POLI 30 Lecture Notes - Lecture 3: Statistic, Observational Error, Push Poll
Document Summary
The four types of variables: nominal, ordinal, ratio, interval. Sometimes we have all the data- a census. Often we study contexts where we can"t get everything else. Or we use methods where we can"t do it all . Sampling is important for questions of inference. Our sample says blah, what can we say about the population?". We want to know whether the pattern we observe is more than just random error. Suppose we find that the percent of our sample (sample statistic) favoring clinton is 69% (sample size = 1000, 690 favor clinton) Well it"s possible that 37,00,00 californians support trump. If your sample is properly conducted, you can say the following. I am 95% sure that 69% plus or minus 3% of californians support hillary. If the sample is close to a 1,000 the margin of error is almost always 3% Essential to political science, especially the study of voting behavior and public.