POLI 311 Chapter 0 : Polls, surveys and sampling error (Wilson)
Document Summary
Example: if you have an urn of white and red balls and you keep drawing samples of one hundred, each sample"s percentage of red would be collected and displayed in a distribuion. It would eventually form a bell curve, clustering around a mean. We know that some samples will be very representaive but others will not be. The larger the sample size the less likely that there will be sampling error; smaller samples will give later curves that are more likely to give values that difer further from the populaion value. Liberal voters is between 45. 6 and 54. 4 percent. this means that 1/20 (or 5% of the ime) we will choose a sample that poorly represents the populaion and is not close to the mean. Rather than having to create a distribuion to igure out the sampling error, we use a formula. It is: 1. 96x the square root of ((our percentage esimate) (100-our percentage esimate)/number of cases in the sample).