The Louisiana Secretary of State has data exports of the election results. The City-Parish Planning Commission has maps of the voting districts which I've used to manually draw the precinct boundaries. I haven't included early voting since I haven't yet found this data broken down between "Yes" and "No".
One thing to pay attention to is that the 2002 and 2010 elections were parish-wide while the 2012 election was only within the city limits of Baton Rouge, Baker, and Zachary. You will see some artifacts of this when comparing the voter counts between the first two elections and 2012: precinct voter counts drop along the edges of the map because fewer voters were allowed to vote.
| 2002 | 2010 | 2012 | 2012-2010 | |
|---|---|---|---|---|
| Vote* | N/A | view | view | view |
| Voter count | N/A | view | view | view |
| Turnout | N/A | view | view | view |
* Early voting not included (See discussion below.)
Clear map
The Louisiana Secretary of State (SoS) website provided the data on election results in various slices. The main set that other people have probably used is the file obtained from clicking the "CSV" link at the bottom of electionresults.sos.la.gov/graphical/ for the April 21 election date (and result type "Parish"). This file (FILE 1) breaks down the Yes/No votes by precinct (not including the early voting) followed by a row of Yes/No votes by early voters (with no information on which precincts the early voters were in).
The SoS also provides a pdf file "2012_0421_ParishStats.pdf" (FILE 2) that breaks down early voter counts by precinct but with no indication of how they voted. Because the Yes/No percentages are missing, this file is only useful for error-checking counts with other data sources.
Another SoS file is "2012_0421_par.pdf" (FILE 3), which lists registered voters and actual voter counts per precinct, useful for calculating turnout by precinct.
The second two files (FILE 2 and FILE 3) also break down voting by political party affiliation and "race" ("white", "black", and "other"), but I'm not interested in that sort of analysis.
Additionally, the SoS website had briefly listed what it called "unofficial turnouts" for each precinct that had only one voting location. (E.g., precinct 1-3 had an unofficial turnout of 21.5% while precincts 1-10A and 1-10B didn't have any posted turnout.) As of this writing, these percentages have been removed, likely replaced with FILE 3.
The following discussion will pertain to the Baton Rouge precincts. The counts in these three files do not reconcile. Subtracting Yes+No counts in FILE 1 from actual voter counts in FILE 2 should give the number of early voters by precinct: {Early and Apr 21 total vote count} - ({Apr 21 Yes vote count} + {Apr 21 No vote count}) = {early voter count}. However, these results are typically different from the precinct counts in FILE 3. To total all precincts together using this formula, FILE 2 minus FILE 1 early votes equal 3046. The FILE 3 precinct total is 3143, a good bit more than this. Furthermore, the "Early Voting" row counts in FILE 1 equal 3102. Since FILE 1 appeared first and since FILE 1 is used in two of these totals calculations, I'm going to assume that the vote counts in FILE 1 are incorrect. By how much is hard to say, but it's surely not on the order of a thousand (2656 being the official difference between total election Yes and No votes in Baton Rouge).
One last thing I should point out, I am unable to figure in the early voting Yes/No votes in my by-precinct analyses because that information has not been provided by the SoS website. (Turnout calculations and total voter amounts do include early voters.)
These are what's called a "stem plot". The first one shows how precincts in Baton Rouge are generally home to like-minded people. To be illustrative, if everyone in a precinct agreed with each other, there would be a single spike, either in the 0 row or in the 10 row, indicating 0% voted Yes (i.e., 100% voted No) or 100% voted Yes. (The Zachary plot is a good example that approximates the 0% precinct type.)
BTR has two spikes, one around 30% and another around 90%. This plot, to me at least, is an interesting view on what the "2012 Vote" map shows, which is a geographic division of tax support.
The single spikes in the Baker and Zachary plots show that each precinct basically contains the same mix of opinion. There aren't the "Yes" and "No" parts of town like in BTR.
The vertical axis is in tens of percents. The horizontal lists occurrences of the ones place for each tens place. Each occurrence shown is a precinct. This plot indicates that precincts mostly voted in two spikes: largely against (displayed as 20-39% "yes") or largely for (displayed as 80-95% "yes") the tax.
34902 voters.
117 precincts. (An "early voting" category is treated as a quasi-precinct in the original data set.)
5 precincts had 0 votes and so aren't shown here.
0 | 1 | 7 1 5 2 | 6 7 4 4 9 8 1 1 5 4 5 6 5 5 1 3 9 3 | 2 5 8 9 9 5 0 8 9 1 1 7 2 0 8 4 | 4 1 1 4 3 2 0 5 | 8 5 0 5 1 4 3 7 3 7 1 6 | 6 7 0 8 1 7 | 5 3 7 9 2 4 6 7 4 8 | 2 9 9 9 6 8 4 8 4 8 1 9 2 3 7 6 4 9 1 7 8 6 9 6 0 9 | 0 0 2 3 2 0 1 0 3 1 2 3 0 0 2 2 1 0 6 1 10 |
Examples:
0 | 9 = 9% 8 | 2 = 82%
Baker data shows that precincts were much more moderate in their voting, indicating either larger precincts or a better mix of people with opposing viewpoints.
3109 voters.
20 precincts. (An "early voting" category is treated as a quasi-precinct in the original data set.)
4 precincts had 0 votes and so aren't shown here.
0 | 1 | 2 | 3 | 4 | 0 4 5 | 4 9 5 3 0 8 3 6 | 3 7 3 1 2 7 | 0 2 8 | 9 | 10 |
Zachary voted extremely against the tax.
2943 voters.
18 precincts. (An "early voting" category is treated as a quasi-precinct in the original data set.)
2 precincts had 0 votes and so aren't shown here.
0 | 0 1 | 7 6 7 3 8 9 8 7 1 2 | 1 0 3 | 9 4 4 | 5 5 | 6 | 7 | 8 | 9 | 10 | 0
These two stem plots may be harder to wrap your head around. I made them to see if they dig up anything interesting in the data. Each digit to the right of the "|" counts for a precinct. The number is the ones place and the number to the left of the row's "|" is the tens. Add them together to get the precinct's vote count, "yes" in the first plot and "no" in the second plot. This isn't so illustrative as it could be because the plots ignore early voting due to the SoS not providing precinct information.
What jumps out first is that there were a large number of precincts where there were 10-19 "no" votes. After some further staring, I can see their shapes are different. Because it's so hard to grasp what this is telling us, this probably indicates that there is a better way to represent the data.
First, I will try combining these two plots into one soon. A common "tens" stem, "yes" precinct counts on the left, and "no" precinct counts on the right. I will also try a plot with an axis of "precinct" and the other of "votes" ("yes" in one direction and "no" in the other). This latter one will probably be more familiar to people.
YES 0 | 0 0 5 0 0 0 1 | 7 0 2 | 3 | 6 8 4 | 0 1 5 | 6 6 | 5 0 0 7 7 | 0 6 8 5 0 1 8 | 5 8 2 1 9 | 2 4 9 7 7 1 2 10 | 6 4 8 11 | 3 8 0 1 4 1 5 5 12 | 6 2 7 1 5 3 3 3 6 5 13 | 9 8 1 9 14 | 1 4 7 0 1 8 4 15 | 9 1 1 16 | 0 9 4 0 8 17 | 9 9 4 5 8 0 18 | 3 1 19 | 6 6 2 20 | 0 9 7 21 | 3 5 9 22 | 5 9 9 23 | 2 24 | 6 0 2 2 25 | 2 4 7 7 26 | 3 27 | 28 | 3 29 | 30 | 4 6 31 | 6 32 | 1 33 | 2 34 | 8 3 35 | 36 | 8 37 | 5 7
NO 0 | 6 5 6 0 0 0 0 0 1 | 8 0 2 9 0 3 3 6 9 7 8 3 4 4 2 5 4 2 | 4 8 6 5 4 3 8 3 7 2 1 3 | 0 6 1 5 7 0 1 6 4 | 7 1 2 5 5 2 2 2 5 | 2 2 9 0 4 0 6 5 6 | 3 7 | 3 8 7 7 8 | 1 8 7 1 4 9 | 8 2 10 | 9 9 11 | 5 9 12 | 4 1 0 9 13 | 7 14 | 8 15 | 9 2 16 | 17 | 7 18 | 6 9 19 | 20 | 21 | 22 | 4 6 23 | 2 1 6 24 | 0 25 | 1 26 | 27 | 28 | 3 29 | 30 | 5 1 31 | 0 9 9 32 | 3 9 33 | 34 | 5 8 35 | 2 0 6 4 36 | 7 1 6 37 | 38 | 39 | 40 | 41 | 8 42 | 43 | 9 7 44 | 45 | 46 | 8