Exit Poll  Methodology

The exit poll results are based on interviews with a probability sample of voters exiting polling places on Election Day.

Sampling Frame

The sampling frame used is an electronic list of all the electoral stations in West Bank and Gaza Strip including Jerusalem Governorate.  The sampling frame showed under-representation of the Annexed Jerusalem area; therefore we decided to use PCBS population figures as of 2002 as a basis for the sample allocation to governorates. The sampling frame contains 3045 Electoral Stations.

Sampling
A stratified two-stage cluster systematic random sample is selected using proportionate allocation to get a self-weighted sample.

The sample is selected in two stages. First, a stratified systematic sample of 152 electoral stations (electoral stations) is selected representing the 16 governorates in West Bank and Gaza Strip including Jerusalem Governorate. Electoral stations are selected with systematic random sampling and the sample allocation is done proportionate to the population in each governorate, the population figures used are the 2002 PCBS population predictions based on the 1997 population census. Each voter in a governorate had the same chance to have his or her electoral station selected. Second, within each electoral station, voters are sampled systematically throughout the voting day at a fixed rate that gives all voters in an electoral station the same chance of being interviewed.  This sample design is used to produce a self-weighted sample so as to be able to do data analysis and produce the main results in as short time as possible.

Stratification

Two levels of stratification are used to maximize the efficiency of the sample through capturing most of the variation exists in the study population, stratification variables used were: Region (West Bank, and Gaza Strip), and Governorate (16 Governorates: 11 in West Bank and 5 in Gaza Strip)

Sample Size

Electoral Stations  (ES) are used as primary sampling units (PSUs) where 152 (5.1 %) ES’s are selected out of the 3045 ES’s exist in the sampling frame, then from each ES voters are selected with systematic random sampling using fixed interval (1 out of 10), hence the total sample size will not be determined before finishing the field work, however, an average of 60 voters are expected to be sampled per ES giving an expected total sample of 9000 voters. 

The sample size was calculated based on the following assumptions:

1. The margin of error in estimating the population parameters will not exceed 1 percentage point

2. The level of risk used for estimating the parameters is 5%

3. The design effect (Deff) equals 1.2

4. Most of the statistics calculated are proportions and percentages

The formula used to calculate the sample size is

                                                                          n= (1.96)² * p* (1-p) * Deff 1 E²

Where

p: Is the proportion/percentage

Deff: Is the Design Effect

E: Is the margin of error

Table 1: Electoral Stations, Population and Sample Distributions by Governorate

Weighting
No weighting is needed to reflect the probability of selection since the sample is a self-weighted one. However, if a significant non response occurred then there will be a need for weighting to adjust for the possible bias resulted from voters who were missed or refused to be interviewed, this weighting adjustment will be based on their observed age, and sex.

Analysis of Individual Governorates

This sample is designed to reflect the views of all voters nationwide, and one should be careful in analyzing voting behavior in geographical areas (governorates.) since the sample would be too small in some governorates giving a much larger margin of error.

Sample Plan (to be used by the Field Work Supervisors and Interviewers)

1. From each selected (assigned) electoral station select, randomly, the voters by the following procedure:

• Start working in the electoral station by the time the voting process starts there

• Select a random number r between 1 and 5 (say 3)

• Start counting the voters as they exit the electoral station (count 1, 2, ..., r)

• Select the voter with the number corresponding to the random number chosen r (3 in the example). And make the interview with him/her.

• Then start counting again from 1 until you reach 5 then make the interview with voter number 5 and start counting again from 1 to 5 interviewing the voter number 5 in each time and so on until the voting process ends in the assigned electoral station.

• If a selected voter refused or couldn’t participate, fill a questionnaire for him/her and write down in his/her questionnaire at least the following information:

a)      Interview result (refused, missed,)

b)      His/her sex

c)      His/her age group (as you observe it).

• Under any circumstances, do not replace the voter who refused to participate, because that might alter the age/sex distribution of the sample

• Make sure to assign a questionnaire for non-respondents (refusals,) because that will help the data analysis staff to re-weight the sample to adjust for non-response.

• Do not change the selection interval (5)

• The sample should spread over the whole Election Day.

2. Do not replace the selected person if you do not like him/her

3. Remember; we do not intend to make the sample representative of the location, all what we intend to do is to make the total sample representative of the total population and this could be achieved if we stick to the sample plan

 Top of this page    |    DSP Home