"Hey, could you send me that presentation that you gave at the symposium last year? Thx" A harmless request like this may turn into a nightmare of digging through the labyrinthine architecture of the computer's folder structure. Perhaps, the presentation is in the default "Presentations" folder or, maybe, it ended up in a Dropbox folder to share it with someone? Or, it's in a folder with the paper draft from that time...? It gets even worse when looking for some documentation that has a less strong memory trace like the final draft of an ethics application or a spreadsheet with some background data. Fortunately, there is an easy way of organising your digital life in a more logical, robust, and accessible form that solves many of these problems. It will introduce a new sense of tranquillity and ease to your day-to-day digital life. Let's spark some joy!
The start of good digital organisation is a consistent folder structure. Many operating systems force a basic structure on us. The operating systems often divide documents by type, e.g. Documents, Photos, Presentations, and so on. This means that your paper draft and your presentation on the same topic live in entirely different places.
Fortunately, there is a solution from the world of software engineering. When writing a python package, developers typically use a template that makes it easier to organise all the subscripts that make up the package. There are even programmes that create the basic folder structure automatically, e.g. cookiecutter. So, how can we translate this for behavioural science? In my experience, most work is structured around projects. Therefore, it makes the most sense to organise the folder structure around the project. For the project, you will probably conduct experiments to collect data and then analyse these data. Finally, you will disseminate your results through papers, blog posts, and presentations. Accordingly, the basic folder structure for any behavioural science project should be:
The folder will contain all files associated with a project - no more digging around for that script or that presentation in some obscure folder that the operating system deemed sensible. You should use an appropriate name that makes it easy for you to identify the project. In later sections, we will discuss the addition of ReadMe files and data dictionaries to add more context. But for now, we’ll just stick with a short descriptive name. Also, note that I used an underscore instead of white space for the name. That’s because white spaces can make life more difficult when you’re working with scripts. So, I recommend that you’re getting into the habit of not using white spaces when naming folders.
In the next level of the hierarchy, I created folder for each major aspect of a data science project: data acquisition, data, analysis, and dissemination. The acquisition folder should contain all the scripts and materials for experimental tasks, all the recruitment letters and information sheets, interview protocols, copies of the specific psychological tests and so forth. So, everything related to any aspect of data acquisition. The data folder should contain all the raw data, including the output from your computerised scripts, scanned pen-and-paper forms etc. The folder should also contain all intermediate data that your analysis scripts produce. The analysis folder will contain scripts for data analysis. That includes scripts to clean and organise the raw data, scripts for data exploration and visualisation, and scripts for the final statistical analysis. The dissemination folder will contain paper manuscripts, abstracts for conference presentations, PowerPoint or similar files for talks, and any other material that relates to the dissemination (blog posts, video blurbs, animations, interpretative dance choreography, …).
As you can see, this basic structure captures what most behavioural science projects require. The template makes it easy to organise most files, but some of the basic folders may be superfluous for some projects. For instance, you may be carrying out analyses on some specific aspects within a larger-scale study. In that case, it makes no sense to include the information sheets and other recruitment materials in every subproject. However, I recommend to still include the ‘acquisition’ folder. You can either add a text file that indicates where the corresponding documentation can be found or add links to the location in your local file system, network of the institution, or web address.
As you may have guessed, the basic folder structure is not enough to keep everything organised. We will need to create subfolders to manage files within the broader domains of the basic folders. These subfolders will differ depending on the specific needs of the project. For instance, some projects may need subfolders within the acquisition folder for different behavioural tasks or different stimulus sets. Or, you may want to create subfolders within the data subfolder for different kinds of data, e.g. reaction time measurements and questionnaire results. Here is an example of what the lower-level hierarchy could look like:
I would recommend keeping the depth of the hierarchy within reason. While a four-level hierarchy is sensible, a structure with more sub-levels quickly becomes tedious. If you think that you absolutely need more sub-levels, consider adding a higher-level folder or splitting the project.
Please spend a bit of time to think about the data structure for the specific project before starting it. This way, you can work with the appropriate structure from the start and, thereby, add things to the right location. This becomes crucial when collaborating with colleagues on a project. It should be immediately obvious for all collaborators where items can be found. This will also make it a lot easier to archive everything when the project is done.
In this section, we have learned how to use a basic folder structure template to organise behavioural science projects. This folder structure will make it much easier for ourselves and our colleagues to find files and documentation that belong to a project. In the next section, we will further improve our data organisation by adding ReadMe files and data dictionaries.