thx!

The findings show that the built environment exerts an important on inequalities in civic participation. We also provide a new way of using computer for inductive inference for image data. 5/5

Finally, we use a causal discovery framework and the quasi-random assignment of voters to polling places to provide evidence that these are not just correlates but that at least some elements of the built environment have a plausibly causal effect on voter turnout 4/5

We use explainable AI tools to identify exactly what features visible in imagery seem to improve turnout: green space, tree cover, official locations, and spacious/breezy buildings -- possibly because they mitigate voting barriers like heat, insecurity, and social distancing fears 3/5

Do the physical characteristics of the polling places to which citizens are assigned affect voter turnout? We use a convolutional neural network applied imagery of every polling place in Mexico to show that building imagery is highly predictive of turnout, over and above standard predictors 2/5

Ok, I think I understand, your reference was to pre-2015 Canada

But there’s been a right wing backlash in Canada too? (Temporarily averted by the counter-reaction to Trump…)

Thanks for this public good..!

Thx for coming!