You are on page 1of 14

Sampling Distribution

* Why do we use samples? Due to the data availability and cost, we have to use samples instead of population data. * Why do we need to know the sampling distribution? Before we use sample to make inference about the population, we need to know the feature of sample distribution.

Sampling Distribution of Statistics


*Statistics: numerical descriptive measures calculated from a sample are called statistics. We will be using means and proportions. *Sampling distribution of a statistic (mean or proportion) is the probability distribution for all possible values of the statistic that results when random samples of size n are repeatedly drawn from the population.

Sampling Distribution of the Sample Mean


You are taking a population (N) and pulling out of that population a sample of size (n) then calculating its mean You do this for all possible combinations of n out of N and you construct the sampling distribution of the sample mean from those means

Sampling Distribution of the Sample Mean


Example: from Berenson and Levine 1996
A certain population consists of 4 typists, the typists make 3, 2, 1, and 4 mistakes respectively. You take a random sample of two typists from that population with replacements, order does matter. 1) What is the population mean and standard deviation? 2) What are all the possible combinations of samples? 3) Construct a sampling distribution of the sample mean from this example, what is the sample mean and standard deviation?

Some facts about sampling distribution of the sample mean:


Fact 1: if a random sample of n measurement is selected from a population with mean and standard deviation , the sampling distribution of the sample mean will possess a mean

x =

And standard deviation (called the standard error) for a sufficiently large population x = n

Fact 2: if the population possesses a normal distribution, then the sampling distribution of x will be exactly normally distributed, regardless of the sample size n

Fact 3: if the population is non-normal, the sampling distribution of x will be closer and closer to a normal distribution with the rise of sample size n

The Central Limit Theorem


If random samples of n observations are drawn from a non-normal population with finite mean and standard deviation , then when n is large , the sampling distribution of the sample mean x is approximately normally distributed with mean and standard deviation:

x =

and

The approximation will become more and more accurate as n becomes larger and larger

*A rule of thumb: sampling distributions of x

will be approximately normal for sample size as small as n=25 for most populations of measurements.
*Example: suppose that you select a random

sample of n=25 from a population with mean=8 and standard deviation=0.6. 1) Find the approximate probability that the sample mean x will be less than 7.9 2) Find the approximate probability that the sample mean x will lie within 0.1 of the population mean 8

The sampling distribution of a sample proportion


*sample proportion: a random sample of n objects is selected from the population and if x of these possess the specified characteristic, then the sample proportion is:

p=x n
Note: We are working with Binomial distributions here useful with survey response data

Sampling distribution of a proportion


It is known that the population proportion of those who like economics is 0.5. You go and ask 6 people if the like economics.
Create the probability distribution table for your random variable the proportion of those who like economics. What is the expected value (average) and standard deviation of your random variable?

Properties of sampling distribution of the sample proportion


1. If a random sample of n observations is selected from a binomial population with parameter p, the sampling distribution of the ^ sample proportion

p=x n
^

will have a mean


pq n

=p
p

and a standard deviation:


=
^

where q=1-p

2. When the sample size n is large, the sampling distribution of sample proportion will be approximately normal. Remember the rule is np and nq both greater than or equals to 5

- this is a result of our ability to approximate binomial distributions with normal distributions

Example: According to the recent poll, about 46% Americans approve of Bushs overall status. Now we select 100 people and ask for their opinions. 1) What is mean of sample proportion 2) How many people in our sample are expected to think the policy is unsuccessful? 3) What is the probability that over 50% of sample approve of Bushs overall status?

You might also like