Week 7: Code, Correlations, and Conundrums

Welcome to Week 7! Feel free to call me an aspiring magician: this week I’ve been all about trying to make the invisible into the visible.

As I predicted last week, data collection is still rolling along as I try to squeeze in those final few participants, but I’ve meanwhile transitioned into the data analysis of my project.

The Calmness Conundrum

Because the CNV is playing hard to get, I’m having to take a long look at my intended evaluation methods.

While testing, I ran into a unique, frustrating, hurdle: my participants aren’t actually that anxious.

When looking at the post-experiment surveys, the reported anxiety levels are trending much lower than I anticipated. Why? Well, it turns out that sitting in a comfy chair in a quiet, heavily controlled room in my house (shoutout to my Week 3 setup!) doesn’t quite trigger the same visceral “pit in your stomach” dread as waiting for a real score or result. 

Despite my best efforts to script a stressful “evaluative threat,” my simulation environment is simply too low-stakes. My participants subconsciously know that whatever happens in this 10-minute session won’t alter their future.

Searching for Signals

Because the anxiety levels are trending lower than expected, it’s having a domino effect on the rest of my data.

Right now, I’m struggling to detect the CNV signals in my data. If you remember, the CNV is that tiny electrical build-up in the brain right before an expected event. Because my participants are relatively relaxed, that anticipatory buildup might be much smaller than anticipated, or it’s simply hiding behind the regular, chaotic background noise of the brain.

Adjusting the Approach

Because of these roadblocks, I’ll likely need to adjust my intended evaluation methods moving forward. I’m trying to look at a relationship: the correlation between the CNV signal (once I can clearly isolate it!), the participant’s actual performance on the cognitive task, and their self-reported anxiety.

The good news is that the preliminary results are actually pointing towards a slight correlation between these factors! It’s a tiny thread, but it’s there.

To pull on that thread and see if it holds up to scientific scrutiny, I need a little more data. A slight correlation in a small sample size could just be a fluke, so I’m motivated to get those last few stragglers into the testing chair this upcoming week to see (read “HOPE”) if the trend solidifies.

Science is messy, and this week was a brilliant reminder of that. You make a hypothesis, you design the best experiment you can, and then you have to wait and see the reality.

Next week is going to be heavy on tweaking, and hopefully, a few clarifying moments. We are nearing the final stretch!

Thanks for following along,

– Josh Peter