CCT208H5 Lecture Notes - Lecture 3: University Of Toronto Mississauga, Snowball Sampling, Observational Error
Document Summary
In general, generalization requires both a large sampling size and a representable sample. Ie. students in high school, corporation, groups in a country, organization etc. matters less. Smaller and specific groups (ie. all canadian females age 20-25 that go to. More likely to reflect the entire population than non-random samples. The more likely you choose randomly, the more likely you have a diverse data for. The less likely you choose randomly, you may end up choosing a person who is very your research different than most people in a large population. Often impossible because they require a complete listing on a population. Ie. random telephone survey can be good. You need to check response rate and service (does the population have access to a telephone) If you see a research project using a random sample, you have to check the response rate as some response rates may be poor. 2016 - 9% response rate on telephones.