STAT 100 Lecture Notes - Lecture 4: Pew Research Center, Non-Sampling Error, Simple Random Sample

• School: Simon Fraser University
• Department: Statistics
• Course: STAT 100
• Professor: Gamage Perera

Document Summary

Chapter 4: sample surveys in the real world. Pew research center did a survey of randomly selected households. Needed a sample size of 1000 responses. Had to call 2879 households to get their 1000 responses. Nonresponse rate was 1879 = 0. 653 (65. 3%) Sampling errors: often caused by poorly designed surveys. Random sampling error: nothing can be done to eliminate this (can explain using. Nonsampling errors: mistakes unrelated to sampling issues. Important: the announced margin of error of a survey result only accounts for random sampling error. The sampling frame is a list of all the individuals from which we will draw our sample. This is often only hypothetical we cannot actually list all individuals in the population of interest. What if we used a list of household addresses? (from tax records?) When we leave certain groups in the population out of our sampling process, the result is undercoverage. Hard to eliminate these sampling errors, but their effect is often minimal.