 Describing and documenting data are crucial tasks in research data management. Who was involved in the data collection? Why were these variables collected and, from where did you get the data? And when? And which methods were used to collect the data, and to process and analyze it? Documenting all this information requires a substantial effort. So why is it so important? Because research results are more easy to review and verify when they are well documented. Besides, they become more reproducible, and it can enhance their impact. It is also important because it enables research data to be understood and reused by others, including by yourself in the future. In other words, it's a crucial step into making your data fair. The earlier you start documenting, the easier it will become. At the start of your project, it is crucial to plan what documentation will be captured, and how you will store it. Once you have a strategy, it is a good idea to keep documenting as you go along, while the steps you've been taking are still fresh on your mind. When talking about documentation, a term that often pops up is metadata. Documentation and metadata are two related concepts, and they are even sometimes used as synonyms. But they are not exactly the same. Documentation is a generic term that refers to any sort of contextual information that allows data to be understandable and reusable. Metadata, on the other hand, is a particular form of documentation. It's highly structured and built up from a set of elements or fields. Metadata is especially meant to be read by computers. For instance, to be used in online applications such as data catalogs or repositories. In this video, we will focus on documentation. Let's have a look at more details. Documentation should happen at different levels. At the study level, at the level of files or databases, and at the variable or item level. Study level documentation refers to high level information about your study. At this level, you describe the research context and design, the data collection methods used, but also the potential sources of the data. Documenting at the study level also involves the recording of any data preparation, manipulation and validation steps. It can also include the summary of preliminary findings and results, and any other contextual information that is relevant. For example, if your research involves surveys, the instructions given to respondents and the informed consent template used should also be part of the study level documentation. This documentation comes in various forms. It might be recorded in notebooks, either analog or digital. Other pre-existing material can be considered study level documentation as well. For example, software guidelines or standardized protocols or standard operating procedures. Part of the documentation might be provided in reports or publications, but also in readme files. File or database level documentation refers to information that describes the structure of the dataset. It typically provides an inventory of the folders and files, and how these relate to each other. Very frequently, you will find this information in readme files. For databases, the documentation often includes diagrams of the database structure, showing the relations between the different tables or entities. In qualitative research, data listings are often used to keep track of the content, the location and characteristics of all the data or files that are generated. Variable or item level documentation, provides information about the content of the datasets, at the level of variables or individual objects. For example, interview transcripts or pictures. For quantitative data, this documentation can be included in so-called code books or data dictionaries. In these code books, all the different variable names and labels are listed, together with their meaning. Attributes of each field or variable are recorded, for example, expected values, measurement units, how missing values are treated, etc. This is also the place to document the codes, abbreviations or classification schemes that are used. Furthermore, documentation at this level should also cover the processing steps that were carried out to clean and prepare datasets. For instance, by aggregation, derivation or grouping of variables into new data. When documenting the processing steps, it is important to not only explain what has been done, but also why these processing steps have been executed. Some data file formats allow variable documentation to be embedded within the data file. This can be handy, but remember that there might be limitations in terms of how much documentation can be embedded. Besides, some information can be lost when converting or migrating to other formats. Therefore, an alternative is to keep your code book in a separate file. If you develop or use a script to process or analyze your data, it is a good practice to annotate your code with human readable comments which explain the different functions that the code is performing. In research that uses qualitative data, other solutions are necessary. For example, when performing interviews, it is a good practice to provide some background or context about each interview, as well as participant details, and record it at the header of the transcription file. Other types of materials such as audio-visual files, might allow you to store background information in the file metadata. This could be for instance, the name of the creator, or image location details. If this is not possible, contextual documentation can be kept as additional fields in your data listing. So as you can see, depending on your project, there are different things that need to get documented, in order to keep your data understandable and reusable, and make your results verifiable. For more information, examples and templates, you can visit our website.