i_felony and i_misdemea are the *outcome* variables on the left hand side though — so I don’t understand the question. They define E[crime].

]]>Thanks Andrew, perhaps the better question to ask is how can we see (calculate) how much of an effect do our “match.out” variables (i_felony & i_misdemea) influence the E[crime].

Similar to how the coefficients from the RHS of a regression influence the outcome variable.

]]>Not exactly sure what your question is Robert. The point estimates you noted are:

Observed – E[Crime]

where E[Crime], “expected crime”, is estimated via the synthetic control areas (constructed from demographics + pre-intervention trends). And the standard errors are estimated via a permutation approach, where we shuffle what micro areas are assigned treatment, redo the whole analysis, and see how much E[Crime] varies over the permutations.

When you say “point estimates that a typical regression would output”, I am thinking about coefficients on the right hand side. This isn’t quite the same, the demographics are matched to be 100% the same for this sample.

You might check out this other blog post I did, showing how to use lasso regression to estimate synthetic control, https://andrewpwheeler.com/2019/12/06/using-regularization-to-generate-synthetic-controls-and-conformal-prediction-for-significance-tests/, to see how the weights can be related to regression. They are not a 100% map to one another though.

]]>Just a quick question for you! From the case study you performed a microsynth I understand that we have two causal relationships we want to discuss.

The first is: i_felony = TotalPop + HISPANIC + Males_1521 + FHHPerc+BlackPerc & the second is: i_misdemea= TotalPop + HISPANIC + Males_1521 + FHHPerc+BlackPerc

How would we go about obtaining the outcome point estimates that a typical regression would output? Would they be just running your 3rd last code chuck and thus the estimates for the outcome variables i_felony and i_misdemea be -10.49 and -21.24?

Thanks!

]]>So we might even have another angle to look at variation — HOAs that implement renters clauses.

]]>