Project Description

For this project, your group should design two opinion polls of the population of Penn State students. Each of the polls should allow you to estimate the proportion of Penn State students who have a particular characteristic or opinion.

• The first poll should be a "good poll," in that it adheres to all of the sound statistical practices of sampling, questioning, analyzing, and interpreting.
• The second poll should be a "bad poll," in that it severely violates one or many of the good polling practices.

The two polls can either pertain to the same topic or not.

For each poll, write a "How To Manual" that tells the other groups clearly, but concisely, how they should conduct your poll. A clear manual will leave the other groups with no questions about how to conduct your poll. The manual should:

• Describe specifically how to select your sample. (The method should be realistic if given the time and a limited budget.)
• Indicate the specific question(s) you want to ask each person in the sample.
• Describe specifically how you want to collect, record, enter, and analyze the data in a software package like Minitab. (One way of collecting and recording the data is by using a web-based or a paper data collection form. Typically, data are entered into software by "coding" the data with numbers, like 1 = "yes" and 0 = "no.")

For each poll, write a "News Article" that describes the results of your poll. Since you won't actually conduct your polls, you'll have to make up believable results. For the "good poll," your news article should explain the results as statistically and soundly as possible, and yet be understandable to a person with limited statistical knowledge. For the "bad poll," your news article should be convincing and yet leave as many questions unanswered as possible.

Useful resources

• Chapter 4, Utts/Heckard, describes different sampling methods, discusses difficulties and disasters in sampling, and addresses how to ask good questions.
• Chapter 10, Utts/Heckard, outlines how to calculate and interpret confidence intervals for a proportion.
• Unit A2, Cyberstats, describes different sampling methods, and outlines how to select a simple random sample and how to select a stratified random sample.
• The National Council on Public Polls has a nice web site that addresses some of the important poll issues you've read about. See for example, 20 questions a journalist should ask about poll results, answers to questions we often hear from the public, and internet polls.
• An example of a web-based data collection form.
• A good New York Times news article that presents the results of a good poll. A little long, but good explanation of statistical methods and interesting results.

What to hand in? and when?

For the good poll:

• Submit a typewritten "How to Manual"
• Submit a typewritten "News Article"

For the bad poll:

• Submit a typewritten "How to Manual"
• Submit a typewritten "News Article"

Make sure you include your group number, and the name of each group member that participated in the solutions. Be prepared to share and hand in your two manuals and two articles at the beginning of next week's lab.