Building Macros for Quick Survey Scoring

Presentation Title
Building Macros for Quick Survey Scoring
Mr. Vincent Chan - Senior Research Analyst
September, 2017
IMPAQ Health

Vincent Chan presented a paper and discussed a technique for programmers to create construct scores.

APA Citation

Chan, V. (2017, September). Building Macros for Quick Survey Scoring. Presentation at the Western Users of SAS Software Conference, Long Beach, CA

Vincent Chan presented a paper that discusses a specific technique to go through an unlimited number of mathematical combinations of variables to create one single score. Surveys are a critical research component when gathering information about a population of people. Researchers often implement surveys after a population has participated in a class or educational program with the aim to capture knowledge, attitudes, and experiences of participants. Survey developers create surveys using psychometric principles to ensure the reliability and validity of each question, also known as an "item." A developer can purposefully create sets of items, which when analyzed as a group, can measure a construct that describes the underlying behavior, attribute, or characteristic of a study participant. The programmer calculates the construct score by combining (e.g. summing or averaging) all component items of the construct.

The programmer is tasked with creating such scores quickly. With an unlimited amount of time to spend, a programmer could operationalize the creation of these constructs by manually entering predefined formulas in the DATA step. More often, time and resources are limited. Therefore, the programmer is more efficient by automating this process using macro programs. In this paper, researchers present a technique that uses an externally created specification Excel file (spec) that contains the construct specifications and corresponding items of each construct. By looping through macro variables created from this spec file, they can create construct scores for an unlimited number of items. This technique can be generalized to other processing tasks that involve any number of mathematical combinations of variables to create one single score.