Baby Rose performed her new album with them and it was awesome

ACIC off to a good BADBADNOTGOOD start

👀!!

We reanalyze to recent studies on forest protection interventions to provide robust evidence that problems of local "leakage" were negligible.

We show how randomization and the simple difference in "circle averages" identifies the AME, and how the Conley spatial covariance estimator with an empirical degrees of freedom correction provides consistent inference under bounds on the extent of spillovers and spatial homophily assumption.

We take an agnostic approach to spatial spillover effects and interference to establish an estimand, the "average marginalized effect" (AME), that characterizes how effects bleed out from interventions at points or polygons in space, accounting for arbitrary spatial interactions.

Insofar as spending freeze hits elementary & secondary public schools, impact will vary by state. NCES has consolidated fiscal data to FY20, which does not include pandemic allocations (with exception of NM & Guam). Here is share of schools revenue from federal funds by state/territory.

Chat GPT prompt: “Based on what you know about me make an image about what you think my life looks like”

Someone actually left these on the book giveaway pile??

Will go down in history

Fortunately there is a pre analysis plan…

Always amusing to make the chatbots trip over themselves...

Many thanks to Rice PoliSci @ricepolisci.bsky.social and Matt Tyler especially for a great visit as part of the Hudspeth lecture series, named in honor of Hank Hudspeth. My talk focused on a study empowering tax payers in Kinshasa. (Paper coming soon.)

GANs for generating synthetic evaluation data in medical research?

From Kearns and Roth’s excellent primer on algorithmic fairness global.oup.com/academic/pro...

🤷🏻‍♂️

My Quant 2 (causal inference) PhD course underway: up today, a primer on design-based statistical inference

Thanks for posting — always good to have this reminder about how to start thinking about political constraints

Eg here is a graph showing how our approach (solid line) maintains good coverage properties as the number of covariates increases, relative to her approach (lasso)