Complex Survey Data Analysis with SAS

Hardcover, 2018 reprint, xiii + 326 pages, NOT ex-library. Some very minor handling wear, book is clean and bright with unmarked text, free of inscriptions and stamps, firmly bound. Issued without a dust jacket. -- Complex Survey Data Analysis with SAS is a practical, technical reference focused on applying the SAS/STAT SURVEY procedures to analyze data derived from complex survey designs. It assumes a foundational understanding of standard statistical methods and is intended for analysts working with data that do not conform to the assumptions of simple random sampling. The book begins with a detailed conceptual introduction to the four features that render a survey "complex": finite population corrections (FPCs), stratification, clustering, and unequal weights. These features are unpacked using hypothetical examples and real-world surveys, such as the National Ambulatory Medical Care Survey, the National Survey of Family Growth, and the Commercial Buildings Energy Consumption Survey, providing applied context for methodological explanation. Each chapter then focuses on a specific analytical tool within the SAS/STAT SURVEY suite. Chapter 2 addresses sampling designs using PROC SURVEYSELECT, including simple, systematic, stratified, and probability-proportional-to-size (PPS) techniques. Subsequent chapters cover descriptive and inferential analysis of continuous and categorical variables using PROC SURVEYMEANS and PROC SURVEYFREQ, respectively, highlighting correct variance estimation under different design features. Linear regression modeling with complex survey data is treated using PROC SURVEYREG, while logistic regression models - binary, multinomial, and ordinal - are analyzed using PROC SURVEYLOGISTIC, with emphasis on proper variance estimation and model fitting. Chapter 7 introduces survival analysis with complex data via PROC SURVEYPHREG and discrete-time models using PROC SURVEYLOGISTIC, demonstrating how time-to-event data from stratified or clustered samples can be analyzed with SAS syntax. Later chapters address more advanced topics often overlooked in introductory texts. Chapter 8 explores domain estimation, including the risks of data subsetting and proper use of the DOMAIN statement for subgroup analysis. Chapter 9 presents replication methods for variance estimation, such as balanced repeated replication (BRR), Fay's method, jackknife, and bootstrap techniques, discussing their applicability, benefits, and implementation in SAS. Chapters 10 and 11 focus on missing data, offering procedures for weight adjustment (including poststratification, raking, and propensity models) and imputation. The imputation chapter explains single and multiple techniques, addresses univariate and multivariate missingness patterns, and discusses how to incorporate design features into imputation models and post-imputation analyses. The book's strength lies in its clear procedural structure and commitment to practical implementation: each method is paired with worked SAS examples, using consistent survey data contexts. Variance estimation is a recurring theme, with each procedure's ability to handle complex design features explicitly demonstrated. Throughout, the text emphasizes the statistical consequences of ignoring survey design features and guides users on using SURVEY procedures instead of their standard SAS counterparts. Rather than attempting to replace theoretical texts on sampling (like those by Kish, Cochran, or Lohr), this volume serves as an applied bridge between theory and software use, specifically within the SAS environment.