Friday, September 13, 2013

Where are undecided voters likely to move?

Honduran presidential election polling has settled into a pretty clear pattern.

That has LIBRE candidate Xiomara Castro in the lead, with National Party candidate Juan Orlando Hernández close, but always behind.

Meanwhile, voters disinterested in any of the candidates running form anywhere from 22% to 31%. If you add decline to state to none of the above in the poll with the largest none of the above totals-- a July poll by Paradigma-- almost half of those polled (48.9%) are not saying they have committed to any of the candidates running.

And that brings the question: can we predict who probably will benefit from movement by those voters?

Our answer: no, we cannot. Paradigma and CESPAD, two of the polling sources, have no previous track record in Honduras for us to use to assess how it might do. Even if there was a wealth of prior polling, none of it would come from the first election after a coup d'etat that tore apart one of the traditional parties, and ushered in two new, and apparently reasonably popular, new parties.

But there are others who do think they can predict what will happen in the coming election. The Economist Intelligence Unit posted a note on the election September 12 (h/t Bloggings by Boz, who has his own post on the way the polling is shaping up, for bringing this to our attention).

The Economist/Intelligence Unit doesn't discuss all the polls we have summarized. We continue to caution that there are clear differences between polling agencies which suggest the best way to assess poll data is across time by an individual pollster.

But the thing we find most surprising is that despite the low level of absolute support for individual candidates, the tightness of the race between the top two candidates, and the large block declaring no preference or absolutely not in favor of any of the existing candidates, The Economist/Intelligence Unit analysis projects a final winner:
We maintain our forecast that Mr Hernández will win the November election as the PN Is likely to gain a larger share of the undecided vote on polling day. However, a low turnout could still result in a LIBRE victory.

Without any explanation of why they think more voters currently registered as undecided will turn to the Partido Nacional, we find ourselves troubled by this projection. We have discussed privately a number of scenarios that we can see happening-- but predicting how the undecideds will break seems, to us at any rate, without basis.

CESPAD-- the latest source of polling data to become available-- provides an illustrative example of how complicated such prediction is in Honduras, this year.

Let's start with how voters polled feel about the Partido Nacional-- which currently controls the presidency (under Porfirio Lobo Sosa) and the Congress (for much of the Lobo Sosa presidency, led by Juan Orlando Hernández). In a word: the people are not satisfied. Since February 2012, CESPAD has found that a majority of the people say that the current president is making things worse (47%) or has no effect on the country's problems (38%). Less than 10% said his administration was helping improve conditions.

Is it unfair of us, then, to doubt that the Partido Nacional is going to attract undecided voters?

CESPAD actually gives us a fascinating look at which voters are still undecided (or uninterested in any of the actual candidates) by party affiliation. (We will spend another post just on this topic. For now, though, we simply want to trace who might be expected to shift to Juan Orlando Hernández.)

About 17% of the Liberal Party members; about 21% of the Partido Nacional members; 11% of Anti-Corruption party members; and only 4% of LIBRE members, report being undecided. So where would the undecided voters that might go Partido Nacional come from?

Liberal Party voters not supporting their own party's candidate are overwhelmingly planning to vote LIBRE (about half of these rogue voters). Unless we assume that the undecided voters represent a much more conservative group within the Liberal Party, it is hard to imagine that 17% of voters breaking for Juan Orlando Hernández. The actual candidate of the Liberal Party represents the more conservative wing of that party, so we think it is unlikely these undecideds feel the Liberal candidate is not conservative enough, which would be the motivation to vote PN. If they cannot find it possible to vote LIBRE, there are other somewhat progressive options available. Or they can reject the entire system and vote Anti-Corruption Party-- as over 6% of Liberal Party Members are already planning to do.

The tiny percentage of LIBRE members who are not saying they will vote for Xiomara Castro indicate preferences either for the Liberal Party, or for an even more progressive minority party. Even if we thought all the undecided LIBRE voters were closet law-and-order pro-business conservatives, that would be a minuscule addition to Partido Nacional rolls.

Salvador Nasralla has support from 82% of the party he founded. Only 1% of the party members plan to vote Partido Nacional-- against about 3% each for the Liberal Party and LIBRE. Again, it seems highly unlikely that the 11% undecided/no preference party members think the Partido Nacional is their best choice-- the whole basis of the Anti-Corruption Party is that present politicians are too corrupt-- and the present politics of Honduras is dominated by the PN, and until recently, had as its most effective leader, Juan Orlando Hernández, as head of Congress.

That leaves us with undecided/no preference voters who said they were members of the Partido Nacional itself. This is a substantial number of potential voters-- 21% of party members. Presumably, what The Economist/Intelligence Unit is expecting is that these voters will settle for their party candidate in the end. Yet if we assume their reluctance to state a preference stems from a lack of enthusiasm for their party's candidate, we might consider whether they will follow the lead of the more than 20% of their fellow party members who have decided to vote for someone else. Here, the main beneficiary has been Nasralla, who has support from about 9% of PN members. Nasralla's Anti-Corruption Party is not particularly progressive/liberal in its politics; it is pro-business and solidly Honduran in its identity. The anti-crime theme of Nasralla's campaign is familiar PN policy. It is worth noting that rogue PN voters support Xiomara Castro in almost as high numbers-- suggesting some of the dissatisfied PN voters may be more progressive.

To sum up, we would echo what CESPAD analysts wrote:
The weakening of the traditional party loyalty, is tending to modify the historic balance of political forces. In the presidential candidacies there exists a clear migration of the vote of the Partido Liberal (PL) and of the Partido Nacional (PN) toward the new political movements. Xiomara Castro is the one that most attracts the vote from other parties: 23% of the Liberal vote and 8% of the Nationalist vote.

CESPAD doesn't offer much support for a claim that the Partido Nacional has support to gain:
Effectively, the electoral preferences for Xiomara Castro and Juan Orlando Hernández have grown since February of 2012 until July of 2013. Nonetheless, the pace of the growth has been greater for Xiomara Castro. Simultaneously there has been registered a growth in electoral sympathies for the LIBRE party (8 points) and a deterioration of the electoral preference for the Partido Nacional (PN) (falling 4 points). [emphasis added]

That seems to us to caution against assuming that the PN has some elasticity to count on. We won't be surprised if Juan Orlando Hernández is declared the winner come election day; but we don't think the data allow for anyone to actually say this is an evident outcome to expect.


paul said...

An important factor in any election is party organization on that day. Who has volunteers to make sure supporters actually vote, and a structure to ensure that happens?
Strategic voting in a multi-party race needs to be considered as well. Nasralla supporters might conclude their votes, at least for president, will be wasted. Will they vote for him anyway, stay home or decide to support their second choice. (I have no idea who that would be.)

RAJ said...

Voter turn out in Honduras may be as much about enthusiasm as it is about organization. Since two of these parties are new, we have little to go on to predict how the get out the vote campaign will proceed. As for strategic voting-- that is what we are trying to assess here. The data on how party affiliates already are breaking may help indicate that Anti-Corruption Party undecideds, for example, if they do vote, might go 3:1 for one of the more liberal options (instead of the national party). As an Anti-Corruption party, we would be surprised if these voters were enthusiastic about the party in power-- the Partido Nacional. But Nasralla's support has mainly been motivated by his own fame; so these may be voters who decide not to bother. Libre voters are motivated and enthusiastic-- the traditional parties, apparently less so.

RAJ said...

Proceso Digital has a new story-- really, it should be labeled an opinion piece-- about how the fragmented voters are breaking. Ignoring the weirdness of their emphasis (LIBRE is the rise of Chavismo from the dead), they come to essentially the same conclusion we do: the center-right is less cohesive today than is the left of center in Honduras. They write that the left:

has achieved a strong intention to vote, the center atomized.

This leads the unnamed author(s) of this piece to insert-- in bold face-- a number of claims about LIBRE intended overtly to motivate the unmotivated center-right voters with the specter of "21st century socialism". It is an amazing article, and along with other bias in their coverage of Xiomara Castro de Zelaya, makes Proceso Digital currently the least objective of Honduras' not-very-objective media.