I think there is gold in it, which is kind of a stone?

Shakshouka!

First time.

Chicken and Noodles, over mashed potatoes.

Regional delicacy anywhere Amish people live.

Galliano root beer float was freaking incredible, thanks.

They're just using the original dialogue at this point.

The elevator pitch of my identification strategy vs. what it looks like written out in potential outcomes

Here in the US, laypeople expect academics to make shit up and then waste each others' time with our made up shit.

Does this work?

We take early/late respondent mean estimates from the Norway in Corona Times (NCT) survey from www.nber.org/papers/w29549 and estimate our model *without control variables or randomized incentives*.

We beat sample means on 4/6 of outcomes, tie on 1, and do worse on 1.

9/n

But Clint! We hate your parametric assumptions!

1. If you’re ok with a standard Heckman correction with a randomized response incentive, no you don’t!

2. We provide an overidentification test that can be used with >2 notifications –“effects” of late reminders on outcomes should be zero.

8/n

Let’s try it out!

In a survey of undergrads, we estimate an entrepreneurship-intention gender gap of 15 pp, at odds with a 20 pp gap among respondents.

This is driven by + (-) selection among men (women).

This mimics some gendered patterns in measurement error, e.g. tinyurl.com/yc4mxt2x

5/n

Fin.

Guys I made homemade sushi.

I am so sorry.