Description

This lesson explores the influence that lurking variables can have on data and statistical inference.

Materials

- Overhead transparencies (if student responses will be displayed on the overhead)
- Marking pens (for overhead or white board)
- Pickles: Death in a Jar! (see Attached File)
- Save Oceana! (see Attached File)

Procedures

1. Distribute Pickles: Death in a Jar! (see Attached File) and read it with a very somber tone in front of the class.

2. Students share their thoughts on the data presented in a class discussion. This discussion should conclude with most students agreeing that pickles are not the demons that they appear to be.

3. Introduce the concept of lurking variables and show how this concept could lead to very misleading statistical inferences about the beloved pickle. Ask students what are some limitations in using statistical techniques and data in making inferences and valid arguments? Use the pickle statistics as examples.

4. Make it clear at this point that the students' jobs as statisticians is to investigate data thoroughly so that incorrect inferences about data are not reached. Students should understand that statistical data in itself (without analysis) can be very misleading.

5. The students will analyze the pickle data and come up with a list of possible lurking variables. (skewed, omitted, presented in a negative or positive light based on the presenter's ideas, compared, contrasted, presented in a circular method or biased method, etc.

6. Place the students' ideas on an overhead or white board and discuss with the class the merits of the responses (being careful to give positive feedback for all of the responses).

7. Continue instruction by tying in the concept of correlation with lurking variables. Pickle consumption and mortality are positively CORRELATED, but pickles do not CAUSE death!

8. Distribute Save Oceana! (see Attached File) and instruct the students to form groups of 3-5 students (depending on class size) and discuss the information therein.

9. Students should be given 10-15 minutes to confer in their groups before reassembling for class discussion. During this period circulate around the room providing formative feedback and providing guidance to the student groups.

10. Students should then reassemble and the instructor should lead a discussion where every group presents their findings and conclusions to the class.

11. Students then write a paragraph answering this question: What are some limitations in using statistical techniques and data in making inferences and valid arguments?

Web Links

This site is a great tutorial on correlation vs. causation in slide show format.
Correlation and Causation