Challenges and Rewards of Editing Complex Survey Data from the National Ambulatory Medical Care Survey

• Activity Number: 488
• Type: Contributed
• Date/Time: Wednesday, August 12, 2015 : 8:30 AM to 10:20 AM
• Sponsor: Survey Research Methods Section
• Author(s): Kelly Myrick*
• Companies: National Center for Health Statistics
• Keywords: Physician ; Visits ; NAMCS ; SAS ; Error identification

Abstract

Editing survey data can be both challenging and rewarding. The National Ambulatory Medical Care Survey, a large nationally representative survey of physician office visits, had significant changes in methodology in 2012. We needed to edit the data efficiently and understand how the 2012 data compared to previous data. This involved analyzing different types of data (i.e. character, numeric, categorical, continuous) that needed differing methods of assessment. A partially automated SAS program was created to compare two years of data to 1.) Identify category response changes that exceed a specified threshold for categorical variables, 2.) Identify differences in the mean/median that exceeded two standard deviations for continuous variables, and 3.) List a user specified number of ranked character values. Some examples of the findings were: there were less missing values for height in feet in 2012 than in 2011, potential errors in decimal places were identified for a lab value, and most verbatim diagnoses matched for more than half of the top 12 values. Several valuable lessons were learned from creating this program; they will also be shared.