Minutes of the Meeting [Week 32/2025] | Reading material for Week 33/2025

[Dammalapati Sai Krishna]

August 10th: 10AM - 11AM

Attendees: @kwatrak...@gmail.com @Karanam Santhoshini and @Dammalapati Sai Krishna 

We introduced ourselves and spoke about our motivations in applying models in our work. Santoshini spoke about her interest in applying some of the concepts in transportation, her field of work. Especially the concept of hyperbolic discounting and immediacy bias, which could explain why drivers violate traffic signals and people avoid public transport. Kartik introduced the Political Economy of Government Statistics, a course he pursued at Azim Premji University and how it got him interested in understanding the data generation processes in public data. He shared an interesting story: Nagaland's population reduced in absolute numbers during the 2001-2011 period. This is not because of any insurgency, famine, or migration. But there is a political-economic explanation that competing sub-groups of population inflated their numbers to seek greater political representation. As a result, the 1991 and 2001 censuses overestimated the population. 

As India awaits the next census and delimitation, this becomes a crucial issue.

We discussed common paradoxes like Simpson's paradox and Berkson's paradox. We thought we needed more Indian case studies of these paradoxes.

We discussed the idea of representing any belief as a probability distribution. Santhoshini gave the example of how people's belief in vaccination changed during COVID and how it can be represented as a probability distribution function (PDFs). Kartik spoke about how it's a very Bayesian practice to represent 'prior' beliefs as PDFs. We all kind of agreed how it's better to draw a PDF line when we state our beliefs and assumptions. It helps us reflect better on what we really believe. I presented my blog—Bayesian Feminist—in which I implemented something similar.

Other topics that we touched at surface level: bootstrapping, spaghetti graphs, how regression models only the mean and quantile regressions help to go beyond that, and utility functions.

Some logistics:

1. This is the document to list all reading material for the group-- Reading list | Model Thinker. Group members can add any book, paper, video, or meme to the list for everyone's perusal. Please bookmark the document.

2. Group members can send an email to modelt...@googlegroups.com to discuss any idea/concept with the group. Any email to modelt...@googlegroups.com would be received by every group member. Each email can be a single thread discussing the idea. 

For the next week, we planned to read till the 5th chapter (Normal Distribution) of Model Thinker.

The call will be on August 17th, Sunday. Timings will be updated during the week.

Please add anything I missed or misrepresented. 

Best,

Krishna