Modules

Variables are formatted within the round or wave sections within each source scrip. Variables are renamed for consistency, valid values are re-ordered but not transformed (so all data is conserved), and missing or invalid values are re-coded to match a common framework shared by all modules. After the formatting process are rounds and sources are merged together for each module.

Each module includes a labeling script and a harmonization script. The labeling script applies variable and value labels to every formatted variable within the module. This happens after data from all sources is merged together to avoid repeating the labeling process within each source. Doing it at this point is more efficient because the same formatted variables often appear across sources where the question and answers are the same. The harmonization script creates common variables that combine data from many sources. Usually this is achieved by reducing the diversity of answers to a least common denominator shared by all or most variables. However, since we maintain the original values of all formatted variables it remains possible to easily create additional harmonized variables.

Finally, each module includes a codebook. We generally maintain one main master codebook that includes all variables from all modules. However, when a module is under active development we parse out the relevant section to create a codebook for just that module. Then when development work is complete we recombine the module codebook into the main codebook. This only happens after undertaking quality control checks on all scripts to fix errors and ensure the data matches the module codebook.