← Back to course
Statistics 9-12 / Week 07 / Thursday
4/6
Week 07 · Does the Model Fit?

Thursday

Testing a claimed rate
// The pacing trial and a run of tails
⏱ about 20 min

Testing a Claimed Rate

"New model," Comet says, pinning the pacing plan to the board. "It claims a runner holds the planned lap 4/5 of the time."

"Then in 10 laps we expect 8 on-pace laps," Wren says. "Let us check the Run Log."

Nova scrolls. "The crew's 10 laps on Tuesday: 5 were on pace."

"5 instead of 8," Comet says. "Is that just chance? What do you notice about the spread?"

"√(10 × 4/5 × 1/5) is about 1.26," Wren says. "So 5 is about 2.4 standard deviations below."

"More than two," Comet says. "The plan's claim is in trouble."

Nova glows. "Would you like a hint? A surprising result questions the model. It does not prove a new one."

"Then we write down what we saw, and we test the plan again," Wren says.

Judging a claimed rate

A claimed rate is a model: "this works with probability p". The crew checks it the same way as the coin.

Expected count n × p = 10 × 4/5 = 8. Spread √(n p (1 - p)) = √(10 × 4/5 × 1/5) ≈ 1.26.

The crew saw 5. That is (8 - 5) ÷ 1.26 ≈ 2.4 standard deviations below the expected count. Beyond 2: surprising.

A surprising result means the model and the data disagree. The crew doubts the claim and tests again with more laps.

ClaimnExpected n × pSpreadSeenJudgment
fair coin, p = 1/21051.583consistent
pacing plan, p = 4/51081.265surprising
THE PACING PLAN
  • Read the question.
  • Tap your answer.
The plan claims a runner holds the planned lap 4/5 of the time. In 10 laps, how many on-pace laps does it expect?
A bar graph of a binomial model with 10 tries and success probability 4/5: the tallest bar is at 8 successes.The plan claims 4/5 of laps are on pace. In 10 laps the crew saw 5. Is that consistent with the claim?
A bar graph of a binomial model with 5 tries and success probability 4/5: the tallest bar is at 4 successes.If the plan's claim of 4/5 is true, what is the probability all 5 of 5 laps are on pace?
A bar graph of a binomial model with 5 tries and success probability 4/5: the tallest bar is at 4 successes.If the plan's claim of 4/5 is true, what is the probability at least 4 of 5 laps are on pace?

How to build a distribution from a sample space

The procedure behind every table this week. Read it first, then put it in order.

  1. List the sample space: every order the tries can come out, and check they are equally likely or know each one's probability.
  2. Define the random variable: the number you count in each outcome, such as heads.
  3. Group the outcomes by the value of the random variable.
  4. Find each group's probability: its count over the total, or the sum of its outcome probabilities.
  5. Write the table of values and probabilities, and check they add to 1.
  6. Find the expected value: each value times its probability, added up.
BUILDING A DISTRIBUTION, IN ORDER
  • ?List the sample space of outcomes
  • ?Write the table and check the probabilities add to 1
  • ?Group the outcomes by the value counted
  • ?Find each group's probability
  • ?Find the expected value
  • ?Define the random variable (what to count)
WHY THIS EXERCISEEvery distribution this week came from these six moves. The formula is a shortcut for the grouping step.
The number you count in each outcome, like heads, is called a random what? Type one word.
Under the plan's claim, how many standard deviations below expected was the crew's count? (1 place)
WHICH CONCLUSION IS JUSTIFIED?
  • Read the question.
  • Tap your answer.
The crew saw 5 on-pace laps when the plan expected 8, about 2.4 standard deviations below. Which conclusion is justified?
Tuesday's coin gave 3 heads in 10. Which conclusion is justified?
StatementTrue or false?
10 × 4/5 = 8?
A result more than 2 standard deviations from the expected count is surprising under the model.?
A surprising result proves exactly what the true rate is.?
The same two-standard-deviation rule works for a coin and for a claimed rate.?
WHY THIS EXERCISEJudging a claim against data is the heart of statistics, and the crew uses one rule to do it fairly.

Careful thinking. Tomorrow you take the model test into everyday life and review the week.

← Wednesday