The applied example illustrates the approaches using psychological distress as a symptom count, with longstanding illness as the outcome. R code is included. Put together with my brilliant supervisors, @rjsilverwood.bsky.social, @georgeploubidis.bsky.social, and Jess Bone.
Our preprint is now out! 📄 We looked at whether five flexible IPTW / propensity score weighting approaches can be used for count exposures, and whether they can be combined with multiple imputation. We also illustrate the methods using BCS70 data. Link: doi.org/10.48550/arXiv.2603.23726
Such a useful update for day-to-day data cleaning. Very excited to finally have look up tables in the tidyverse!
If you use survey data collected through mixed modes (f2f, telephone, online), check out this @royalstatsoc.bsky.social event with @rjsilverwood.bsky.social @georgiatomova.bsky.social and others.
What percentage of the health research funding pot do you think goes on methods and meta science?
Around 2% (if you add together all mention of research design and methodologies across all areas from research.hscni.net/sites/defaul...)
Just 2% spent on the part that underpins EVERYTHING we do.
We are looking for some community feedback on 3 new dplyr functions!
- replace_when()
- recode_values()
- replace_values()
New tools for recoding (think, lookup tables!) and replacing (think, replace `-99` with `NA` in `col`) in the tidyverse - I'm pretty excited about these!
📢 Job opportunity! 📢
I am recruiting a 2-year postdoc to join me at @clscohorts.bsky.social to work on a SUPER exciting project on the quantitative study of intersectional inequalities in youth mental health across time and place!
See www.ucl.ac.uk/work-at-ucl/... for more info and how to apply!
Excited (and a little nervous!) to be giving a talk on 30 June at the Turing’s CIIG seminar series. If you're interested in IPTW/propensity score methods, or in seeing some cat photos 🐾 (with the former obviously being more likely!), everyone’s welcome to join.
Midwit meme left panel: causal claims based on often untestable assumptions middle panel: nooo you cannot make claims like that our study merely shows that X predicts Y above and beyond Z future experimental studies are needed to what even is a cause according to hume right panel: causal claims based on often untestable assumptions
BCS70 at age 51 webinar data and photo of man and woman walking in the hills
✨ Have you booked yet? Join us for our free webinar NEXT WEEK introducing 1970 British Cohort Study data at age 51. Discover more about the experiences of generation X in midlife: cls.ucl.ac.uk/events/bcs70... #BSC70 #ageing
Just came across this gem on survey designs and survey weights with R. Free online version available.
1/2 Thrilled my 1st PhD study on how adolescent social connections influence BMI trajectories has been published🎉
Huge thanks to my coauthors @yvonne-kelly.bsky.social @tattanbirch.bsky.social @martindanka.bsky.social, Liam Wright, Daisy Fancourt, Ellie Iob 🤝
www.sciencedirect.com/science/arti...
Our JSS article is out!
And now I get to focus on {marginaleffects} 1.0.0. Stay tuned.
www.jstatsoft.org/article/view...
Excited to see a new causally-informed epi textbook 🎉
Thanks, Peter - much appreciated!
Also had the chance to give a peer workshop on causal inference. It was rewarding to see how much clarity the frameworks bring to diverse areas and to learn from the discussions. The takeaway is that causal inference is less intimidating when seen as a way asking clear questions. #causalinference
It was wonderful to finally meet one of my causal inference heroes @pwgtennant.bsky.social in person at our Soc-B PhD workshop. A brilliant and inspiring talk.