The last participants

Well I had other topics in mind for the day, but I’m so excited that I wanted to make a little post about it. I’ve finally confirmed the last three participants for my dissertation project. I have 7 out of 10 datasets collected and needed three more for this first aim. For the past few days I was nervous that I wasn’t going to get it done this week. Sometimes that happens, things don’t work out how you plan, but that isn’t the case! I’ve confirmed all three participants for my project.

It’s the last thing I have to do to graduate, my dissertation. It’s been a long, long four years (going on five now…), but at the beginning of the year I defended my dissertation proposal and I’m finally collecting my data after a long, drawn out wait for the equipment I needed. Things have been going quickly too. I’ve managed to get two or three datasets collected every week since the equipment has arrived and I quickly ran out of candidates I could ask to participate.

After asking around I had several maybe’s a few flat out no’s and no yes’s. As time was running out to hit my goal, it was harder and harder to ask someone when it’s so last minute. However, today I managed to confirm with everyone that they would be available and they would do it, so unless something goes wrong or the weather doesn’t play nice I now know that this weekend I’ll have all ten datasets needed to start my analysis.

Okay, I could start doing data analysis at any time. You don’t need to have all the data collected to process it, but it does make batch processing easier, so having all the data collected isn’t a bad thing. I can go through each dataset one by one and get it through each step (or automate most of the processes, but I’m not a fan of that). It’s an assembly line method, you do step one to all of the datasets, then step two, and so on.

This isn’t the only way to process data, but it’s my preferred way. I like it because just like data collection, you get a feel for the data as you go, so you are less likely to make mistakes and you have a fresh recollection about problems you’ve run into with other datasets. Most of the stuff can be automated, but a lot of it needs to be done by trial and error. Making sure the data looks good before moving to the next step, so automating it saves time, but leads to data that aren’t as “clean” or artifact free. Which in turn affects the analysis and any conclusions you want to draw.

Basically I prefer doing 98% of the data processing manually and it’s time consuming, but I firmly believe you end up with a better result. There’s no substitution for good data to begin with, but there is always pre-processing steps you need to do (cleaning) no matter how good your data look when collecting it.

Which of course means that the next several weeks will be a ton of working on processing the data and trying to make sense of it all. It’s five dimensional data, so there are headaches associated with presenting the data in a way that makes sense to people, but I’m going to try to do my best.

But this is all getting way ahead of myself and the intended point of the post, which was simply that I now have all my participants confirmed and that’s a ton of stress off my shoulders. There’s still a lot of work to be done, but after this weekend, a good portion of it will be finished! There’s still data to be collected, but I’m hoping if I need to do aim 2, I can shrink the sample size. We’ll see and again that’s a future problem and not the point of the post exactly.

I guess I can’t help thinking ahead now that I have this big step nearly finished! Plus with DARPA right around the corner, I can’t afford to have any more delays.