Downstream analyses: Common approaches and use cases

[wendtke]

Right now as I'm trying to figure out the best approach, I need to know some common characteristics in downstream analyses. Some detailed use cases will help. For example, what are some frequently used statistical models? Are modeling usually done for each and every subject, or across some kind of summation of a group?

Originally posted by @iqis in #58 (comment)

[wendtke]

@iqis

Examples:

• repeated measures ANOVA (there are limitations with this approach; see Vasey & Thayer, 1987)
• latent growth curve modeling
• time series analyses such as multilevel autoregressive modeling
• general multilevel modeling approaches

Approaches differ based on individual vs. dyadic (two people) data and number of data streams (e.g., combining HRV/RSA and EDA for index of autonomic coordination; see here for example). I focus on dyadic physiological processes. See here and here.

@MalloryJfeldman and I can add more examples and papers to the Google Drive and will write detailed use cases for visualization #34 and analysis #60.

[iqis]

Thanks, I'm starting to finally get a grasp on what on earth you guys are really doing. I took a glance at the papers and found Helm very interesting. I also wonder how much does what is called psychophysiology are generalizeable to ... well.. regular physiology?

[wendtke]

Yes, definitely. I think of psyphophysiology as the nesting of physiological processes within a psychological framework of analysis; we try to make inferences about individuals' and dyads' psychological functioning based on patterns of physiological responses.

[MalloryJfeldman]

A lot of the physio we do utilizes multilevel modeling. More recently we've been interested in dynamical systems modeling, structural equations, and unsupervised clustering techniques though I only have experience with the latter. I'll keep thinking on this!

[wendtke]

Is modeling usually done for each and every subject, or across some kind of summation of a group?

Originally posted by @iqis in #58 (comment)

Both/either. Depends on the analytic approach.