Practical Tools for Designing and Weighting Survey Samples

<div>Preface</div><div>Acknowledgements</div><div><br></div><div>1 An Overview of Sample Design and Weighting</div><div>1.1 Background and Terminology</div><div>1.2 Chapter Guide</div><div><br></div><div>Part I Designing Single-Stage Sample Surveys</div><div>2 Project 1: Design a Single-Stage Personnel Survey</div><div>2.1 Specifications for the Study</div><div>2.2 Questions Posed by the Design Team</div><div>2.3 Preliminary Analyses</div><div>2.4 Documentation</div><div>2.5 Next Steps</div><div><br></div><div>3 Sample Design and Sample Size for Single-Stage Surveys&nbsp;</div><div>3.1 Determining a Sample Size for a Single-Stage Design&nbsp;</div><div>3.1.1 Simple Random Sampling</div><div>3.1.2 Stratified Simple Random Sampling</div><div>3.2 Finding Sample Sizes When Sampling with Varying Probabilities&nbsp;</div><div>3.2.1 Probability Proportional to Size Sampling</div><div>3.2.2 Regression Estimates of Totals</div><div>3.3 Other Methods of Sampling</div><div>3.4 Estimating Population Parameters from a Sample</div><div>3.5 Special Topics</div><div>3.5.1 Rare Characteristics</div>3.5.2 Domain Estimates<div>3.6 More Discussion of Design Effects</div><div>3.7 Software for Sample Selection</div><div>3.7.1 R Packages</div><div>3.7.2 SAS PROC SURVEYSELECT</div><div>Exercises</div><div><br></div><div>4 Power Calculations and Sample Size Determination&nbsp;</div><div>4.1 Terminology and One-Sample Tests</div><div>4.2 Power in a One-Sample Test</div><div>4.3 Two-Sample Tests</div><div>4.3.1 Differences in Means</div><div>4.3.2 Differences in Proportions</div><div>4.3.3 Special Case: Relative Risk</div><div>4.3.4 Special Case: Effect Sizes</div><div>4.4 R Power Functions</div><div>4.5 Power and Sample Size Calculations in SAS.&nbsp;</div><div>Exercises<div><br></div><div>5 Mathematical Programming</div><div>5.1 Multicriteria Optimization</div><div>5.2 Microsoft Excel Solver</div><div>5.3 SAS PROC NLP</div><div>5.4 SAS PROC OPTMODEL</div><div>5.5 R Alabama Package<div><br>&lt;<div>6 Outcome Rates and Effect on Sample Size</div><div>6.1 Disposition Codes</div><div>6.2 Definitions of Outcome Rates</div><div>6.3 Sample Units with Unknown AAPOR Classification</div><div>6.4 Weighted Versus Unweighted Rates</div><div>6.5 Accounting for Sample Losses in Determining Initial Sample Size</div><div>6.5.1 Sample Size Inflation Rates at Work</div><div>6.5.2 Replicates</div><div>Exercises</div><div><br></div><div>7 The Personnel Survey Design Project: One Solution&nbsp;</div><div>7.1 Overview of the Project&nbsp;</div><div>7.2 Formulate the Optimization Problem</div><div>7.2.1 Objective Function&nbsp;</div><div>7.2.2 Decision Variables&nbsp;</div><div>7.2.3 Optimization Parameters</div><div>7.2.4 Specified Survey Constraints&nbsp;</div><div>7.3 One Solution&nbsp;</div><div>7.3.1 Power Analyses</div><div>7.3.2 Optimization Results</div><div>7.4 Additional Sensitivity Analysis</div><div>7.5 Conclusion&nbsp;</div><div><br></div><div>Part II Multistage Designs&nbsp;<div>8 Project 2: Designing an Area Sample&nbsp;</div><div><br></div><div>9 Designing Multistage Samples&nbsp;</div><div>9.1 Types of PSUs&nbsp;</div><div>9.2 Basic Variance Results&nbsp;</div><div>9.2.1 Two-Stage Sampling&nbsp;</div><div>9.2.2 Nonlinear Estimators in Two-Stage Sampling&nbsp;</div><div>9.2.3 More General Two-Stage Designs&nbsp;</div><div>9.2.4 Three-Stage Sampling&nbsp;</div><div>9.3 Cost Functions and Optimal Allocations for Multistage Sampling&nbsp;</div><div>9.3.1 Two-Stage Sampling When Numbers of Sample PSUs and Elements per PSU Are Adjustable&nbsp;</div><div>9.3.2 Three-Stage Sampling When Sample Sizes Are Adjustable&nbsp;</div><div>9.3.3 Two- and Three-Stage Sampling with a Fixed Set of PSUs&nbsp;</div><div>9.4 Estimating Measures of Homogeneity and Variance Components</div><div>9.4.1 Two-Stage Sampling&nbsp;</div><div>9.4.2 Three-Stage Sampling&nbsp;</div><div>9.4.3 Using Anticipated Variances&nbsp;</div><div>The lme4 R package has been updated so that the syntax in the 1st edition no longer works. We will revise the examples in this section for the new version of the package.</div><div>9.5 Stratification of PSUs&nbsp;</div><div>9.6 Identifying Certainties&nbsp;</div><div>Exercises</div><div><br></div><div>10 Area Sampling</div><div>10.1 Census Geographic Units</div><div>10.2 Census Data and American Community Survey Data</div><div>10.3 Units at Different Stages of Sampling</div><div>10.3.1 Primary Sampling Units</div><div>10.3.2 Secondary Sampling Units</div><div>10.3.3 Ultimate Sampling Units</div><div>10.4 Examples of Area Probability Samples</div><div>10.4.1 Current Population Survey</div><div>10.4.2 National Survey on Drug Use and Health</div><div>10.4.3 Panel Arbeitsmarkt und Soziale Sicherung</div><div>10.5 Composite MOS for Areas</div><div>10.5.1 Designing the Sample from Scratch</div><div>10.5.2 Using the Composite MOS with an Existing PSU Sample</div><div>10.6 Effects of Population Change: The New Construction Issue<div>10.7 Special Address Lists</div><div>10.7.1 Allocations in ABS using Mathematical Programming&nbsp;</div><div>Mathematical programming allows efficient allocations to be made to domains (e.g., age groups) using information on housing units that can be purchased from commercial list makers. &nbsp;Discussion and examples will be added to illustrate this technique. &nbsp;The following article will be the basis for examples:</div><div><br></div><div>Valliant, R., Hubbard, F., Lee, S., Chang, W. (2014). “Efficient Use of Commercial Lists in Household Sampling”, Journal of Survey Statistics and Methodology, 2, 182-209.</div><div>Exercises</div><div><br></div><div>11 The Area Sample Design: One Solution</div><div><br></div><div>Part III Survey Weights and Analyses</div><div>12 Project 3: Weighting a Personnel Survey</div><div><br></div><div>13 Basic Steps in Weighting</div><div>13.1 Overview of Weighting</div><div>13.2 Theory of Weighting and Estimation</div><div>13.3 Base Weights<div>13.4 Adjustments for Unknown Eligibility</div><div>13.5 Adjustments for Nonresponse</div><div>13.5.1 Weighting Class Adjustments</div><div>13.5.2 Propensity Score Adjustments</div><div>13.5.3 Classification Algorithms</div><div>13.6 Collapsing Predefined Classes</div><div>13.7 Weighting for Multistage Designs</div><div>13.8 Next Steps in Weighting</div><div>Exercises</div><div><br></div><div>14 Calibration and Other Uses of Auxiliary Data in Weighting</div><div>14.1 Weight Calibration</div><div>14.2 Poststratified and Raking Estimators</div><div>14.3 GREG and Calibration Estimation</div><div>14.3.1 Links Between Models, Sample Designs, and Estimators-Special Cases</div><div>14.3.2 More General Examples</div><div>14.4 Weight Variability</div><div>14.4.1 Quantifying the Variability</div><div>14.4.2 Methods to Limit Variability</div><div>Exercises</div><div><br></div><div>15 Variance Estimation</div><div>15.1 Exact Methods</div><div>15.2 Linear Versus Nonlinear Estimators</div>15.3 Linearization Variance Estimation</div><div>15.3.1 Estimation Method</div><div>15.3.2 Confidence Intervals and Degrees of Freedom</div><div>15.3.3 Accounting for Non-negligible Sampling Fractions</div><div>15.3.4 Domain Estimation</div><div>15.3.5 Assumptions and Limitations</div><div>15.3.6 Special Cases: Poststratification and Quantiles</div><div>15.3.7 Handling Multiple Weighting Steps with Linearization</div><div>15.4 Replication</div><div>15.4.1 Jackknife Replication</div><div>15.4.2 Balanced Repeated Replication</div><div>15.4.3 Bootstrap</div><div>15.5 Combining PSUs or Strata</div><div>15.5.1 Combining to Reduce the Number of Replicates</div><div>15.5.2 How Many Groups and Which Strata and PSUs to Combine</div><div>15.5.3 Combining Strata in One-PSU-per-Stratum Designs</div><div>15.6 Handling Certainty PSUs</div><div>Exercises&nbsp;</div><div><br></div><div>16 Weighting the Personnel Survey: One Solution</div><div>16.1 The Data Files</div><div>16.2 Base Weights</div>16.3 Disposition Codes and Mapping into Weighting Categories</div><div>16.4 Adjustment for Unknown Eligibility</div><div>16.5 Variables Available for Nonresponse Adjustment</div><div>16.6 Nonresponse Adjustments</div><div>16.7 Calibration to Population Counts</div><div>16.8 Writing Output Files</div><div>16.9 Example Tabulations</div><div><br></div><div>Part IV Other Topics</div><div>17 Multiphase Designs</div><div>17.1 What is a Multiphase Design?</div><div>17.2 Examples of Different Multiphase Designs</div><div>17.2.1 Double Sampling for Stratification</div><div>17.2.2 Nonrespondent Subsampling</div><div>17.2.3 Responsive Designs</div><div>17.2.4 General Multiphase Designs</div><div>17.3 Survey Weights</div><div>17.3.1 Base Weights</div><div>17.3.2 Analysis Weights</div><div>17.4 Estimation</div><div>17.4.1 Descriptive Point Estimation</div><div>17.4.2 Variance Estimation</div><div>17.4.3 Generalized Regression Estimator (GREG)</div><div>17.5 Design Choices</div><div>&nbsp; &nbsp;17.5.1 Multiphase versus Single Phase</div>&nbsp; &nbsp;17.5.2 Sample Size Calculations</div><div>17.6 R Software</div><div>Exercises</div><div><br></div><div>18. Non-probability Samples</div><div>18.1 Types of Non-probability Samples</div><div>18.2 Potential Problems</div><div>18.3 Quasi-randomization Approach</div><div>18.4 Superpopulation Modeling Approach</div><div><br></div><div>19 Process Control and Quality Measures</div><div>19.1 Design and Planning</div><div>19.2 Quality Control in Frame Creation and Sample Selection</div><div>19.3 Monitoring Data Collection . . .</div><div>19.4 Performance Rates and Indicators</div><div>19.5 Data Editing</div><div>19.5.1 Editing Disposition Codes</div><div>19.5.2 Editing the Weighting Variables</div><div>19.6 Quality Control of Weighting Steps</div><div>19.7 Specification Writing and Programming</div><div>19.8 Project Documentation and Archiving</div><div>&nbsp;</div><div>Part V. Backmatter&nbsp;</div><div><br></div><div>Appendix A: Notation Glossary</div><div>Appendix B: Data Sets</div>Appendix C: R Functions Used in this Book</div>1 r="" overview<div>C.2 Author-Defined R Functions</div><div>References</div><div>Solutions to Selected Exercises</div><div>Subject Index<div><br></div></div></div></div>