Success! 2 of 10 dissertation experiments
Well today went better than yesterday and yesterday went pretty good to being with, so I’ll take it. All in all there were some things that I needed to iron out and a few things I still need to figure out, but it was successful and that’s all that matters. Since yesterday I already went over the rough outline of how it went and doing that again would be redundant even for me, I figure we can talk about how even the best laid plans can change once you start experimenting.
Okay so I’m FINALLY! in the middle of collecting data for my dissertation. There’s still a long way off between finishing the data collection and graduating, but I’m hopeful… mostly. It’s a lot, but progress is slowly being made. With any luck we’ll have five out of ten done by the end of next week. I’m still planning it, but I’m hopeful. It’s just a scheduling thing and I’m thankful that the people in the school lab are so flexible.
Anyway so let’s talk about why no matter how you plan you always end up needing to fine tune things as you go. In research plans are important. It’s how we estimate the time it will take to perform an experiment, it’s how we make sure we get all the data we want, and it’s how we ensure that if things go wrong, we prioritize the parts of the experiment that mean the most to us. Unfortunately, plans only go so far.
It’s like building a house, even with the best blueprints, things don’t always turn out the way you expect. On paper things look good, but when you start building you realize there are things that need to be adjusted, moved, removed, or changed altogether. It’s not a bad thing, at least you have a plan going in. With experiments it’s exactly the same issue. Things look good on paper, but once you’re physically doing the experiment you may find a better way to do something or maybe part of the experiment isn’t as feasible as you thought.
For me, a lot of planning (literal years worth) went into these experiments. I’ve had a lot of time to think about all the details, but things still change. So what did I find that went wrong with my experiments? Mainly the order of operations needed to be adjusted. My experiments, like a lot in our lab, are task based. So the order of the tasks (conditions) needed to be changed for the comfort of my participants, but it also sped things up on my end. Some of the setup had to be adjusted to account for several different things I didn’t think of when I was planning. This happens a lot because you can’t think of every possible thing that could go wrong.
After yesterday’s experiment ended, I gave a lot of thought on how to do it better. I went over everything I had done, wrote it all out and came up with some improvements. By adjusting a few things, not only did I save ~30 minutes in the actual experiment, I also managed to squeeze in a new condition I wanted to try. I’ve also decided at the end of the experiment to throw in a one off trial to test the limits of “super secret technique” and to find some other applications for it. There’s a wide range of ways my dissertation project could be used, so I’m cycling through 10 or so (one each participant) different ideas about the use cases.
Since we’re talking about the experiments, I guess I should make a small note (if only for my future reference since I do enjoy reading back on these years after the fact) about how today went. Setup went much faster than yesterday, which was to be expected, until we hit a snag with the impedance of the electrodes. For whatever reason the impedances just would not drop to save my life. After about an hour (literally!) of working the gel around the electrode — fun fact you literally swirl the blunt tipped needle to help it bridge the gap between the scalp to the sensor — I finally managed to get the impedances into an acceptable range. They were on the higher end of acceptable, which was kind of a bummer, but they were good enough that we didn’t have to keep trying.
Even with that the experiment ran for about three hours with the participant there (as always I was there several hours before and several hours post) and we always tell our participants to expect to be there for up to four hours, so this was good news for me. Setup takes forever and there’s not a whole lot you can do about it some days.
And that’s the update for the day. Maybe tomorrow we can talk about why ten participants may not be the end of this little project. Or maybe I’ll hold off on explaining that whole mess until I can figure it out. In any case, the big news is that we’re now ~20% done, with this bit of the project anway and if all goes well next week we’ll be 50%. Not too bad, if I do say so myself!
Ah and before I forget, I took a quick shot of the cap I was setting up after I got one of the bundles all attached. This is about as much as I can share until we publish something (assuming we find something to publish!)
“No plan survives contact with the real world” is a maxim I’ve seen to be true in a variety of contexts … even those that don’t involve working directly with people.
What could have been the cause of that impedance problem? Are there variances in hair or skin that can make it harder to get contact?
Anyway, I continue being happy to hear about these experiments and how well things are going. Do you think you’ll be left with any time to peek at the data, or does that have to wait until all ten Phase I collections are done?
LikeLiked by 1 person
August 13, 2022 at 8:44 pm
Very true, plans only go so far. Finding that out the hard way…
There’s a lot of reasons for it, it could be a thick hair issue. Typically people with long and thick hair will make it harder to get the impedances low. You can keep adding gel, but eventually you could bridge two sensors (cause gel to span two sensors) and that basically gives you the same measure at both, so we try to avoid that by just rotating the syringe around to push some of the hair away and let the gel settle, once it warms up its semi-fluid so impedances improve over time typically. Alternatively the sensors themselves can oxidize a bit and that will increase impedance so they may have just not been as clean as they could be.
I’m hoping to at least start basic processing before too long. I need to segment the data and even though we add flags and things to it, it’s better to do it while the memory of the experiment is fresh because you’ll forget what happened or what certain notes mean. Looking at the data afterwards removes almost all context of what was going on at that time other than what the very short notes say (almost all the flags we add to our recordings are just acronyms).
Once I have each individual dataset segmented and organized the analysis will go smoother, so I want to try to avoid doing all the segmenting work at once, even if I can’t find time to do the analysis before I absolutely need it done.
LikeLiked by 1 person
August 14, 2022 at 12:09 pm