Dr Lois Player
banner
loisplayer.bsky.social
Dr Lois Player
@loisplayer.bsky.social
Behavioural Scientist | Climate Policy & Behaviour Change 🌏 | Novel Computational Methods - NLP & LLMs 👩‍💻 | Research Consultant | Currently Contracting at European Climate Foundation ✨

loisplayer.com
🧵 4/4

To avoid falling support for climate policies, and a shift to parties who won't support them - we unquestionably need to rebuild trust and build climate policies *with* the public 🙌
July 30, 2025 at 1:56 PM
🧵 3/4

The review even shows that when governments support workers, involve communities, and communicate clearly, the backlash fades and support grows.
July 30, 2025 at 1:56 PM
🧵 2/4

When people feel targeted by or left out of policies, they are more likely to back these politicians - who promise to roll them back.

What’s driving this isn’t climate denial, but *distrust* of the government to do what's best for people.
July 30, 2025 at 1:56 PM
Huge thanks to my co-authors - Ryan Hughes, Kaloyan Mitev, @lwhitmarsh.bsky.social, @christinademski.bsky.social, Nick Nash, Tris Papakonstantinou, & Mark Wilson — a dream team of qual, quant, & ML expertise.

Thanks also to AAPS CDT- without whom Ryan and I would never have begun this project! 🚀
April 8, 2025 at 9:59 AM
The code is totally open-access - and we are building a user-friendly interface now! 👩🏼‍💻

Want to participate in an early pilot or receive updates?

Sign up: tinyurl.com/DECOTAform
Contact Form
Interested in using DECOTA? DECOTA (DEep COmputational Text Analyser) is a scientifically validated tool designed to help you rapidly analyse survey response data through an automated process based on...
tinyurl.com
April 8, 2025 at 9:59 AM
Even as a preprint, DECOTA drew significant interest from policymakers for its potential to:

• Provide rapid insights in time-sensitive situations
• Scale analysis to huge datasets that could not be analysed manually
• Support more interpretative qual work 💭
April 8, 2025 at 9:59 AM
We tested DECOTA on 4 human-coded datasets. It identified:

• 92% of all human-coded sub-themes
• 90% of main themes

…in just 10 minutes and $0.82 per dataset 🚀
April 8, 2025 at 9:59 AM
DECOTA uses fine-tuned LLMs to analyse open-ended text data - with no human input needed.

It’s a transparent 6 step process, that outputs:

- Themes & sub-themes in the data
- Their prevalence
- How they vary by demographic group
- Example quotes
April 8, 2025 at 9:59 AM
The mad stampeed when they announce the platform 😂
February 20, 2025 at 10:01 PM
Thanks Jake! Great to see these trends coming to the fore 👑
February 18, 2025 at 5:33 PM
Thank you so much @masharemskar.bsky.social 🥹 what a journey it was 🤍
January 23, 2025 at 8:34 PM
Thanks so much Jake! ☺️
January 23, 2025 at 8:34 PM
Thank you so much Emma!
January 23, 2025 at 11:55 AM
A huge thank you to my amazing supervisory team
@lwhitmarsh.bsky.social & @christinademski.bsky.social for your unfaltering support always - and many other incredible people along the way 🌟
January 23, 2025 at 11:29 AM