The Poll Project

In this assignment, you will perform a thought experiment in which you construct a poll. Our primary goal is to familiarize ourselves with some of the issues involved in conducting legitimate surveys. We will imagine that you are polling the Dartmouth student population about an issue relevant to you. Your goal is to design an “M&M” style survey for implementation on real people. This can be quite complicated! In class we handed out information that dealt with some of these complications; also read the discussion of polling issues at the American Statistical Association website.

For the written part of the assignment, please do the following:

1. Think of a topic on which the opinions of others genuinely interests you, and design a survey that attempts to capture this information.

2. Design two methods for choosing a random sample of Dartmouth students. Explore and articulate whether you feel there is any natural bias built into either of your survey methods. If so, decide whether the methods of weighting, clustering, or stratification (all found in the above general references) might help improve your findings. Explain why they might help.

3. Decide how many students you will survey and explain why you chose this number. Compute the margin of error associated to this sample size and explain what a margin of error means in general, and with respect to your results.

Contrary to the original statement on the class schedule, a first draft of this assignment is not due. However, we will be happy to look at first drafts submitted before Wednesday.

The finished assignment is due at the start of class on Friday, October 18.