How to Automatically Document Data With the codebook Package to Facilitate Data Reuse

Abstract

Data documentation in psychology lags behind not only many other disciplines, but also basic standards of usefulness. Psychological scientists often prefer to invest the time and effort that would be necessary to document existing data well in other duties, such as writing and collecting more data. Codebooks therefore tend to be unstandardized and stored in proprietary formats, and they are rarely properly indexed in search engines. This means that rich data sets are sometimes used only once—by their creators—and left to disappear into oblivion. Even if they can find an existing data set, researchers are unlikely to publish analyses based on it if they cannot be confident that they understand it well enough. My codebook package makes it easier to generate rich metadata in human- and machine-readable codebooks. It uses metadata from existing sources and automates some tedious tasks, such as documenting psychological scales and reliabilities, summarizing descriptive statistics, and identifying patterns of missingness. The codebook R package and Web app make it possible to generate a rich codebook in a few minutes and just three clicks. Over time, its use could lead to psychological data becoming findable, accessible, interoperable, and reusable, thereby reducing research waste and benefiting both its users and the scientific community as a whole.