Chapter 9.1: Sampling Distributions

Chapter 9.1: Sampling Distributions

Mr. Lynch

AP Statistics

The Heights of Women

The heights of women in the world follow: N(64.5, 2.5) … Explain …

Let’s draw a sketch that helps illustrate this MATH … PRB … 6:randNorm(64.5,2.5) Stand up if your value is between [62, 67] Stand up if your value is between [59.5, 69.5] Stand up if your value is between [57, 72]

The Heights of Women

MATH … PRB … 6:randNorm(64.5,2.5, 100) STO L1 1-Var Stats: Mean? Median? S? STAT PLOT 1: Histogram … L1, 1 WINDOW: X:[57,72, 2.5] …Y:[-10,60,10] STAT PLOT 2: Boxplot … L1, 1 TRACE Histogram … Enter frequencies is chart Repeat three times … fill out frequency chart as

shown

The Heights of Women

Interval Set #1 Set #2 Set #3 Total %

57-59.5 3 2 3 8 2.7

59.5-62 8 15 11 34 11.3

62-64.5 39 33 33 105 35.0

64.5-67 37 38 32 107 35.7

67-69.5 11 10 18 39 13.0

69.5-72 1 2 2 5 1.7

95.070.799.4

Pooled Data Period 03 – January 2008

Interval LynchRow

1Row

2 Row 3Row

4Row

5Row

6 Total %

57 - 59.5 11 27 37 23 40 38 20 196 2.7%

59.5 – 62 40 158 208 120 180 155 126 987 13.7%

62 - 64.5 91 423 529 306 392 401 298 2440 33.9%

64.5 – 67 102 418 503 318 409 398 323 2471 34.3%

67 - 69.5 43 148 184 106 147 158 111 897 12.5%

69.5 – 72 13 25 39 26 32 50 19 204 2.8%

The Heights of Women

How did the “Empirical Rule” work out for you? What do the Shape, Center, and Spread look

like? Let’s look at the n = 7500 histogram! How are we doing now? Conclusion: This distribution is just a miniature

version of the population distribution with same mean and standard deviation

The Heights of Women

Now, take 4 samples again … and one at a time – Use 1-Var Stats to get the mean .

Write that value on one of your post-it notes. Repeat this 3 more times. Place the notes upon the board CAREFULLY

in the correct slots to build a histogram! Let’s record the values in L2.

X

The Heights of Women

How did the “Empirical Rule” work out here? Compare a Boxplot for L2 in PLOT 3 – to the one we

did in PLOT 2 for the population. What do the Shape, Center, and Spread look like for

THIS NEW distribution? Let’s look at the new SAMPLING DISTRIBUTION of

Sample means of n = 100 histogram! Conclusion: What is the relationship between the

mean of the population and the mean of the X bars? What about the standard deviation of the population and that of the X-bars?

Terminology

Population Parameter-– Numerical value that describes a population– A “mysterious” and essentially unknowable –

idealized value.– A theoretically fixed value– Ex: Population Mean, Population Standard

Deviation, Population Proportion, Population Size

, , ,p N

Terminology

Sample Statistic– Numerical value that describes a sample (a subset of

a larger population)– An easily attainable and knowable value– Will vary from sample to sample– Used to estimate an unknown population parameter– Ex: Sample Mean, Sample Standard Deviation,

Sample Proportion, Sample Size

ˆ, , ,X s p n

Example and Exercises

EXAMPLE 9.1: MAKING MONEYEXAMPLE 9.2: DO YOU BELIEVE IN GHOSTS?EXERCISE 9.2: UNEMPLOYMENTEXERCISE 9.4: WELL-FED RATS

Sampling Variability

What would happen if we took many samples?

EXAMPLE 9.3 BAGGAGE LUGGAGE

Sampling Variability

Sampling Distribution: of a statistic is the distribution of values in ALL POSSIBLE samples of the same size

EXAMPLE 9.4 RANDOM DIGITS

Describing Sampling Distributions

EXAMPLE 9.5: ARE YOU A SURVIVOR FAN?

1000 SRSs; n = 1000; p = 0.371000 SRSs; n = 100; p = 0.37

Using the same x-axis scale as to the left!Using a scale to show shape!

UNBIASED vs. BIASED

A Statistic is said to be UNBIASED if the mean of the sampling distribution is equal to the true parameter being estimated

When finding the value of a sampling statistic, it is just as likely to fall above the population parameter as it is to fall below it.

VARIABILITY of a STATISTIC

The larger the sample size, the less variability there will be

EXAMPLE 9.6: THE STATISTICS HAVE SPOKEN– 95% of the samples generated: Mean ± 2 Sd – With n = 100 …0.37 ± 2 (0.05) = 0.37 ± 2 (0.05)

[0.32, 0.42]– With n = 1000 …0.37 ± 2 (0.01) = 0.37 ± 2 (0.01)

[0.35, 0.39]

The N-size is irrelevant! Accuracy for n = 2500 is the same for the entire 280M US, as it is for 775K in San Fran

BIAS & VARIABILITY (Revisited)

Precision

versus

Accuracy

BIAS & VARIABILITY (Revisited 2)

Homework Example EXERCISE 9.9: BEARING DOWN

p = 0.1; 100 SRSs of size n = 200 Non-conforming ball bearings out of 200 are shown:

(a) Make a table that shows the frequency of each count! Draw a histogram of the p-hat values.

(b) Describe the shape of the distribution.(c) Find the mean of the distribution of p-hat; mark it on

the histogram. Any evidence of bias in the sample?

(d) What is the mean of “the sampling distribution” of all

possible samples of size 200?

(e) What is we repeated this exercise, but instead used SRSs of size 1000 instead of 200? What would the mean of this be? Would the spread be larger, smaller or about the same as the histogram from part (a)?