national difference in difference analysis

[aberman6]

I did a very messy first pass mainly to set up a good pipeline that we can use after we settle the data selection criteria. This analysis:

• Included all states
• Excluded state years with missing VEP_Counted

Just to be extra clear, I'm not saying this is our final result but it's a good baseline to have.

The causal effect is ~19 % point increase in votere turnout.

[katevcoulter]

hey i think we need to figure out which states we are including before we go any further on this that way we can also work on gathering the county level data / stop pulling data so we don't run into version issues

[aberman6]

With the updated exclusion criteria, the D&D looks like this:

The causal effect is ~10% point increase in voter turnout.

[katevcoulter]

YAY

On Apr 16, 2020, at 2:19 PM, Anna Berman notifications@github.com wrote:

With the updated exclusion criteria, the D&D looks like this: https://user-images.githubusercontent.com/16299065/79491868-15ec3600-7fed-11ea-978b-cfe967f7efca.png The causal effect is ~10% point increase in voter turnout.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/MIDS-at-Duke/unifying-data-science-final-project-primaries/issues/6#issuecomment-614816788, or unsubscribe https://github.com/notifications/unsubscribe-auth/AKLBR5ZV5BVHZ36TCO6WV33RM5DZJANCNFSM4MH5F3VA.

[nickeubank]

Damn!

[aberman6]

We realized our dataset was incorrect in reporting of primary vs. caucus. With the updated election types we now have 4 treatment states and 20 control states:

Treatment group: Colorado, Washington, Minnesota, & Maine

Control Group = Alabama, Arizona, Arkansas, Florida, Iowa, Massachusetts, Michigan, Mississippi, Missouri, Nevada, New Hampshire, North Carolina, North Dakota, Oklahoma, South Carolina, Tennessee, Texas, Utah, Vermont, & Virginia

Our causal effect is now 15.7 points!