calpoly athletic programs.

• Less than 1% of non-Calpoly alumni make donations to Calpoly athletic programs.

• What are the two variables here, and are they independent or dependent?

Test Question

• Are the variables

“Blind/Not Blind”

and

“Has Good Color Vision/Has Bad Color Vision”

independent?

Red Blue Green

Square

Oval

Create the population so that Create the population so that color and shape are color and shape are independentindependent

Check Your SolutionReset

Populate It

Today’s Lecture

ExcitementExcitementLevelLevel

DifficultyDifficultyLevelLevel

Hypothesis TestingFrom now on: Your research hypothesis must be stated in the “there IS an effect” form,

e.g.,

Hypothesis: Adderall makes people better test takers. (From: Switzerland)

NOT:

Hypothesis: Adderall has no effect on test takers.

Hypothesis Testing

Hypothesis: Adderall makes people better test takers. Results of Experiment (Data):

Scored High on Exam

Scored Low on Exam

Take

Adderall

60 40

No Take Adderall

40 60

Are we done?

(Have we proven that Adderall makes people better test takers?)

Hypothesis Testing

Hypothesis: Adderall makes people better test takers.

Results of Experiment (Data):

Scored High on Exam

Scored Low on Exam

Take

Adderall

40 60

No Take Adderall

60 40

Can we say that Adderall helped these students?

We would like to compute: p(Adderall makes people better test takers | Data).

But this is not possible, because the research hypothesis is not ‘exact’:

How much better are test takers who take Adderall?

So, we need to think hypothetically:

What do we expect to see in our data if the research hypothesis is FALSE?

The Super Bowl Coin Flip Problem

Is the type of coin used at the start of the super bowl “Fair”?

Flip the coin n times.

Suppose that the coin is “fair”. Is the number of times it came up heads in the n flips very surprising?

If so, conclude that the coin is UNFAIR (publish an article about it).

Otherwise, keep your mouth shut (no publication).

Inferential Statistics: Step OneInferential Statistics: Step One

If you were to run your experiment over and over again, How much variability would there be in the sample means that you get?