Analysis of Survey Data
Description: This course is designed to teach students appropriate analytic methods for answering research questions using existing survey datasets. We will begin by discussing the creation of analysis plans, which will include basic descriptive and inferential analytic methods along with more advanced regression methods. Students will learn how to apply regression methods to survey data, so must have a solid foundation in linear regression and a basic understanding of logistic regression. We will focus on learning the types of data that can be used with particular analytic methods, as well as on learning which types of analytic methods are appropriate for which types of research questions. As such, we will learn when and how to manipulate existing data through the creation of dummy variables or composite variables, through data transformations, and through coding procedures for changing text into analyzable numeric data. We will also learn how to assess the degree of missing data and how to handle problematic levels of missing data, how to interpret statistical results, and how to present findings in text and in graphic format. This course does not include data mining techniques, as we will focus on research question-driven analysis. We will use SAS 9.2 to conduct our analyses, and students will learn SAS procedures appropriate to those analyses. Students who have limited experience with SAS should attend optional Friday SAS workshops, in which they will receive basic instruction on navigating the SAS environment, importing datasets, and writing SAS code.
Learning Outcomes: At the end of this course you should be able to: - Assess the degree of missing data and determine whether missing data are associated with a particular set of respondents - Use appropriate coding techniques to transform text responses into analyzable numeric data - Manipulate data by creating dummy and composite variables - Create an analysis plan for answering the guiding research questions - Perform basic descriptive and relational analyses on collected data - Determine effects of explanatory variables on outcomes of interest using linear and logistic regression - Report survey findings in a clear, understandable manner that focuses on overall meaning rather than individual results