ballot

As the next election approaches for the West Contra Costa Unified School District (among other jurisdictions), I thought it would be a good time to analyze previous recent elections in WCCUSD. I’ve set up a project on Github where I detail how I gathered this information and store an R script for analyzing the information.

I was able to go back to the 1995 election. Starting with this election, there have been 10 WCCUSD board elections, 65 candidacies (47 unique candidates), and 25 victors (2 seats then 3 seats are up for each two elections).

The main sources I used to compile the database (CSV file) are from these two web pages:

I have more work to do (including adding in the campaign contributions), but here are a few things I noticed already.

Total Voting

total-voting

This graph shows the difference in turnout for Gubernatorial-year (like 2002), Presidential-year (like 2004), and off-year (like 1995) elections. Until 2010, the board elections were in lower-turnout off years. It’s interesting to note the slight decline in Presidential-year election turnout. Also, the drop-off of 2014 from 2010 seems significant.

One caveat is that I assume if three seats are open the voter selects three candidates. (There’s actually no way to check this from available data.) My “total voting” variable is calculated by dividing votes cast by seats that can be voted on. A drop in “total voting” from one year to the next may simply reflect voters selecting fewer candidates than they’re allowed for strategic or preferential reasons.

Advantage of Incumbency

incumbent-advantage
This shows the success rate for incumbent candidacies as greater than 50% (55.6%), but I expected the advantage to be more. The overall success rate for all candidacies is 38.5%.

Advantage of Ballot Position

ballot-position
ballot-distance
I also looked to see if there was any effect from ballot position on the chances of winning. A commonplace for down-ballot elections is that people are more likely to use “eenie-meenie-minie-moe” techniques for choosing candidates. One technique is to pick candidates based on ballot position.

This doesn’t seem to be the case for WCCUSD elections. The first graph shows success rates for just ballot position. (The 9 and 10 are outliers. They only occurred once as positions.) The second graph tries to eliminate the differences in the significance of a particular position from election to election as the numbers of candidates varies. In the graph, I use a measure “insideness” from the first or last positions.

What Next

I’ll look at other factors like gender and ballot designation along with cleaning things up more. You can always see where things are by checking the Github repository. At the end, I’ll tie it all into a document and post about it.