We're a little crazy, about science!

Analysis fails

I’m an idiot. So for the past few weeks I’ve been working with our super fun new dataset! Like a big kid, I did the statistics, made the checks to ensure that the assumptions held, then got my result. Everything was going well until I realized there was a typo (here) in my code. Well it turns out that wasn’t my only problem, but everything should be fixed now… should.

A little over a week ago I realized the analysis I had performed was flawed. There was a lovely typo in my code which meant I ran a different type of analysis on the data than I had thought, no worries, I caught it, spent the time to update the figures, life was looking good. Figure making isn’t an easy thing, so I spent a few hours making everything perfect, twice. Then came the realization that I had a whole other problem on my hands, my assumptions about the data were wrong.

It’s not that I didn’t check, I did check. One of the assumptions you make when you run a parametric test (here) is that the data are normal. Well being the exhausted and overworked grad student I have been the past few months I ran the test on the averaged data, not the raw data, big difference. While the averaged data were normal the raw data were not and that’s the data I ran the statistics on (because you can’t really run statistics on averaged data, at least not the way we are doing things).

This meant that once again I was back to square one with the data and I needed to rerun the analysis this time using a non-parametric test. Long story short (for once) I ended up with roughly 1500 lines of code written to redo the analysis which took a few days of hard brain labor. Then I had to regenerate the figures, make everything look pretty, and go. Which also took some time, but I’ve finished.

It’s been an exhausting week with all this work I’ve had on my plate because I’m an idiot and didn’t double check the basics. It’s cost me roughly the whole week now, but it’s done. I’m just frustrated because I wanted to do the analysis right the first time, or hell at least the second time, but I guess it took three tries. Thankfully it’s correct now, or at least as correct as I can get it considering I’ve only taken a few statistics class, it’s not my degree focus. It was basic(ish) stuff as far as statistics are considered, but it was just time intensive and there was a lot of figuring out how to run the actual anyalysis.

Like anything involving programming the answer isn’t as straightforward as it seems going in. I had multiple tests to run and I had no real way to do it, but I figured it out and got the data in the form I needed. I also had a few mistakes along the way that I caught, but I really, REALLY don’t want to get into that. Now I need to update the paper, add in the new figures, and figure out what the heck I’m doing.

I was super angry with myself over this mistake (well mistakes), not going to lie, but now that I have it all worked out I feel slightly better about the outcome. Now I can get back to finishing the thing and get this off my plate (hopefully) my Co-PI is being super hands on with writing this paper, which has sped up the process significantly, but I’m still behind schedule. I’m hoping for a break soon, but if we look back you’ll see I’ve been hoping for one for about six months or so now with no end in sight.

That being said I need to get to work, so short(er) update, but at least it’s the daily update. Sometimes a PhD feels like voluntary torture. I know why I signed up for this, I know why I’m doing it, but fuck it hurts so much.

But enough about us, what about you?

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.