Afghanistan: 2004 results and 2009 Polling

On the eve of presidential voting in Afghanistan, let us take a final look over what we know going into this election. The Taliban, a Pashtun faction, ruled Afghanistan until 2001, when they were forced out of power by a combined Northern Alliance (a group largely made up of the Tajik, Hazara and Uzbeks who populate the northern portion of the country) and US-NATO forces. In 2004, the first post-Taliban election was held, which saw Pashtun ally of the anti-Taliban movement Hamid Karzai prevail with 55.4% of the vote.

The 2004 election was contentious but without major protest afterwards. While Karzai had his base of support in the Pashtun regions, he ran strongly second in many Northern Alliance provinces, even winning a few. Irregularities were many, and allegations of undue influence by the US and allies were fairly common.

Fast forward to 2009, where a quite similar situation is playing out. Hamid Karzai is running fairly strong in polls taken this summer, and continues to have net positive approval ratings, though his star has fallen significantly since 2004, both with national allies and international forces, including the US.

As previously mentioned, softened support for Karzai with the Northern Alliance faction has been the main hurdle for his re-election. The candidate running second in most evaluations of the contest, Abdullah Abdullah, is the candidate of the Afghanistan National Front, which is largely composed of Northern Alliance members. How many votes can be swayed (or forcibly moved) from the north into each camp will likely be the deciding factor.

Two major US-based polling organizations have taken the temperature of public opinion Afghanistan since summer began, with all three polls funded almost entirely by the US government. The non-profit International Republican Instititute (IRI) conducted in person and telephone interviews across Afghanistan in May and late July, while Glevum and Associates, a consulting firm, did the same in mid July. IRI did a similar exercise ahead of the 2004 elections, and ended up overestimating Karzai's overall victory margin by about 20 points.

Assuming that we can trust these results at a rough level (both pollsters have experience in Afghanistan, have learned lessons since 2004, and have relatively sound methodologies given the terrain), a couple key issues stand out. The main difference between the July polls is the undecided category, something that gave us a lot of trouble when we looked at polling in Iran back in June, when over 27% of Iranians told the American pollsters that they were undecided.

Were IRI's interviewers better able to earn the trust of the population, given their longer history in the country? Or instead, did they do their questioning in a way that made undecided a less attractive option, and therefore got decided, but possibly untrue responses? One other unknown question is whether "Other" candidates in the Glevum study were indeed other names, or a convenient way to "decided", but actually uncommitted.

In particular, a refusal rate of 38% in the Northern region, and overall refusal rates of 33% by Pashtuns and 32% by Tajiks means that key constituencies are underrepresented in the Glevum study. In the study, Tajiks are split between Abdullah (34%) and Karzai (30%), whereas Pashtuns lean far toward Karzai (44%), with Abdullah receiving just 11% of the Pashtun vote.

Nevertheless, if both sets are correct, the minority candidates Bashardost and Ghani seemed to proportionally gain the most of undecided registered voters, while Karzai and Abdullah mainly kept the same proportion as they got previously.

Looking back to IRI's May poll is significant mainly in its illustration of a very wide field back in May, where more than a dozen candidates received 2 % support or more. Dr. Abdullah officially registered as a candidate on 6 May, after speculation that he might drop out, which could explain his weak support during that period, while Karzai made significant progress with many loose allies during this period. The May edition of the IRI report goes into detail with each candidate over 2%, which, rather than lumping them into one "other" category, is as it should be done.

The final results will likely raise more questions than the answer for numbers-oriented people, in large part because the tallies that come from the electoral commission in Afghanistan (IEC) do not have a great track record. Indeed, it is possible that these polls, flawed as they might be, could better approximate the Afghan public sentiment than the results will. International observers quoted by the BBC after the 2005 parliamentary elections as saying that the elections had "very significant fraud," something that remains a major concern. Along with basic security, the integrity of the voting system for this year has again been widely challenged.

Even so, given the requirement of a run-off in the case that there is majority share winner, we could see a two-man second round following tomorrow's voting. A run-off, which could be forced if Karzai loses just 6 points from his 2004 share, would indicate a small but serious rupture in his coalition. Whether this would be a positive development (Competitiveness! Democracy!) or a negative one (Weak, fragmented government; Corruption; Taliban influence), is up for interpretation.
---
Renard Sexton is FiveThirtyEight's international columnist and is based in Geneva, Switzerland. He can be contacted at sexton538@gmail.com