Menu

Lesson plan · Grade 11 · 50 minutes

Poll Workshop

I can explain how the way a sample is chosen and a question is worded bias a survey, estimate a proportion with a margin of error, and explain what the margin of error does not cover.

▶ Open the tool Download the tool
Standard
S-IC.A CCSS-Math
S-IC.B CCSS-Math
S-ID.A.4 CCSS-Math
Shape
Explore first
Grouping
Pairs on devices, with whole-class pauses.
You need
The tool on the projector and on student devices. Calculators.
Tool
curioclass.org/tools/poll-workshop/
  1. Hook5 min
  2. Explore17 min
  3. Make sense10 min
  4. Apply13 min
  5. Exit check5 min

Before class

  • Open Workshop with Simple random, Neutral and n = 400.
  • Millbrook is an invented town of 20,000. The tool is seeded, so a class that loads together sees the same town.
  • Find a real survey question from an ad or the news for the Apply part.

Hook 5 min

  1. "In 1936, a magazine polled 2.4 million people and predicted the president would lose in a landslide. He won 523 electoral votes to 8. How?"
  2. "Today you find out how a giant poll can be so wrong."

Explore 17 min

  1. Press Run poll and read the headline card. Point at the map: filled dots said yes, hollow dots said no.
  2. Press Run 50 more like this. "Where do you think the true answer is?" Then press Reveal the truth. About 95 percent of the plus or minus bars catch it.
  3. Pairs switch to Mail to car and phone owners, slide n up to 2000, and run 50 more. "More people didn't fix it. Why?"
  4. Pairs try other methods and the Leans yes and Leans no wordings. They record the average and how many polls caught the truth for each design.

Make sense 10 min

  1. "The mail poll printed plus or minus 2.0 points. What does that number promise, and what doesn't it cover?" (Only the luck of the draw, not bias.)
  2. Work the margin of error for n = 400 and p = 0.45 on the board: 1.96 × √(p(1 − p)/n).
  3. "Why does Cluster spread about twice as wide with the same n?"
  4. Turn on Normal curve and look at the spread of each row.

Apply 13 min

  1. Pairs open Case file and run both buttons: the Digest's method and a fair method. Discuss the three prompts.
  2. Pairs open Budget desk for round 1 and design the best poll they can for $10,000, then Publish.
  3. Pairs rewrite the real survey question you brought as a neutral question.

Exit check 5 min

On a card: "A poll of 900 randomly chosen students finds 54 percent support a later start. Find the margin of error and write a headline using the tool's rules." Then: "A poll of 5,000 people who visit a gaming website finds 80 percent support. What's wrong with it?"

Support and stretch

  • Support: compare only two designs: Simple random and the mail method.
  • Support: use Show to filter the wall to one method at a time.
  • Stretch: play round 3 of Budget desk, the close race. "Why can't a $10,000 poll say 'backs' with confidence?"
  • Stretch: explain why systematic samples spread a little less than simple random ones in Millbrook.

Short on time

  • Run 50 more like this for Simple random, then the mail method (10 min).
  • Reveal the truth and discuss (5 min), then the exit card (5 min).

How each button works is in the teacher notes. From Curio Class, free under CC BY 4.0: change it, print it, share it.