# Unit 9: Surveys: Percentages and Averages

## Text reading and homework:

Read chapters 20, 21 and 23 of FPP and do the following review exercises:
Chapter 20 (pages 371-373): 1, 3, 4, 5, 6
Chapter 21 (pages 391-394): 1, 2, 3, 4, 11
Chapter 23 (pages 425-428): 1, 2, 3, 4, 5

"Pollster discusses McCain, Obama and the problems with polls," by Jeremy Duda, November 7, 2008, Arizona Capitol Times

Available via LexisNexis Academic(make sure to select the "News" tab instead of "General")

## Possible Essay Questions:

• What type of error does the official "statistical error" reported with most polls refer to? If the polls are "subject to statistical error of 2-3 percentage points", how can two polls differ by more than 6 points?
• What kind of problems does the article identify are made by typical election polls?
• How can we achieve the same margin-of-error by sampling 400 people, regardless of the population size?
• How can polls be used to influence subsequent polls, or the election for that matter?

## Computer project:

1. Use a spreadsheet to carry out 30 simulations of the two strategies (i) and (ii) discussed in Problem 9 of Chapter 17 (page 305). How do your simulation results compare with your theoretical answers for 9a), b) and c)? Explain.
2. A multiple-choice quiz has 25 questions. Each question has 5 possible answers, only one of which is correct. Four points are given for each correct answer, but a point is taken off for a wrong answer. Use a spreadsheet to simulate the scores of 100 students answering all of the questions at random. How many, and what percentage, of the students score at least 10 points? Compare your results with those obtained by theoretical computation using expected value and standard error. Repeat the simulation and comparison with 400 simulated students guessing randomly.

#### Simulation Tips:

Use each column of the spreadsheet as a single simulation of 1000 plays. The number in each cell should contain the winnings on one spin of the roulette wheel. [i.e. for 9(i) -- the columns strategy -- the contents of a cell might be
`=IF(RAND()<(12/38),2,-1).]`
Below each simulation, calculate the total winnings.

To count how many of the simulations "came out ahead", won 100 or lost 100, use the FREQUENCY function with the "data array" being the row of total winnings, and the "bin array" being -100,0,100 (put these numbers in three adjoining cells.) The function FREQUENCY counts how many times an outcome was less than each bin label: So, the number next to -100 is the number of times we lost more than \$100. The number next to 0 is the count between -\$100 and \$0. The number next to \$100 is the count between \$0 and \$100 and the last number is the count greater than \$100. (Note: four counts for three bin numbers.)