MaizePrepSTATS 250
Exam 1 · Lec 05

The normal distribution and the Central Limit Theorem

Normal curves, standard scores for statistics, pnorm with the right spread, and why sample means and proportions end up bell-shaped.

25 min

Key ideas

Normal curve

Symmetric, bell-shaped, set by its mean and SD. Total area 1. Shifting the mean slides the curve; a larger SD makes it wider and flatter. Nearly all area is within 3 SDs of the center.

Central Limit Theorem

The sampling distribution of μ^\hat\mu is approximately normal if the population is normal (any nn) or n≥25n\ge25; for π^\hat\pi, if nπ≥10n\pi\ge10 and n(1−π)≥10n(1-\pi)\ge10. Averaging lets random pulls cancel, which builds the bell.

pnorm(q, mean, sd, lower.tail)

Area to the left of qq (lower.tail = TRUE) or to the right (FALSE). Lecture called lower.tail the most important argument: ask "am I shading below the cutoff?"

The what-if chain

When nn changes: does the center move or the spread? Does the curve compress or stretch? How many SD hops is the cutoff now from the center? Then read the area.

Formulas

Standard score of a statistic
z=μ^−μσ/nz=π^−ππ(1−π)/nz=\frac{\hat\mu-\mu}{\sigma/\sqrt n}\qquad z=\frac{\hat\pi-\pi}{\sqrt{\pi(1-\pi)/n}}
Same "observation − mean over SD" idea, with the sampling distribution's SD.
Probability in R
pnorm(0.32, 0.40, 0.04, lower.tail = TRUE)\texttt{pnorm(0.32, 0.40, 0.04, lower.tail = TRUE)}
Equivalent standardized form: pnorm(-2, 0, 1, lower.tail = TRUE); lecture preferred the standardized version.
Worked example

Satisfied subscribers

40% of a streaming service's subscribers rate it 5 stars. For a random sample of 150 subscribers, how likely is a sample proportion of 0.32 or less?

> sd <- sqrt(0.40 * 0.60 / 150)
> sd
[1] 0.04
> pnorm(0.32, 0.40, 0.04, lower.tail = TRUE)
[1] 0.02275013
  1. Check shape: 150(0.40)=60150(0.40) = 60 and 150(0.60)=90150(0.60) = 90, both at least 10, so π^\hat\pi is approximately normal.

  2. Model: center 0.40; spread 0.24/150=0.0016=0.04\sqrt{0.24/150} = \sqrt{0.0016} = 0.04.

  3. Standard score: (0.32−0.40)/0.04=−2(0.32 - 0.40)/0.04 = -2: the cutoff is 2 SDs below the center.

  4. Area: "0.32 or less" means shade below, so lower.tail = TRUE: about 0.0228.

Answer

About 2.3% of random samples of 150 subscribers would give π^≤0.32\hat\pi \le 0.32: unusual but possible.

Check yourself

1
Lec 05

A fitness tracker's step-count errors have mean μ=50\mu = 50 steps and σ=12\sigma = 12 steps. For random samples of 36 readings, what is the standard score of a sample mean of 53 steps? Give the exact value.

Math shortcuts are on for STATS 250. Type ^ then digits for powers, _ then digits for subscripts.
2
Lec 05

Continuing: which R command gives the probability that the mean of 36 readings exceeds 53 steps?

3
Lec 05

A shop's true return rate is 10%. Compared with samples of 200 orders, for samples of 800 orders the probability that π^\hat\pi is above 0.13 is:

4
Lec 05

Select every situation where the course says a normal model for the sampling distribution is justified.

5
Lec 05

A machine fills cups with a mean of 250 mL. Before a repair, a 238 mL cup had a z-score of −2; after the repair it has a z-score of −3, and the machine's mean is still 250 mL. What did the repair do?

6
Lec 05

The sampling distribution of μ^\hat\mu is normal with center 80 and SD 3. Between which values does nearly all of it lie?

MaizePrep is an independent study resource. It is not affiliated with, sponsored by, or endorsed by the University of Michigan. Course names and numbers are used only to describe which class the material covers. Every practice question is original, written in the style of the course.