@andrewjdbell.bsky.social
Senior Lecturer in Quantitative Social Sciences, University of Sheffield. Multilevel models, age period cohort, Intersectionality, etc https://linktr.ee/andrewjdbell
Now officially out with nice formatting and all 🥳 "Thinking clearly about age, period, and cohort effects" -- a gentle introduction to the age-period-cohort problem and how to "solve" it through various types of assumptions.
journals.sagepub.com/doi/10.1177/...
A promotional graphic for the Research Methods Rendezvous, which takes place online on 10 September and 29 October 2025.
NCRM has opened applications to participate in the Research Methods Rendezvous!
This free event will explore the process of turning early-stage ideas into #research projects.
#RMR2025 takes place online on 10 September and 29 October 2025.
Apply: www.ncrm.ac.uk/training/RMR...
Thanks to everybody who chimed in!
I arrived at the conclusion that (1) there's a lot of interesting stuff about interactions and (2) the figure I was looking for does not exist.
So, I made it myself! Here's a simple illustration of how to control for confounding in interactions:>
Our research on the effects that PIP changes could have on local economies, written by authors @nataliecbennett.bsky.social @profbambra.bsky.social & @lukemunford.bsky.social, was featured on the @itvpeston.bsky.social programme last night - you can watch the episode on catch up
22.05.2025 08:46 — 👍 5 🔁 8 💬 0 📌 0Another q for the stats people!
People worry about collinearity (cf blog post below).
Consider a scenario in which the collinear predictors are just controls to account for confounding.
Including both of them doesn't impair the precision with which the effect of interest is estimated, does it?
Multilevel Models: Practical Applications
Curious about multilevel modelling but not sure where to start?
Learn to recognise, build & interpret multilevel models using MLwiN or R, through real-world examples & hands-on practice.
Find out more: bit.ly/3Yj5I4f
#ESS2025
Still time to apply for this RA post- developing and applying the intersectional MAIHDA approach with @clarerevans.bsky.social and others. If you're a quantitative social scientist with an interest in quantitative methods and methods development, please apply! DMs open www.jobs.ac.uk/job/DMR190/r...
29.04.2025 08:50 — 👍 2 🔁 1 💬 0 📌 0
Text reads: "An introduction to multilevel modelling for intersectionality research: the MAIHDA Approach. 19th June 2025. Instructors: Prof George Leckie and Dr Andrew Bell. There are logos for Centre for multilevel modelling, NCRM, University of Sheffield, and UKRI
Come join George and I in Bristol, to learn about intersectional MAIHDA! bristol.ac.uk/cmm/software... @ncrm.ac.uk
04.03.2025 14:50 — 👍 3 🔁 3 💬 0 📌 0
It’s my first project too! In fact, it’s also my first in epidemiology since I come from philosophy. :) I find stratified graphs based on social categories (like Hernández-Yumar, 2018) really interesting.
22.11.2024 13:09 — 👍 3 🔁 1 💬 1 📌 0A little story about this article & how teachers make a difference:
In 2013, I was a student in a social epidemiology class taught by the esteemed David Williams. He observed, with some frustration while we looked at a series of graphs, that we often lump people into the “Hispanic” category — 1/
George Leckie and I have created a new short MAIHDA tutorial, with videos and practical exercises:
www.ncrm.ac.uk/resources/on...
It's based on our tutorial paper in SSM Pop Health, but shorter and more interactive.
Hope it's helpful! @ncrm.ac.uk
Starting out using #intersectional MAIHDA analyses, so need to dig into this great tutorial paper by @clarerevans.bsky.social , Leckie, Subramanian, @andrewjdbell.bsky.social Juan Merlo 👏
www.sciencedirect.com/science/arti...
New paper published in Social Science and Medicine 'An analysis of intersectional disparities in alcohol consumption in the US'. Led by @sophiebright.bsky.social, this study identifies several understudied groups who may have higher alcohol consumption than traditional methods would suggest.
19.11.2024 11:23 — 👍 5 🔁 8 💬 0 📌 0
Abstract Psychological researchers are interested in how things change over time and routinely make claims about, for example, age effects (e.g., personality changes with age) or cohort effects (e.g., differences in intelligence between cohorts). The age-period-cohort identification problem means that these claims are not possible based on the data alone: Any possible temporal pattern can be explained by an infinite number of combinations of age, period, and cohort effects. This concern holds regardless of the study design—it also applies to longitudinal designs covering multiple cohorts—and regardless of the number of observations available—it also applies if we observe the whole population. Researchers rely on statistical models that impose assumptions to pick one specific combination of effects. But these assumptions are often opaque and researchers may be unaware of them, resulting in a lack of scrutiny. Here,...
New preprint! osf.io/preprints/ps...
The age-period-cohort problem is something that many researchers are vaguely aware of. There have been very cool advances in how to reason about it which don't seem to be well-known in psych. So, I've written a primer!
Rereading @andrewjdbell.bsky.social's 2020 paper on APC analysis & this is such a great example of how plots imply certain interpretations of the data.
Same underlying data but depending on how you connect the lines, the implied age effect looks completely different.
