Recently, in writing a letter nominating someone for a scientific award, a colleague and I wrote something like “It’s difficult to convey just how strong a candidate Dr. So-and-so is.” Which is not a letter-writing approach I would recommend. There are in fact words that can convey just how strong a candidate Dr. So-and-so is, and you should write them. Which we did, of course–our letter didn’t actually rely on the bald assertion that Dr. So-and-so is too awesome for words. But still, that sentence could be considered a minor violation of the standard advice that I (or anyone) would give about how to write a reference letter.
Hence my question: is there scientific advice that you give to others, or teach to students, that you don’t follow yourself? Or only follow sometimes?
Here are some of mine:
- I don’t use reference management software, even though I advise my students to do so. My reason is sheer inertia. I even bought a copy of Endnote last year, but haven’t taught myself to use it yet. It turns out that I cannot force myself to use something just by spending money on it!* I am good at many bits of sciencing, but frankly I suck at forcing myself to learn new technical skills that I would be glad to have learned once I learned them. I don’t mean that as a humblebrag, it’s just the truth. We all have our weaknesses as well as our strengths, and that’s one of my weaknesses.
- Until recently, I didn’t use R Studio. Or even the R scripting window. I either used the command line**, or used R Commander to operate R with drop-down menus and push buttons, or else I wrote my code in Word and then copied and pasted into R.*** You’ll be glad to know that I just started using R Studio and now regret not using it years ago.
- I tell my students not to bullshit about the purported applications of their fundamental research when writing papers or grant proposals. But like most everyone, I do it myself.
- I blog, but I wouldn’t advise anyone else to. More precisely, I think it’s unlikely that my own good reasons for blogging will apply to most other people.
So, about what bits of science do you say to others “Do as I say, not as I do”?
*Because now I feel like I’d be committing the Concorde fallacy (aka sunk cost fallacy) if I spent time learning to use Endnote that I didn’t really feel like spending. Thus illustrating that awareness of logical fallacies doesn’t necessarily make you “rational” in any useful sense.
**An undergrad who worked in my lab and took intro biostats with me once saw me do this. Her eyes got really wide and she said “You taught us never to do that!”
**We now go to a live shot of every reader of this blog.
Dynamic Ecology 
Ditto reference management systems and bullshitting about application. Also backing up work and time management (gantt charts etc.) I’ll produce a gantt chart when I have time to do it…
I bet that bullshitting about applications of fundamental work is a common one.
It is in the UK. But then we are almost forced into it by the Research Councils who demand to know what the “Pathways to Impact” will be in advance. Impossible to get funding for basic science unless you show that there’s going to be some kind of application. Consequently my non-applied/conservation work is largely unfunded.
“Don’t forget to go on holidays”
“Have proper breaks from work; do not check work emails during weekends”
Might I suggest that there would seem to be a significant difference between “things I say that I’d really like me to follow too, but that (like reference-manager software) I haven’t managed to convince myself to listen to”, versus “things I say that are good general advice but that (like the ‘too awesome for words’ reference letter content) can be violated to good effect if one knows exactly what one is doing”?
In trying to figure out the boundaries of these concepts in some kind of cost/benefit/expectation space, it seems like there are additional distinct categories as well – for example, “things I say that are important for the vast majority of people to follow, but that (like speed limits) if only a few people violate, those people have a high expectation of significant benefit”.
And to answer the question:
Same same on reference management systems. I’ll claim that it’s because I haven’t found anything that does a good job with BibTex. That’s my story and I’m sticking to it.
“Document your code”.
“Write test cases first”.
“Never drive through deep water”.
“lock out equipment before servicing”.
Yes, there is a distinction between “I personally have good reason not to follow my own advice” and “I am a hypocrite”. My fourth bullet is an example of the former; my third bullet is an example of the latter.
Of course, humans are good at rationalizing, and so it can be hard for you to tell which category you’re in.
I don’t know if it does a good job, but Zotero claims to interface well with BibTex.
Re anonymous’s comment, zotero does indeed interface with bibtex well. I use Zotero to grab references from the web via the Chrome extension, then I either use it with Word to insert citations and generate a bibliography or I export the library as a bibtex file and use it with latex. It’s pretty great. And free!
JabRef is a really neat (native) BibTeX manager, I wholeheartedly recommend it (and of course free software too). Pretty easy and straightforward to get started with it… As a bonus, it can rely on Zotero transparently to grab things from the web with the JabFox extension (for Firefox only?), which is however not compatible yet with the latest Firefox.
Via Twitter, the likely thread-winning comment:
Thank you. :”>
Document your field/lab/analysis methodology at the time, because the time lag to publication means you might have forgotten what you did
Ok, that’s one that I tell others to do, *and* that I do myself.
I do use EndNote and have used a reference manger system since the 1980s. gannt charts – don’t use them, but tell my students to do so, similarly with R, I really must go on a course one day 🙂