5/2/20131STA291Statistical MethodsLecture 2Common Languageo Populationo Sampleo Sampling Frameo Census versus censuso Parametero StatisticUniversity Health Services at UK conducts a survey about alcohol abuse among students. Two-hundred of the 30,000 students are sampled and asked to complete a questionnaire. One question is “Have you regretted something you did while drinking”?• What is the population? Sample?For the 30,000 students, of interest is the percentage who would respond “yes”.• Is this value a parameter or a statistic?The percentage who respond “yes” is computed for the students sampled.• Is this a parameter or a statistic?Example5/2/20132The Current Population Survey of about 60,000 households in the United States in 2002 distinguishes three types of families: Married-couple (MC), Female householder and no husband (FH), Male householder and no wife (MH).• It indicated that 5.3% of “MC”, 26.5% of “FH”, and 12.1% of “MH” families have annual income below the poverty level.• Are these numbers statistics or parameters?Example 2Samplingsimple random sample: sample drawn so that every possible sample of the size we plan to draw has an equal chance of being selected; usually abbreviated SRS.Other (good) Sampling Methodso stratified random samplingo cluster samplingo multistage sampleo systematic sampling5/2/20133Other (not-so-good) Sampling Methodso voluntary responseo convenienceSampling: Famous (?) Exampleo 1936 presidential electiono Alfred Landon vs. Franklin Roosevelto Literary Digest sent over 10 millionquestionnaires in the mail to predict theelection outcomeo More than 2 million questionnairesreturnedo Literary Digest predicted a landslidevictory by Alfred LandonSampling: Famous Example (cont’d)o George Gallup used a much smallerrandom sample and predicted a clearvictory by Franklin Roosevelto Roosevelt won with 62% of the voteo Why was the Literary Digest predictionso far off?5/2/20134Exampleo TV, radio call-in pollso “Should the UN headquarters continue tobe located in the US?”o ABC poll with 186,000 callers: 67% noo Scientific random sample with 500respondents: 28% noo The smaller random sample is much moretrustworthy because it has less biaso undercoverageo nonresponseo response biasOther Problems, Sampling and OtherwiseWhy the concern?o Biaso Sampling variability versusNon-sampling variability5/2/20135Experiments Looking backo Simple random sample, or SRSo Other sampling methods, good and not so goodo Sources of sampling and non-sampling erroro