Overview

Abstract (English)

Every researcher has to prepare datasets, recode or create variables based on specific research hypotheses. These steps are inherent to the secondary data analysis research process. As they are idiosyncratic to each research project, these steps could not be routinized. In contrast, the preparation of a database for longitudinal analysis is the same, whatever the research hypotheses are. In the case of the National Population Health Survey (NPHS), data cannot always be directly used for longitudinal analysis as they are provided in a person-level format, whereas longitudinal analysis requires a person-period format. Commonly available statistical software therefore provide easy way to convert datasets, but the ease with which this can be done hinges on a convention around variable names. Indeed, to use macros, arrays or any routines in statistical software, variables should all have a name ending with a numerical form. In this regard, the naming system in the NPHS poses two challenges to researchers: First, the cycle is mentioned in the middle of the variable name, not at the end. Second, the identification of the cycle itself is a mix of numbers (from 1994 to 2002) and letters from (from 2004 onward). This presentation discusses these challenges and introduces a package developed to facilitate the transposition of data matrices from the NPHS from a wide to a long format, and back.

Abstract (French)

Please note that abstracts only appear in the language of the publication and might not have a translation.