Clinical psychology researcher | adjunct professor | applied statistics geek | so called #RStats influencer
Sep 19, 2022 • 5 tweets • 1 min read
If you wanted to use *age* as a control variable in a regression model and its distribution looked like this, what would you do with it?
a) mean center
b) standardize
c) take the log
d) other
For those commenting, please extend me the benefit of the doubt that I already understand regression models don't make distributional assumptions about the predictor variables.
May 31, 2019 • 4 tweets • 2 min read
Shopping visuals:
Got me a 2-DV (QoLS and WHODAS) longitudinal multilevel model. There are random epsilons. This leaves me with 6 variances and 15 covariances. I’m not a fan of correlation tables. Thoughts on this visual?
For context,
u_0 = random intercepts,
u_1 = random slopes, and
u_2 = random epsilons.
Mar 29, 2019 • 6 tweets • 3 min read
Psychometric twitter: What are your favorite articles addressing how to approach multidimensional constructs? I’m interested in statistical methods and philosophy of science.
@JkayFlake@EikoFried@MelissaGWolf@edgarmerkle@oscar_olvera100@hardsci
While pondering multidimensional constructs, I recalled part of my clinical training. Take IQ, which is often measured with the WAIS. You give 10+ subtests, the results of which are summarized by 4 indices (e.g., verbal comprehension) and the full scale IQ...
Here are my thoughts:
2/n It probably goes without saying, but just in case you missed it, make sure you check out McElreath’s lectures on his text, too youtube.com/channel/UCNJK6…. He has three semester’s worth and they’re overall really great.