New blog post! If you're a substantive researcher, you mainly do statistics to answer substantive questions. So let's stop putting the statistical model first. www.the100.ci/2024/08/27/l...
Summer in Berlin – the perfect time and place to explore the city, take a walk in the Görli, go skinny dipping in the Spree, attend an overcrowded, overheated conference symposium on cross-lagged pane...
Also a good reason to not leave statistics to the statisticians (less of an issue in psych, I believe, but I am looking at medicine and epi here...)
While I am strongly aligned with articulating questions before seeking answers, I personally also really believe in “tinker with new models as to learn which estimands they can get you, which you didn’t know exist, and now you can answer substantive questions you never knew were feasible”
I really appreciate statistical/methodological blogs of this nature. Thanks Julia. I realize you don't get much "credit" in terms of your CV for writing blogs (vs. journal articles), but please know that we are reading! And it's all giving us much to think about.
I never understood why all those named-SEM models never had a "your control/confounding variables go here"... But now I understand 🤦♂️
Looking forward to reading this!
Yes! Bookmarked for future reference. Sometimes I just sit and think about how hilariously crazy our world is that we have to ask anyone to please figure out what they want to know first before deciding on their methods. We are so used to this behaviour that we don't appreciate how weird it is.
Finally gave this a proper read. Great work Julia, and not 1% to disagree with! I followed through some citations and was pleased to find that even the EU has set some guidelines in this area: www.ema.europa.eu/en/documents... 🇪🇺
One of my concerns with latent growth curve models (which are popular in lifecourse epidemiology) is I have no idea what estimand is being targetted. Also, the summary latent variables have wide temporal footprints, which (in my mind) makes them susceptable to intermediate confounding...
As the trope says, biostatistics consultations largely consist of the statistician repeatedly asking the clinician "yes, but what is the research question?"