Mike Gechter

Assistant Professor of Economics @uawaltoncollege . @YaleRISE affiliate. Working on external validity, cities, and firms in developing countries.

[Mike Gechter @MikeGechter]

Tere I'm so proud of you, you did it! #SUGrad23

[Mike Gechter @MikeGechter]

@hlanthorn @economeager So the more info the better. Representativeness is good to have. I have another paper (https://t.co/qNn3pHVOZp) thinking about representativeness in outcomes and characteristics. But we always wonder about unobserved reasons for site selection and their correlation with the TE

[Mike Gechter @MikeGechter]

@hlanthorn @economeager Great question @hlanthorn . In this paper, we don't know exactly how sites were selected, which is why we need an instrument. If we knew exactly how sites were selected, we could control for those characteristics (think selection on observables).

[Mike Gechter @MikeGechter]

@sadish_ @economeager But this could be an interesting route to a new IV, which we really encourage!

[Mike Gechter @MikeGechter]

@sadish_ @economeager Interesting idea! We picked JPAL/IPA because many of their services are specifically targeted towards RCT studies. Whereas other survey firms may work on a mix of RCTs and observational studies. So they differentially change the cost of RCTs vs. observational.

[Mike Gechter @MikeGechter]

@sadish_ @economeager A challenge I can see with survey firms in general is finding a systematic way of cataloging their number over time because I would imagine the cross sectional number is correlated with growth which likely influences treatment effects. Maybe industry associations would help.

[Mike Gechter @MikeGechter]

@economeager @angrist_noam Yes, thanks! Hope alls well.

[Mike Gechter @MikeGechter]

@DavidRhysBern @economeager Nice to hear from you, David! The observational studies are selected in their own way since they will be done when its too expensive to experiment. So if costs are correlated with, this will have an impact.

[Mike Gechter @MikeGechter]

@economeager @anthlittle @soashworth @TomPepinsky Just as I was grateful to have @economeager working on the Bayesian parts!

[Mike Gechter @MikeGechter]

@rglenner @economeager @anthlittle @soashworth @TomPepinsky Interesting! So there are treated and untreated people in the surrounding area, and the RCT encourages people to take the treatment?

[Mike Gechter @MikeGechter]

@besttrousers @economeager Depending on your objective (most precise inference on the experimental TE for a given observational study, most precise inference on the mean experimental TE across studies...) you could fit our model to get at the question, just not something we've looked at.

[Mike Gechter @MikeGechter]

@besttrousers @economeager We're after something a bit different. Some studies are observational, some experimental. We want to draw aggregate conclusions. At our most ambitious, we want to predict the experimental TE in an observational study using the RCTs.

[Mike Gechter @MikeGechter]

I appreciated the chance to read these articles by @bNeethipudi, a student in my economic development source: https://t.co/HGIqfPtQQm https://t.co/fqA8xqaxBJ . Making the social sector inclusive is so important

[Mike Gechter @MikeGechter]

@EricVerhoogen will do, looking forward to seeing it.

[Mike Gechter @MikeGechter]

@EricVerhoogen any chance you're willing to share slides from your SCID-IGC keynote?I thought of doing metrics work on PF estimation in DCs

[Mike Gechter @MikeGechter]

Made a last-minute donation to the Democratic party via @actblue https://t.co/aeIa7yPXbX

[Mike Gechter @MikeGechter]

Grateful to receive imagery grant from @DigitalGlobeFDN to investigate equity and efficiency effects of land policy in dev country cities!

[Mike Gechter @MikeGechter]

.@falleneconomist interesting. I'd be surprised if the difference cited in the headline result is statistically significant, though.

[Ben Wood @bdkwood]

.@MikeGechter's pointers on conducting "external" #replication research https://t.co/ix6p1csxrU (it's hard). @anbrowndc @3ieNews

[Mike Gechter @MikeGechter]

@akshaykore @mumbaidata Thanks! How did you convert the PDFs to shapefiles? Would you be willing to share the shapefiles you created?