Simply put, metadata are data that describe other data. In practical terms, metadata are bits of information that allow people and/or machines to understand the meaning of the data being presented, such as the content typically found in the header row of a table or spreadsheet, or along the axes of a graph.
| Without metadata, data are difficult or even impossible to interpret. For example, if the header were missing from the table below, would it be clear that 2009-07-18 is a purchase date? It could just as easily be a publication date, the date the item was added, or some other code. |
Confusing or incomprehensible data can't be reviewed or used by others - and may not be of long-term value to you either - so aim to provide unambiguous and complete metadata for all your data-sets.
Metadata is highly-structured information describing your data - from individual variables to entire data-sets. Metadata are generally entered into pre-determined fields, e.g., Name, Program, Date, as metadata need to be both human-readable and machine-readable - which is more difficult when metadata are entered as unstructured free-text.
Documentation: covers a wide range of contextual information that makes your data comprehensible - both to yourself and others - at the project, file/database and variable levels, such as codebooks, questionnaires, file version history logs etc. To learn more see the Accompanying Documentation page in this guide.
