Data Ab Initio

Every once and a while, I run across a topic and think, “surely I’ve covered this one my blog?” The most recent of these topics, which came up when I wrote December’s Exit Strategy post, concerns file formats.

File formats and exit strategies go hand-in-hand, as you don’t want to be stuck with inaccessible data in an unreadable file format. Proprietary, out-of-date, little used formats lower the chance that you’ll be able to access your data when you need it. Ask anyone with 20-year old files if they can use those files and you’ll likely see why file formats matter.

When it comes to choosing file formats that last, file types actually exist a spectrum. The best formats are open, well-documented, and in wide use. Clear examples are .TXT instead of .DOCX, or .CSV instead of .XLSX or .SAS. In the middle of the spectrum we find something like .PDF, which is an Adobe file format but in such wide use that it will be usable for many years. Also note that the spectrum of preferred file formats shifts over time (hello Lotus Notes and WordPerfect!).

Since there is no one right answer for any data type, the key thing for picking a good file format is to ask yourself if your content is currently in a file format that is uncommon or can only be opened by a specific software program. If the answer is yes, now is the time to make a backup copy of that data in a more open format. Even if you lose some formatting in the process, it’s better to have some data in an open format than having no data because it’s locked in an unreadable format. By making a copy, you also don’t have to lose the performance of the original file format while gaining the sustainability of the new; you can, of course, wholly switch to a better format if that is feasible. It’s also worth reviewing old data for formats that are no longer popular or supported.

Good data management is the sum of a number of small practices and picking good files formats is a piece of this puzzle. The more you are aware of how closed your current file formats are, the better you can plan for making that data usable into the future.

One phrase that’s bound to come up at every data management conference is “carrot versus stick” vis-a-vis incentivizing researchers to manage their data better. Carrots are rewards for good practices and sticks are requirements and their consequences relating to data management. There is inevitably discussion over which method is more effective for implementing data management.

Another phrase that I often hear in similar settings is “eating our own dog food” or “drinking our own champagne”. This is another way of saying “practice what you preach”, in that data experts should apply their advice to their own files.

These phrases are used so often that I’ve decided that they need to be combined as “eating our own carrot sticks”. It’s at least more appetizing than some of the other “eating our own…” options and a bit of snarkiness provides relief from predictability.

But to say something serious in this blog post, all of these phrases emphasize the importance of *doing* data management. It’s not enough to have the knowledge or to be given the incentive. It is only in the act of actually managing the data that we get value.

So I challenge you, whether you are a data management novice or an expert, to find one new data management practice to implement this month. Because a carrot stick a day keeps the data disaster away*.

* Okay, now I’m taking it too far, I know. I can’t help myself.

Twice in the month of January, I had to find files from an old project. With resignation, I delved into old folders only to find that, wow, there’s a “FinalDocuments” subfolder with everything I need all laid out for me in a well documented way. Both of these times I was so, so thankful that my past self had the forethought to organize things for the future.

Past Kristin strikes again! I looked up files from a 2015 conference I planned and, hey look, a "FinalMaterials" folder! Nicely documented outputs from the conference in one subfolder and conference planning documents/write up (plus README.txt) in the other. #pastKristinRocks

— Kristin Briney (@KristinBriney) January 25, 2018

Looking back through the blog archives, I realize that I wrote a “Wrapping Up a Project” post 4 years ago! (Related: how the heck is this blog 5 years old?!) I still stand beside the advice I gave in that post that researchers should: back up their notes, convert file formats, use README.txt files, and keep everything together. These are generally useful strategies for all of your files that make it likely for you to still have everything in 5-10 years. However, most often when you go back to your old files you are looking for something specific.

This is why, in this post, I’m recommending that you add a step during project wrap up to select key information to copy to a “FINAL” folder (or some obvious variant of that name). Example documents include: a copy of the final publication, the raw dataset and the analyzed dataset, finalized scripts, key protocols, and JPEG files for figures. Basically you should identify the information that you will mostly likely need to refer to later and place all the final versions together in one folder. And then write a README.txt file to describe the contents of that folder.

Without this added step, you will still have your files and can open them, but you’ll likely waste a lot of time looking for exactly what you need. And even with this step, there will be times when you’ll have to dig through all of the project files to find something specific. But 90% of the time, you will save time by placing these key documents in an obvious place.

Trust me, your future self with thank you for taking 20 minutes, while you still understand the files and their organization, to set aside the important stuff for later.