Thread on silly late night musings regarding chaos and ecological momentary assessment:
I was watching these videos and playing around with data. 1/
I was watching these videos and playing around with data. 1/
There's this idea that the extent of chaos can indicate system failure, or destabilisation such as shown by this awesome work by @OlthofMerlijn 2/ psych-networks.com/how-to-study-e…
I wondered, what kind of OBJECTIVE data I'd have that could show periodicity and chaos in time (like in fig) and realised I could play around with the inter-response intervals from our study, where office workers were beeped 5/day to answer motivation surveys
3/
3/
Because we have nights and weekends where participants are not answering EMA, we'd expect to see a period-3 attractor in inter-response intervals: people would answer ca. every 2 hours during daytime, have ca. 16-hour breaks between days, and ca. 64-hour breaks during weekends.
So I made some plots, which is great now that I finally have learned to speak #purrr and can do massive iteration jobs for every participant at once. Surprise surprise: some people were more deterministic (compliant?), others less so. /5
(I've been following the same protocol for a while now, tracking my motivation to do things in my life; a pretty nice period-2 orbit as there are no weekend breaks) /6
Not sure if I really learned much here, just wanted to put it out there in case someone else gets ideas. My preliminary thoughts were to try connect the deviance from optimal periodicity to warning signs of study dropout, or as a quality check for the analysis /7
... That is, if the deviance metric (however it's derived) correlates with any of your outcomes, it might be a problem for the generalisability of your findings. Or maybe it could be connected to demographics. Maybe there are better ways of doing this already? /8
Anyway, it was a fun thing to play around with. Let me know if this sparks any thoughts or if you want to share your EMA data quality assessment toolbox!
[Thread might be of mild interest to @EikoFried, @FredHasselman, @RichieLenne, @DanPowell83)
[Thread might be of mild interest to @EikoFried, @FredHasselman, @RichieLenne, @DanPowell83)
