Tweet
Press Release
Afghan Presidential Election - Second Round
Glevum Exit Polls - Key Findings
Glevum Exit Polls - Key Findings
(Face-to-Face and Telephone Exit Polls)
Summary:
Following on from the two pre-second round polls that Glevum Associates
undertook between June 3rd and June 11th 2014, we now present the findings of two exit
polls that we undertook during and after the June 14th election. We conducted face-toface
interviews at 51 polling stations in ten provinces, where we interviewed 2,206
Afghans who had just voted. These voters where asked whom they had just voted for in
this election. A majority of those interviewed, 53%, said that they had voted for Ashraf
Ghani and 47% said they had voted for Dr. Abdullah Abdullah, which suggests that
Ashraf Ghani will be the next President of Afghanistan. As expected when asked about
their ethnicity, the result shows that the majority of Pashtun and Uzbek voters chose
Ghani and the majority of Tajiks and Hazara chose Abdullah. We also conducted a
telephone poll of 2,749 respondents who identified themselves as voters. These voters
also favored Ashraf Ghani (54%) over Dr. Abdullah Abdullah (46%) as the next
President of Afghanistan. These findings are in line with the polls conducted last week
by Glevum, which showed Ghani on 48-49% and Abdullah on 42-45% with 7-9% of
likely voters then undecided.
undertook between June 3rd and June 11th 2014, we now present the findings of two exit
polls that we undertook during and after the June 14th election. We conducted face-toface
interviews at 51 polling stations in ten provinces, where we interviewed 2,206
Afghans who had just voted. These voters where asked whom they had just voted for in
this election. A majority of those interviewed, 53%, said that they had voted for Ashraf
Ghani and 47% said they had voted for Dr. Abdullah Abdullah, which suggests that
Ashraf Ghani will be the next President of Afghanistan. As expected when asked about
their ethnicity, the result shows that the majority of Pashtun and Uzbek voters chose
Ghani and the majority of Tajiks and Hazara chose Abdullah. We also conducted a
telephone poll of 2,749 respondents who identified themselves as voters. These voters
also favored Ashraf Ghani (54%) over Dr. Abdullah Abdullah (46%) as the next
President of Afghanistan. These findings are in line with the polls conducted last week
by Glevum, which showed Ghani on 48-49% and Abdullah on 42-45% with 7-9% of
likely voters then undecided.
Introduction
Glevum Associates is again proud to present the findings of another first for an Afghan
election. Two simultaneously executed, company initiated and sponsored,
election. Two simultaneously executed, company initiated and sponsored,
Exit Polls,
conducted in Afghanistan during and after polling day on June 14
th 2014. Glevum and
our partner, Kabul-based Asia Innovative Research and Communication (AIRC),
conducted a face-to-face poll at 51 polling stations on Election Day, June 14th 2014. This
poll, which interviewed 2,206 actual voters, was undertaken in twelve cities in ten
Afghan provinces – Mazar-e-Sharif/Balkh, Maimana/Faryab, Jaghori and Andar/Ghazni,
Herat/Herat, Kabul/Kabul, Kandahar/Kandahar, Matoon/Khost, Kunduz/Kunduz, Asad
and Abad/Kunar, and Jalalabad City/Nangahar. We also undertook a second
simultaneous telephone poll beginning June 14th and continuing through June 16th,
interviewing 2,749 self-identified voters.
our partner, Kabul-based Asia Innovative Research and Communication (AIRC),
conducted a face-to-face poll at 51 polling stations on Election Day, June 14th 2014. This
poll, which interviewed 2,206 actual voters, was undertaken in twelve cities in ten
Afghan provinces – Mazar-e-Sharif/Balkh, Maimana/Faryab, Jaghori and Andar/Ghazni,
Herat/Herat, Kabul/Kabul, Kandahar/Kandahar, Matoon/Khost, Kunduz/Kunduz, Asad
and Abad/Kunar, and Jalalabad City/Nangahar. We also undertook a second
simultaneous telephone poll beginning June 14th and continuing through June 16th,
interviewing 2,749 self-identified voters.
Survey Information and Methodology
Overview – Face- to-Face Exit Poll
Our face-to-face exit poll was conducted in 12 cities in 10 provinces in Afghanistan.
Fully 2,206 voters were asked for whom they had voted, as they were leaving polling
2
stations on June 14, 2014. The polls are sponsored by Glevum Associates and were
conducted by Kabul-based Asia Innovative Research and Communication (AIRC), who
have conducted multiple surveys for Glevum over a five-year period. The interviews
were conducted in Dari and Pashto. Respondents were initially asked if they had voted in
the June 14 runoff election. Next, they were asked for whom they had voted. If
respondents indicated that they had not voted, the interviews ended. Voters were also
asked their ethnicity.
The 51 polling stations were in the following locations:
Fully 2,206 voters were asked for whom they had voted, as they were leaving polling
2
stations on June 14, 2014. The polls are sponsored by Glevum Associates and were
conducted by Kabul-based Asia Innovative Research and Communication (AIRC), who
have conducted multiple surveys for Glevum over a five-year period. The interviews
were conducted in Dari and Pashto. Respondents were initially asked if they had voted in
the June 14 runoff election. Next, they were asked for whom they had voted. If
respondents indicated that they had not voted, the interviews ended. Voters were also
asked their ethnicity.
The 51 polling stations were in the following locations:
City/Province
Sample
size
Number
of
polling
stations
visited
Sample
size
Number
of
polling
stations
visited
Mazar--‐e--‐Sharif/Balkh
210
5
Maimana/Faryab
160
3
Jaghori,
Ghazni,
Andar/Ghazni
230
5
Herat/Herat
300
6
Kabul/Kabul
649
13
Kandahar/Kandahar
200
5
Matoon/Khost
160
3
Kunduz/Kunduz
160
4
Asad
Abad/Kunar
100
2
Jalalabad/Nangarhar
250
5
TOTAL
2,206
51
210
5
Maimana/Faryab
160
3
Jaghori,
Ghazni,
Andar/Ghazni
230
5
Herat/Herat
300
6
Kabul/Kabul
649
13
Kandahar/Kandahar
200
5
Matoon/Khost
160
3
Kunduz/Kunduz
160
4
Asad
Abad/Kunar
100
2
Jalalabad/Nangarhar
250
5
TOTAL
2,206
51
The total sample size was determined by past sample sizes used in the extensive polling
Glevum has conducted in Afghanistan and also on AIRC’s capacity to undertake a faceto-
face exit poll in a politically volatile country that is unaccustomed to such surveys.
Sample sizes per province/city are proportionate to the populations of the 10 provinces,
although data was weighted by population to account for any variances.
A representative sample of polling stations was determined to ensure that the ethnic
makeup of the sample reflected that of the country. For example, using data from prior
representative samples gathered by Glevum, Kabul is assessed as 33% Pashtun, 39%
Tajik, 21% Hazara, and 7% other ethnicities. Therefore, for this province, about a third
of interviews were held at polling stations in predominately Pashtun areas. This strategy
was used to achieve an ethnic balance in the final results.
Males and females were dispatched to polling stations to conduct interviews throughout
Election Day. A specific number of respondents were interviewed per hour to ensure that
interviews occurred throughout the day. Every fifth person was interviewed until that
specific number was reached.
The margin of error is estimated at 2.07% with a 95% level of confidence.
3
Glevum has conducted in Afghanistan and also on AIRC’s capacity to undertake a faceto-
face exit poll in a politically volatile country that is unaccustomed to such surveys.
Sample sizes per province/city are proportionate to the populations of the 10 provinces,
although data was weighted by population to account for any variances.
A representative sample of polling stations was determined to ensure that the ethnic
makeup of the sample reflected that of the country. For example, using data from prior
representative samples gathered by Glevum, Kabul is assessed as 33% Pashtun, 39%
Tajik, 21% Hazara, and 7% other ethnicities. Therefore, for this province, about a third
of interviews were held at polling stations in predominately Pashtun areas. This strategy
was used to achieve an ethnic balance in the final results.
Males and females were dispatched to polling stations to conduct interviews throughout
Election Day. A specific number of respondents were interviewed per hour to ensure that
interviews occurred throughout the day. Every fifth person was interviewed until that
specific number was reached.
The margin of error is estimated at 2.07% with a 95% level of confidence.
3
Overview – Telephone Exit Poll
From June 14-16, 2014, interviewers spoke to 2,749 respondents with all interviews
being conducted in Dari and Pashtu. The survey utilized a randomly generated sample of
six-digit numbers for area codes within Afghanistan. A total of 200,000 numbers were
then entered into SPSS by our U.S. based analyst and 30,000 numbers were randomly
selected for the Afghan interviewers to call. Those 30,000 phone numbers were dialed to
obtain 2,749 completed interviews.
Interviewers phoned respondents and asked if they had voted on June 14 and, if so, for
whom they voted.
All calls were made to cell phones because landlines are rare in Afghanistan. It is not
uncommon for Afghans to share cell phones and phone numbers although the extent of
this practice is unknown. In the survey, the person who answered the phone was deemed
the respondent. S/he was simply asked if s/he had voted on Election Day to keep the
interview short and minimize refusals. Randomly selecting respondents from all users of
a phone number selected in the sample would be too lengthy. Because incoming calls are
typically also charged to respondents, it was vital to keep the interviews short.
Determining specifically with whom to speak would first require asking if the cell phone
number was shared or used only by one person. If shared, then it would be necessary to
ask questions about with whom it was shared and then randomly selecting interviewees
and reaching those individuals, if they had not answered the phone. Glevum was
therefore concerned that respondents would end interviews that were too long. For the
same reasons, demographic questions were not asked.
The margin of error is estimated at 1.78% with a 95% level of confidence.
This survey was conducted separately but simultaneously with a similar face-to-face exit
poll, the results of which we are also releasing today. The aim of sponsoring two surveys
was to ensure that we were able to provide the most comprehensive coverage with the
limited resources we have available for this effort
being conducted in Dari and Pashtu. The survey utilized a randomly generated sample of
six-digit numbers for area codes within Afghanistan. A total of 200,000 numbers were
then entered into SPSS by our U.S. based analyst and 30,000 numbers were randomly
selected for the Afghan interviewers to call. Those 30,000 phone numbers were dialed to
obtain 2,749 completed interviews.
Interviewers phoned respondents and asked if they had voted on June 14 and, if so, for
whom they voted.
All calls were made to cell phones because landlines are rare in Afghanistan. It is not
uncommon for Afghans to share cell phones and phone numbers although the extent of
this practice is unknown. In the survey, the person who answered the phone was deemed
the respondent. S/he was simply asked if s/he had voted on Election Day to keep the
interview short and minimize refusals. Randomly selecting respondents from all users of
a phone number selected in the sample would be too lengthy. Because incoming calls are
typically also charged to respondents, it was vital to keep the interviews short.
Determining specifically with whom to speak would first require asking if the cell phone
number was shared or used only by one person. If shared, then it would be necessary to
ask questions about with whom it was shared and then randomly selecting interviewees
and reaching those individuals, if they had not answered the phone. Glevum was
therefore concerned that respondents would end interviews that were too long. For the
same reasons, demographic questions were not asked.
The margin of error is estimated at 1.78% with a 95% level of confidence.
This survey was conducted separately but simultaneously with a similar face-to-face exit
poll, the results of which we are also releasing today. The aim of sponsoring two surveys
was to ensure that we were able to provide the most comprehensive coverage with the
limited resources we have available for this effort
Most Compelling Findings – Face-to-Face Exit Poll
Below are the most compelling findings from our face-to-face poll.
Candidates Support
Below are the key findings of this exit poll also presented in graphical form.
According
to the findings of this exit poll, Ashraf Ghani appears to be the clear winner of this
second round and therefore likely the next President of Afghanistan. He is chosen
by 53% of voters interviewed compared to 47% who supported Abdullah.
to the findings of this exit poll, Ashraf Ghani appears to be the clear winner of this
second round and therefore likely the next President of Afghanistan. He is chosen
by 53% of voters interviewed compared to 47% who supported Abdullah.
4
The ethnic breakdown of the voters interviewed shows that the majority of voters were,
as expected, split between the two candidates largely along ethnic lines with Ghani
securing the vast majority of the Pashtun and Uzbek who voted (85%/90%) and Abdullah
picking up the vast majority of the Tajiks who voted (83%). He also secured a majority
of the Hazara vote (73%). Of interest is the fact that Ghani and Abdullah picked up
roughly the same percentage of each other’s core vote. Ghani gained the support of 17%
of Tajik who voted and Abdullah secured 15% of the Pashtun vote. However, Ghani did
much better with the Hazara vote (27%) than Abdullah did with the Uzbek vote (10%).
This difference probably had significant impact on the outcome of the election.
The vote for each candidate by city/province is included below. Perhaps of most interest
is the percentage of Pashtun who voted for Ghani in Kandahar (72%), which reversed a
slight lead that Abdullah had held in last week’s Glevum poll. Ensuring a high turnout of
the Pashtun vote and winning their support was a prerequisite for a Ghani win and he
seems to have achieved both objectives.
as expected, split between the two candidates largely along ethnic lines with Ghani
securing the vast majority of the Pashtun and Uzbek who voted (85%/90%) and Abdullah
picking up the vast majority of the Tajiks who voted (83%). He also secured a majority
of the Hazara vote (73%). Of interest is the fact that Ghani and Abdullah picked up
roughly the same percentage of each other’s core vote. Ghani gained the support of 17%
of Tajik who voted and Abdullah secured 15% of the Pashtun vote. However, Ghani did
much better with the Hazara vote (27%) than Abdullah did with the Uzbek vote (10%).
This difference probably had significant impact on the outcome of the election.
The vote for each candidate by city/province is included below. Perhaps of most interest
is the percentage of Pashtun who voted for Ghani in Kandahar (72%), which reversed a
slight lead that Abdullah had held in last week’s Glevum poll. Ensuring a high turnout of
the Pashtun vote and winning their support was a prerequisite for a Ghani win and he
seems to have achieved both objectives.
5
Most Compelling Findings – Telephone Exit Poll
Below are the most compelling findings from our face-to-face poll.
Candidates Support
Below are the key findings of this telephone exit poll also presented in graphical form.
According to the findings of this exit poll Ashraf Ghani again appears to be the clear
winner of this second round and therefore likely the next President of Afghanistan.
He is chosen by 54% of voters interviewed compared to 46% who supported
Abdullah.
Exit Poll Comparison
winner of this second round and therefore likely the next President of Afghanistan.
He is chosen by 54% of voters interviewed compared to 46% who supported
Abdullah.
Exit Poll Comparison
In both exit polls, Ghani is chosen by a majority of those voters interviewed, strongly
indicating that he has won the second round of this election and is probably the next
President of Afghanistan. These results are also in line with the two pre-second round
polls conducted by Glevum and released last week which also had Ghani (49/48%)
leading Abdullah (42/45%) with between 7% and 9% of likely voters undecided.
indicating that he has won the second round of this election and is probably the next
President of Afghanistan. These results are also in line with the two pre-second round
polls conducted by Glevum and released last week which also had Ghani (49/48%)
leading Abdullah (42/45%) with between 7% and 9% of likely voters undecided.
6
Conclusion
In the face-to-face pre election poll that Glevum conducted last week, Ghani led
Abdullah by 48-45 with 7% undecided. Given that Ghani clearly had the momentum
going into the second round, the outcome of these exit polls are not surprising. Ghani
appears to have picked up most of the undecided vote and consolidated his core vote to
secure a close and hard fought but nevertheless conclusive win.
Interestingly Pajhwok Afghan News reporting of early results has Ghani at 52.45% and
Abdullah at 47.52%.
Abdullah by 48-45 with 7% undecided. Given that Ghani clearly had the momentum
going into the second round, the outcome of these exit polls are not surprising. Ghani
appears to have picked up most of the undecided vote and consolidated his core vote to
secure a close and hard fought but nevertheless conclusive win.
Interestingly Pajhwok Afghan News reporting of early results has Ghani at 52.45% and
Abdullah at 47.52%.
All early indicators therefore suggest that Ashraf Ghani is the
winner of the second ballot.
Further Information
winner of the second ballot.
Further Information
A copy of our PowerPoint presentation for this survey can be obtained either by email
directly from Glevum or by accessing the website of Glevum (glevumusa.com). It will
be available after 11 am EST June 17
directly from Glevum or by accessing the website of Glevum (glevumusa.com). It will
be available after 11 am EST June 17
th 2014.
Contact Information
Andrew.j.garfield@glevumusa.com
Telephone: 571 228 4813
Comment