Sampling Distributions
Reese’s Pieces candies come in three colors: orange, brown, and yellow. Which color do you think has more candies in a package: orange, brown or yellow? I am asking you toguess,based on this picture, or on your own experience.
-
From my own experience I think that I pull out a lot of brown M&M’s. If I had to guess based off this picture I would choose orange because it seems like it makes up about 50% percent of M&M’s shown.
Math 243
Guess the proportion of each color in a bag. Give your answers as percentages:
Orange______48%______ Brown______30%______ Yellow_____22%________
Let’s think about theorange candies. Suppose we had an unimaginably large bag of Reese’s Pieces. If each student in our class took a sample of 25 Reese’s Pieces, would you expect every student to have the same number oforange candies in their sample? Explain.
Let’s say that 10 students each took a sample of 25 Reese’s Pieces. Write down the number of orange candies you might expect in each of these 10 samples, based on your answer to question 2:
________ ________ ________ ________ ________ ________ ________ ________ ________ ________
Of course, the number of orange candies in each sample shouldn’t be the same. These numbers represent theexpectedvariability you would expect to see in the number of orange candies in 10 samples of 25 pieces each.
Write your answers to the previous problem asproportions—either as decimals or percentages:
________ ________ ________ ________ ________ ________ ________ ________ ________ ________
The proportions in question 5 are calledsample statistics.This is a generic term – asample statistic is anything we use to summarize a sample. In this case, we could also call thesesample proportions, because we computed aproportion to summarize each sample.
We use the termpopulation parameter to refer to the value that summarizes the whole population. In this case, we are exploring thepopulation proportionof orange Reese’s Pieces.
For the purposes of the next five questions, proceed as though you and all the other students in the classactually drew a sample of 25 candies each.
Notation:
Thepopulation proportion is usually denoted by a lower-casep.
Thesample proportion is usually denoted by a lower-case p with a “hat” on it:p̂.We say “p-hat”.
Do we know the value of thepopulation parameterp (thepopulation proportion
Do we know the values of thesample statisticsp̂’s (thesample proportions of orange candies in each student’s sample of n = 25 candies)?
Does the value of the population parameterchange
Does the value of the sample statisticchange
Instead, we will simulate the activity using a web applet. Go to
http://www.rossmanchance.com/applets/OneProp/OneProp.htm?candy=1
You will see a cartoon container of colored candies: this represents the POPULATION.
You will see that the probability of orange is set at 0.5, so this is thepopulation parameter,which in this case is thepopulation proportionp. (People who have counted lots of Reese’s Pieces came up with this number, for example,this guy).
How does 0.5 compare to the proportion of orange candies in your guess in question 2? Were you close?
On the left side of the Reese’s Pieces web page, change “Number of orange” to “Proportion of orange”:
Click on the “Draw Samples” button. One sample of 25 candies will be taken and the proportion of orange candies in this sample is displayed in blue (under the candy machine, it will say something like, for example,p̂ = 0.640,depending on your sample and sample proportion), and also plotted on the graph.
Repeat this two or three more times.
Do you get thesame ordifferent
In general, about how close is eachsample statistic (sample proportion,p̂) to thepopulation parameter(population proportion,p
Why aren’t the values of the sample proportions theSAME as the population proportion (0.5)?
Unclick the “Animate”box and change “Number of samples” to 100.
Click the “Draw Samples” button, and turn your attention to the resulting distribution of sample proportions. (This is the graph with the dots below the candy machine.)
This distribution is calledThe Sampling Distribution ofp̂.
How is this distributiondifferent from the distributions we have worked with so far?
Repeat the “Draw Samples” exercise a couple times, until you start to see the general characteristics of the distribution. Describe theshape, center, and spread of this distribution.
Now we want to know: what happens to this distribution of sample statistics as we change the number of candies in each sample (sample size)?
First, change the sample size (“Number of candies”) to 10 and draw 100 samples.
How close is eachsample statistic to thepopulation parameter
Describe theshape, center, and spreadof this distribution of sample proportions with n = 10, particularly the similarities and differences with the sampling distribution we got when n = 25.
Next, change the sample size to 100 and draw 100 samples.
How close is eachsample statistic to thepopulation parameterDescribe theshape, center, and spreadof the distribution of sample proportions with n = 100, particularly the similarities and differences with the sampling distributions we got when n = 10 and n = 25.
Use the table to organize and summarize your results:
Sample Size
|
Shape of Sampling Distribution
|
Center of Sampling Distribution
|
Spread of Sampling Distribution(you can give the range here, as in “from about 0.24 to about 0.76”)
|
n = 10
|
|
|
|
n = 25
|
|
|
|
n = 100
|
|
|
|
How well do the sample statistics resemble the population parameter? Does it depend on the sample size?
What is the effect of sample size on theshape
What is the effect of sample size on thecenter
What is the effect of sample size on thespreadLet’s think about opinion polling. If we want to know: how a group of voters might vote, how a group of consumers feels about a product or service, how a group of students feels about a class, we might devise a question, select a sample, conduct a survey, and compute a sample proportionp̂.
How does the sampling distribution ofp̂help us to understand the relationship between our sample and the population? Put another way, how does the sampling distribution help us to determine how well our single sample statisticp̂ reflects the true population parameterp
">