Coding is time consuming. I didn't really consciously think about this till my advisor asked me about the roadblocks to my progress in my dissertation. I thought back and it occurred to me that I was
- learning a somewhat new language (R)
- trying to code in it using concepts from statistics that I already have
- trying to generalize code so that it could be less repetitive and automate chores and processing
- trying to splice other languages such as latex and perl to interact with R code
- learning THOSE other languages
- trying to prevent errors in their interaction
- trying to write functions that would serve a purpose similar to "macros" on SAS but yet be understandable to me at a somewhat-novice level.
- debugging to see where it all goes wrong when errors turn up
- differentiating what is my code's fault, what can be changed in R preferences, and what is inherently the fault of the R build on the particular linux distro I am using.
Why am I doing this when I have a data manager for the data I am dealing with? Because it's insane to go whining to the manager the minute you want some data processing done. He has own Ph.D. to complete and has limited time and innumerable demands. So there is really no option but to power through all of this on my own. Besides, I think population sciences are *really* about statistics and data processing even though population scientists who don't do all this dirty work may persuade themselves that it is just about interpretation. It is as much about quality control in the data and seemingly trivial logical checks as it is about analyzing super-clean data in the end and figuring out what it could mean in the bigger picture of million other studies. I think it's important to acquire this skill so I can use it in my future. Not everyone has a data manager, a software coding bloke and a statistician to always help them.
The process itself is exhausting because it's like learning a completely new language without having ANY experience in how the semantics work. Apparently some people have observed this exhaustion and set up sites like "Software Carpentry":
It sets up a formalized framework within which you could learn these skills. I think courses like this should be mandatory to population sciences coursework. At some point, we need to realize that population sciences and bioinformatics are very similar in their scope and application. This is more true than ever as we progress rapidly towards HUGE datasets, complex data structures, multidimensional variables and the necessity of channeling all of this into interpretations simple enough to make sense to health-related and population policy-making.
You could always mix in any essential oil to the mix.
What else is in your deodorant?
I like tea tree oil in my deoderant but I am sure I am a much stinkier individual being an active, greasy Italian fur ball.