Sampling Distributions!
•! What are Sampling Distributions?!
•! The Central Limit Theorem!
•! Probability!
•! Standard Error!
What are Sampling Distributions?!
•! Frequency distributions of statistics (usually
Ms) obtained by selecting all possible
samples of a given n from a population!
•! Allow us to determine:!
1.! if a sample is representative of a
population!
2.! the probability of obtaining any
particular sample from a population!
1
What are Sampling Distributions?!
Example: Distribution of sample means!
Repeat this process until all possible
samples have been drawn
n=2
2, 4, 6, 8
µ=5
What are Sampling Distributions?!
Example: Distribution of sample means!
SCORES SAMPLE MEAN
Sample First Second (M)
2
What are Sampling Distributions?!
Example: Distribution of sample means!
p(M < 3) = ?
= 1/16
Sample means from samples of n = 2
What are Sampling Distributions?!
Example: Distribution of sample means!
Characteristics!
1.! Sample means (M) pile
up around population
mean (µ)!
2.! Distribution of sample
means is approx.
Sample means from samples of n = 2
normal!
3.! With larger ns the
distribution becomes
less variable!
3
What are Sampling Distributions?!
The Central Limit Theorem!
For any population with mean µ and standard deviation !,
the distribution of sample means for sample size n will:!
1.! have a mean = µ!
2.! have a standard deviation that is called
a standard error and is = !/# n !
3.! approach a normal distribution as n
approaches infinity (n " 30)!
The Central Limit Theorem!
1.! Mean = µ ! central tendency!
M is an unbiased estimator of µ!
The mean of the dist. of sample
means is called the expected value
of M!
4
The Central Limit Theorem!
2.! Standard error = !/" n! variability!
•! the symbol for standard error is !M!
•! !M is measuring the standard distance between M
and µ!
•! !M is telling us exactly how different (on average)
M will be from µ!
•! as n increases !M decreases and therefore M
becomes a more precise estimator of µ !
The Central Limit Theorem!
2.! Standard error = !/" n! variability!
Effect of sample size (n)!
Sample! Standard Error!
Size! For ! = 10!
=!
=!
=!
=!
=!
=!
=!
=!
5
The Central Limit Theorem!
3.! Distribution is normal! shape!
IF!
•! The population from which the samples were
drawn is normal!
OR!
•! The number of scores (n) in each sample is " 30!
The Central Limit Theorem!
So What?!
Because the Distribution of Sample Means:!
1.! is normal in shape!
2.! has a predictable mean !
3.! has a predictable amount of
variability!
We can use z-scores & the Unit Normal Table to determine:!
1.! if any given sample is representative of the
population!
2.! the probability that any particular sample came
from a given population !
6
Probability!
Example!
Normally distributed population of SAT scores with the
following parameters: µ = 500 ! = 100!
If we randomly sample 25 students, what is the prob. that
M > 540?!
Based on CLT we know that the dist. of sample means:!
1.! is normally distributed!
2.! has a mean of 500!
3.! has a standard error of !M = 20!
!M = !/# n = 100/# 25 = 20!
Probability!
Using the Unit Normal Table To Calculate Probability!
•! Draw the distribution!
Step 1!
•! Shade the area you
are trying to find!
•! Convert M to z-score!
M - µ!
Step 2! z=!
! M!
540 - 500!
=!
20!
Step 3! Look up z-score in Table! p(z > 2.00) = 2.28%!
7
Probability!
Example!
Normally distributed population of SAT scores with the
following parameters: µ = 500 ! = 100!
If we randomly sample 25 students, what sample mean (M)
are we likely to obtain? !
Must define likely with a specific prob level; for this
example we will use 80%.!
Question becomes: What is the range of M values that
we would expect to obtain 80% of the time?!
Probability!
Using the Unit Normal Table To Convert Probability to M!
•! Draw the distribution!
Step 1!
•! Shade the area of
interest!
•! Look up % in Table
to determine z-score!
Step 2!
z = ± 1.28!
474.4! 525.6!
Step 3! Convert z-score to M!
M = µ + z!M!= 500 + (1.28)(20) = 525.6!
M = µ + z!M! = 500 + (-1.28)(20) = 474.4!
8
Standard Error!
•! Defined as !!/# n !
•! Specifies the average discrepancy between our
sample mean (M) and the population mean (µ)!
•! Specifies the reliability of our sample mean!
•! It is good practice to report SE along with M!
Standard Error!
Reporting SE in a bar graph!
9
Standard Error!
Reporting SE in a line graph!
The End!
For next class - read chapter 8!
For next weeks tutorial -!do the problems at
the end of chapter 7!
10