ABS-CBN - PULSE ASIA'S MAY 2010 ELECTIONS EXIT POLLS

TECHNICAL DETAILS

I.GEOGRAPHICAL COVERAGE

The ABS-CBN – Pulse Asia exit poll conducted on May 10, 2010 covers the entire Philippines: National Capital Region (NCR), CAR, Regions 1 to 12, ARMM and CARAGA. The survey targeted 12,405 registered  voters who actually voted. However, only 12,024 questionnaires made it to the cut-off period for data processing. Of these, 7,296 were completed interviews yielding an unweighted completion rate of 61%.

Data was obtained from the respondents through face-to-face interviews conducted in their homes after they voted. Respondents were first asked to fill up a sample ballot to reflect the votes they cast and to drop the sample ballot in a sample ballot box. They were then asked further election-related questions, a good portion of which focuses on the automated election system and the overall conduct of the elections.

II.SAMPLING DESIGN
    
A.Domains and Strata

The principal domains of the survey are the National Capital Region (NCR), Balance Luzon (Luzon excluding NCR), Visayas and Mindanao. Barangays in each region were stratified by urbanity based on the 2000 Barangay Reference File (BRF) prepared by the National Statistics Office (NSO)(1) . Thus there were 33 strata (one in the case of NCR, two in the rest of the regions).

B.Sampling Units

The ABS-CBN – Pulse Asia exit poll employs a multi-stage sampling design.  A total of 2,700 target respondents was allocated to NCR, while the remaining 9,705 respondents were proportionally allocated to the 32 strata based on registered voter population. Barangays were drawn with probability-proportional-to-size with replacement (ppswr) from each stratum in the first stage. Enumeration areas or spots were then randomly selected without replacement in the second stage and households within the selected barangay systematically selected at the third stage. Fifteen (15) respondents were pre-listed per urban spot, 12 per rural spot. A registered voter within the household was then selected via simple random sampling (using a Kish grid).

III.FIELDWORK PROCEDURES

1.Supervision
Supervisors reporting to the field manager monitored survey operations full time. They observed at least 20% of total interviews, and conducted follow ups and surprise checks on field interviewers.

2.Spot-checking
During spot-checking, at least 30% of the unsupervised interviews were re-interviewed/back-checked.

3.Number of Calls and Substitution
A target household that gave an “outright refusal” was substituted with one having the same socio-economic class and gender. Call backs were conducted until 3 p.m. in cases where the door of a target household was locked. After 3 pm. respondents who were still not available were substituted with one having the same socio-economic class and gender.

Interviews and callbacks were done until 1 am. of May 11, 2010.

C.Field Editing
After each interview, the interviewer was asked to go over her own work to check for consistency.
All accomplished interviews were submitted to the assigned group supervisor who, in turn, edited every interview.

D.Transmission of Data
Supervisors transmitted the data using both SMS and fax.
 
IV.RESULTS OF FINAL CALL

Only 61% of the 12,024 questionnaires that made it to the cut-off time for data processing were completed, i.e., had responses. Some 7% of respondents eventually said that they did not vote, with only 2% citing disorderly voting procedure as reason. About a fifth of target respondents were not available for interview (mostly falling under the category of ‘locked doors’), higher than the 2007 exit poll figure of  17%. It is possible that the incidence of non-availability for interview is higher because of the longer period it took many voters to cast their votes.

Across the domains of the study, NCR had the lowest completion rate, at about half, while Mindanao had the highest at 70%. This is due to the high refusal rate in NCR (16%); the refusal rate in Mindanao was the lowest, at  about 2%.

Across regions excluding NCR, Central Visayas exhibited the lowest completion rate (51%) mainly due to non-availability (37%).On the other hand, Region 11 (Southern Mindanao) had the highest completion rate (77%). Completion rates were also higher in rural areas than in urban areas (67% vs 58%) and in the case of socio-demographic groupings, among females than among males (67% vs 58%).

With these patterns in the completion rates, some care is needed in interpreting voter preferences of candidates who are known to have strong support in the areas with higher and/or lower completion rates, i.e., NCR, Central Visayas and Region 11.

____________________________________________

(1) Urbanity of barangays found in the 2010 COMELEC data of registered voters but not found in the 2000 BRF will be imputed based on the modal urbanity in the municipality to which these belong.