Slide 1Practice ProblemsAdditional Reading and ExamplesSlide 4Motivating ExampleArticlesArticlesProducing DataSampling ProceduresExperimental DesignsA. Sampling DesignsA. Sampling DesignsRepresentative SampleBiasSelection BiasSelection BiasSelection BiasNonresponse BiasNonresponse BiasResponse BiasResponse BiasResponse BiasSlide 23Haphazard SamplesVolunteer Response SamplesBiasRandom SamplesMotivating ExampleMotivating ExampleMotivating ExampleMotivating ExampleMotivating ExampleSTAT 210Lecture 3Producing DataAugust 30, 2017Practice ProblemsPages 38 through 42Relevant problems: II.1 and II.11Recommended problems: II.1 and II.11Additional Reading and ExamplesPages 26 through 29Particularly read #7 on page 29Top HatMotivating ExampleIt is of interest to determine what percentage of all VCU students who, when asked to list three things that they enjoy doing, include “working out/playing sport/gym” on their list.ArticlesThere are two articles, one on pages 11 through 14, and one on pages 15 and 16.Read the article on pages 15 and 16 of the Supplement and answer the three clicker questions related to this.Top Hat 3ArticlesOutside of class, read the article on pages 11 – 14and think about the population of interest, the sample, and any parameters and statistics.Producing DataTo perform any type of statistical analysis or statistical inference, we need data. How do we get this data?Sampling ProceduresWe select a sample of the population and only measure (or contact) the subjects in the sample.Experimental DesignsA treatment is imposed on the experimental units or subjects and a response is observed. You then can compare the effect that the treatment is having on the response.A. Sampling DesignsA population is a large set of subjects that the researcher wants to make statements about.Example: all VCU studentsThe problem is that in many cases time, cost, or inconvenience make it impossible to contact or measure every member of the population.A. Sampling DesignsInstead, we obtain a representative sample of the population and use the data in the sample to make inferences about the population.Example: We can easily contact a sample of 200 VCU studentsRepresentative SampleThe sample should be as representative of the population as possible, meaning that the characteristics of the sample should mimic the characteristics of the population.Example: If the population is 60% female, then the sample should be roughly 60% female as well if we think a person’s sex will influence the response.BiasBias exists when some subjects or outcomes are systematically favored over others. A sample which is representative of the population should be free of bias. If the sample is not representative, then the results will be biased in favor of the responses of those which are over-represented.Selection BiasSelection bias occurs when one or more types of subjects are systematically excluded from the sample.Selection BiasSelection bias occurs when one or more types of subjects are systematically excluded from the sample.Example: Suppose the population consists of all students at this university, and of interest is the opinion of students regarding an increase in student fees to pay for an expansion of the basketball arena. Contacting all students for their opinions is not possible, so instead a sample of 200 students in attendance at a basketball game is selected and they are asked their opinion. This would likely create selection bias because the opinions of students not attending the basketball game were not included. Most likely those attending the basketball game would be more willing to support an increase in fees for this purpose, while those who do not have an interest in basketball would be less likely to support such a fee increase.Selection BiasWhen selection bias exists, the results from the sample can only be inferred to part of the population. The inference cannot be made to the entire population, but only a subset of the population. This is referred to as undercoverage.Example: Instead of using the 200 responses to make inferences about the opinion of all students, the 200 responses can only be used to make inferences about the opinion of all students who attend basketball games. Opinions of students who do not attend the basketball games are not represented and hence inference cannot be made about these students – this is the undercoverage.Nonresponse BiasWhen an individual randomly chosen to be a part of the sample cannot be contacted or fails (or refuses) to respond, then we have a nonresponse bias. This is often a big problem in surveysor polls, in which the person either throws the survey away or refuses to answer the questions of those conducting the poll.Nonresponse BiasExample: Suppose the U.S. Government is interested in the opinions of citizens regarding the rights of illegal aliens in this country. A sample of 500 citizens is chosen such that a representative number of citizens from each racial and ethnic group is included. However, if all the Latino Americans refuse to respond, then we would have nonresponse bias if their opinions are different from those of other racial or ethnic groups.Response BiasWhen respondents give inaccurate information or if the interviewer influences the subject to respond in a certain way due to the way the questions are phrased, this is response bias. This is especially a concern with legal or social behavior issues.Response BiasWhen respondents give inaccurate information or if theinterviewer influences the subject to respond in a certain way due to the way the questions are phrased, this is response bias. This is especially a concern with legal or social behavior issues.Example 1: “Have you ever committed a crime that involved inflicting injury to the body of another person?”Due to social and legal concerns, the number of “no” responses will likely be higher than the actual number of people for which the correct response is “no”.Response BiasWhen respondents give inaccurate information or if theinterviewer influences the subject to respond in a certain way due to the way the questions are phrased, this is response bias. This is especially a concern with legal or social behavior issues.Example 2: Goal is to determine opinions about how safe people feel flying different airlines. Consider the question:“Do you feel safer flying on Delta Airlines as opposed to one of the other airlines?” This question would cause a
View Full Document