Top-notch research, jerome

The New Orleans Mayoral Election was held this weekend.  Before the election, there was a lot of controversy about accomodations for displaced voters.  The election was postponed from February, and the state set up satellite early voting stations in other parts of Louisiana, but resisted setting up any outside of the state.  For voters displaced to other parts of the country, the options were to vote absentee by mail, or travel to Louisiana.

You've probably heard by now that voter turnout was 36%.  But what does that mean?  How many votes came from voters in the city, in the state, or outside?  How does it compare to past elections?  Finding this information in more detail has been difficult, but after a number of calls to state and city departments, and a search of what's available online, I have some of it and want to share.

The total population of New Orleans before Katrina was about 462,000.  According to the Lousiana Elections Division, 297,909 of them were registered voters.  108,348 cast votes for mayor, and this is where the number you've been hearing comes from:

108,348 / 297,909 = 36.4%

Voter turnout is an ambiguous term.  36.4% represents the percentage of registered voters who voted for mayor.  By this definition, if you have an election where only 1 out of 10 eligible voters register, and all of them vote, "turnout" is 100%.  Another meaning of voter turnout is the percentage of eligible voters who vote in an election; in my opinion, a more meaningful number.  The EAC's 2004 election survey estimated that the voting age population of New Orleans was 343,365, of whom 334,379 (97.4%) were citizens.  So,

108,348 / 334,379 = 32% of those eligible voted for mayor in this election

(I did not go past the decimal point on that percentage because I do not know what the margin of error on the EAC estimate is)

I had two key questions I wanted to explore:

1. How does this compare to past elections?
2. What is the breakdown of these numbers in terms of displaced voters and those back in the city?

Compared to Past Elections

Many news outlets reported that this year's 36% turnout was significantly lower than the almost 46% turnout in 2002.  Those who argue that this is not a big problem go back one election further, to 1998, when turnout was 38.6%.  To see the real pattern, let's look at the Louisiana Elections Division's historical election data...

ElectionVotes cast for Mayor 1994 round 1 152,702 1994 round 2 170,824 1998 round 1 116,543 (there was no runoff required in 1998) 2002, round 1 133,601 2002, round 2 130,475 2006, round 1 108,348 ElectionVotes cast for Governor 1991, Duke vs. Edwards 199,665 I included the oft-cited 1991 election for Governor to illustrate what I think I see in this table: Voter participation in New Orleans fluctuates from election to election, and is likely affected by the issues at stake and how they motivate voters.  When an ex-Klansman was on the ballot for Governor, more than 2/3 of New Orleans voted.  Other people may see something else: A slow, fitful decline in voter participation.  Either way, this year represents the lowest turnout yet, both percentagewise and in absolute numbers, even though interest in the election was probably higher than for any election the city has faced since Duke vs. Edwards.

Conclusion: Voter turnout this year was significantly lower than in the past.

Votes Cast in the City and Away

A quick recap: Many of New Orleans' residents were displaced by Katrina...

• Some have returned and reside in the city again, or in the metro area.
• Some are further away, but still in or near Louisiana.
• Some are in far-flung portions of the country.

There were several methods of voting available...

• Vote in person, on election day - available in the city of New Orleans only.
• Vote early, at a satellite voting station - throughout Louisiana, but not on election day.
• Absentee vote by mail - anywhere... but not allowed for some people (such as new registrants who had never voted in person)

There's been a lot of confusion about these methods of voting, and how they're tallied and reported.  First, the satellite voting stations were all early voting locations only.  Some voters in other parts of the state thought they needed to go to their nearest satellite voting station on election day, only to find it closed.  And many people reporting and looking at the numbers think that election day in person voting totals include satellite voting.  In fact, votes cast at these voting stations were counted as absentee votes, which leads to the second confusion: People mistakenly think that the number of absentee ballots reported, represents absentee by mail voting.  In fact, it represents the sum of votes cast by mail and at satellite voting stations.  Finally, note that not all of the satellite voting stations were far from New Orleans.  Many current, non-displaced residents of New Orleans voted early at satellite stations.

Got that?  Good.  Now you can correctly interpret the numbers I found:

There were 108,348 votes cast for mayor...
In person on election day                         "Absentee" votes85,846     22,502 11,326 by mail11,176 at satellite stations   about 6,000 (estimate) cast inside the cityabout 5,000 (estimate) cast elsewhere in Louisiana
(these numbers and unofficial estimates courtesy of Jennifer at the Louisiana Secretary of State's office - thank you!)

What we really want to know is, how many current residents voted, and how many currently displaced former residents voted.  The simple, but wrong answer that you've probably seen reported in the press is that "about 20,000" evacuees voted, not counting those who returned to the city to vote in person.  If you're with me so far, you can tell where this error comes from: failing to take into account that a majority of satellite votes, which were included in the "absentee" tally, were actually cast in the city.

We can break total into three parts:

• 85,846 (election day) + about 6,000 (satellite) = about 92,000 votes were cast in the city.
• About 5,000 (satellite) votes were cast outside of the city.
• The remaining 11,326 (mail-in) votes could have come from inside or outside the city.

The only way to tell for sure would have been to look at the postmarks, and as far as I can tell, nobody kept a tally.  So we don't know how many of them came from people inside the city who voted by mail for traditional reasons such as disability, and how many came from displaced residents who wanted to vote without having to travel to Louisiana.  But, we can estimate.  Since there were no satellite voting stations around the state in past mayoral elections, the rate of absentee voting in those elections should tell us approximately what the "normal" rate for mail-in voting is:

1994 primary, 1,836 absentee ballots for mayor, 1.2% of 152,702.
1998 primary, 1,006 absentee ballots for mayor, 0.9% of 116,543.
2002 primary, 2,171 absentee ballots for mayor, 1.6% of 133,601.

We don't know how many of those votes came from inside or outside the city, either.  But I think we can safely estimate that any significant increase in absentee voting this year, represents displaced voters.  If we take the normal level of absentee voting by the resident population to be a high 2%, then about we should've seen about 2000-2200 from the resident population this year.  If we take the normal level to be 1.2%, then we should've seen about 1300.  So, of the 11,326 mail-in votes, I estimate that about 10,000 came from outside the city.  Since some percentage of "normal" mail in voting also comes from people who are away, consider this a low estimate.  In other wards, I am erring on the side of overestimating the in-city vote.

In summary:

* 86,000 (election day) + 6,000 (satellite) + 1,300 (mail-in) = 93,300 votes cast in the city.
* 5,000 (satellite) + 10,000 (mail-in) = 15,000 votes cast from outside the city.
(we're working with only about 3 significant figures here, but it does add up to 108,300, and I didn't plan it that way)

Displaced Voter Turnout

To put it all together, we need one more pair of numbers: How many of New Orleans' voters are back in the city, and how many are still displaced?  That, unfortunately, is the most elusive.

Neither the Secretary of State's office nor the City of New Orleans were able to give me current estimates of the displaced population, let alone the displaced voter population, and they were not able to suggest anyone who does have those estimates.  The city did commission a "Rapid Population Estimate Project" a few months ago.  Based on surveys taken January 28-29 - 5 months after Katrina, and 3 months before the election - and statistical methods, they estimated 160,500 - 202,200 residents were back.  They then collected data from college dorms, hotels, and cruise ships providing housing, which add 28,000-29,000 more, for a total nighttime population estimate of about 210,000.  How many of these are voting age citizens?

[ Note: I am not a statistician, and may be a little out of my depth in this next section.  I welcome comments from actual statisticians. ]

One private study found that only about 44,000 registered New Orleans voters have filed change of address forms since Katrina giving a new address away from the metro area.  Based on this, they conclude that a large majority of voters are back in the city, even though more than half of the total population are away.  I find this study's claim dubious, because they also found that about 168,000 New Orleans voters had not filed change of address forms at all, and want us to assume this means they were not displaced.  If we take their 80% estimate at face value, there are about 238,000 registered voters in back New Orleans: considerably higher than the city's upper estimate of total population at the end of January!

A better way to estimate voters, I think, is to assume that the percentage of the current population who are voting age citizens is about the same as it was in the past.  The Census Bureau estimated the 2004 population of New Orleans as 462,269.  That means that about 297,909 / 462,269 = 64% were registered, and about 334,379 / 462,269 = 72% were voting age citizens.  

Multiplying the city's population estimate range by 64% gives us an estimate of 103,000 - 129,000 registered voters back in the city.  If we assume the same percentage for college dorms, hotels, and cruise ships (which are probably housing a lot of temporary reconstruction workers who vote elsewhere), that adds about 18,000 more voters.  Let's split the middle and estimate that there are about 10,000.  Using the rapid population estimate of 181,400 residents, we have 181,400 * 64% + ~10,000 = 125,000 registered voters back in the city, + / -27,000 (64% of 202,200-160,500).  Doing the same calculation for voting age citizens gives us 181,400 * 72% + ~10,000 = 140,000 voting age citizens, + / -30,000, or about 110,000 - 170,000 voting age citizens currently in the city.  Because of the dorm, hotel, and ship residents, I think I am once again erring on the side of overestimating the in-city population.

Finally:

Total pre-Katrina population: 462,269
... 334,379 (72%) voting age citizens, 297,909 (64%) registered voters
Displaced population: 252,000 (230,000-274,000)
... 194,000 voting age citizens, 173,000 registered voters
Back in New Orleans: 210,000 (189,100-230,800)
... 140,000 voting age citizens, 125,000 registered voters

In-City Voter Turnout:
93,300/140,000 = 66% of voting age citizens,
93,300/125,000 = 75% of registered voters

Displaced Voter Turnout:
15,000/194,000 = 8% of voting age citizens,
15,000/173,000 = 9% of registered voters

At this point, there are so many assumptions built in that the margin of error is very high.  At one extreme, for example, if the resident population is close to the highest estimates, then we have only 55% turnout of voting age citizens in the city (61% of registered voters), and 9% / 10% displaced turnout.  But that's about as far as we can go in that direction, and the difference is still striking.  Even if we err very far in the direction of overestimating the number of people back in the city, and underestimating the number still displaced, two things are very clear:

1. In the city, this was one of the highest turnout elections in New Orleans' history, rivaling the 1991 Duke vs. Edwards election.
2. There is a gaping maw between the rate of voter turnout among those back in the city, and those still displaced.

Hypothesis: There was very high voter interest, but inadequate effort was made to inform displaced voters about the election, and to accomodate them.  Many many more displaced residents would have voted if they had been informed and if satellite voting stations had been made available outside of Lousiana.  As Ted Shaw of the NAACP Legal Defense Fund said in a recent interview,

"If we could have out of nation polling places for Iraqis living in the United States, we certainly should have had it for people displaced by Katrina living in the United States; democracy should begin at home."

Disclosure: I am a paid blogger for John Bonifaz, running for secretary of state in Massachusetts.  See my post at johnbonifaz.com about Bonifaz's take on what secretaries of state should have done to help Katrina victims in their states.