Case Study:
Vote-By-Mail Analysis

The Client: Washington Monthly

Washington Monthly, an independent nonprofit magazine interested in find ways to make government work bette

The Challenge: Determine the effects of Utah’s universal vote-by-mail systems

We needed to find ways to compare different counties with different voting methods while accounting for historical differences in population and turnout patterns. Were the new county-level Vote By Mail systems having an effect on voter turnout?

The Solution: Use areas without vote-by-mail as pseudo-controls

We found several avenues of analysis, all leading to the same conclusion. Primarily, we used difference-in-difference comparisons to examine expected turnout versus actual turnout; this analysis showed that voters exceeded their expected turnout in VBM counties — while merely voting at expected rates in non-VBM counties. We then confirmed these results by creating an individual-level regression model, finding similar turnout effect sizes (around five to seven percentage points). Finally, we found a particularly illustrative example in a mountain town that just so happened to straddle the line between a VBM county and a non-VBM county. We even called election administrators in these counties to make sure we were properly understanding the data! Read the full study here.