Ulat ng Bayan June 2021 Tech details
A. LOCATION & FIELDWORK SCHEDULE
|National Capital Region||June 7-16, 2021|
|North/Central Luzon||June 7-14, 2021|
|South Luzon||June 7-15, 2021|
|Visayas||June 7-13, 2021|
|Mindanao||June 7-16, 2021|
Respondents for the survey were 2,400 adults (18 years old and above) who were interviewed face-to-face on current political, social, economic issues as well as personal and household information.
C. SAMPLING METHOD
Sample sizes and Error Margins
Below is the distribution of the sample by area and the corresponding error margin at the 95% confidence level.
|AREA||SAMPLE SIZE||ERROR MARGIN|
|National Capital Region||600||±4%|
|Cordillera Administrative Region||20||±22%|
The number of sample barangays was distributed among the 17 regions proportional to population size. In each sample barangay, five (5) probability respondents were selected.
|AREA||%||SAMPLE BARANGAYS||PROBABILITY RESPONDENTS|
|National Capital Region||25%||120||600|
|Cordillera Administrative Region||1%||4||20|
For the National Capital Region:
Stage 1: Selection of Sample Barangays
One hundred twenty (120) barangays were distributed among the 17 cities and municipalities in such a way that each city/ municipality was assigned a number of barangays that is roughly proportional to its population size. An additional provision was that each municipality must have one sample barangay. Barangays were randomly selected without replacement from within each city/municipality.
Stage 2: Selection of Sample Households
In each sample barangay, interval sampling was used to draw 5 sample households. A starting street corner was drawn at random. The first sample household was randomly selected from the households nearest to the starting street corner. Subsequently, every 6th household was sampled.
Stage 3: Selection of the Sample Adult
In each selected household, a respondent was randomly chosen from among household members who were 18 years of age and older, using a probability selection table. To ensure that half of the respondents were male and half were female, only male family members were pre-listed in the probability selection table of odd-numbered questionnaires while only female members were pre-listed for even-numbered questionnaires. In cases where there was no qualified respondent of a given gender, the interval sampling of households was continued until five sample respondents were identified.
For the rest of the Philippines:
Stage 1: Allocation of sample barangays to Regions
The number of barangays allocated to each region was proportional to population size.
Stage 2: Allocation and Selection of Sample Cities/Municipalities to Regions
Within each region, 15 cities/municipalities were allocated to the regions in proportion to household population size. Sample cities/municipalities were selected without replacement and with probability proportional to household population size.
Stage 3: Selection of Sample Barangays
Once the cities/municipalities have been selected, the allocated number of barangays were distributed among the sample cities/municipalities in such a way that each city/municipality was assigned a number of barangays roughly proportional to its household population size. However, it was ensured that each city/municipality will be assigned at least one sample barangay.
Sample barangays within each sample city/municipality were randomly selected without replacement.
Stage 4: Selection of Sample Households
Within each sample barangay, five households were selected through interval sampling. In sample urban barangays, a random corner was identified, a random start generated, and every 6th household was sampled. In rural barangays, the designated starting point could be a school, the barangay captain’s house, a church/ chapel, or a barangay/municipal hall and every other household was sampled.
Stage 5: Selection of the Sample Adult
In each selected household, a respondent was randomly chosen from among household members who were 18 years of age and older, using a probability selection table. To ensure that half of the respondents were male and half were female, only male family members was pre-listed in the probability selection table of odd-numbered questionnaires while only female members was pre-listed for even-numbered questionnaires. In cases where there was no qualified respondent of a given gender, the interval sampling of households was continued until five sample respondents were identified.
D. RESEARCH METHODOLOGY
The questionnaire was prepared in Filipino.
The final, Filipino version of the questionnaire was translated into Bicolano, Cebuano, Ilocano and Ilonggo by language experts.
Training of field interviewers was conducted in four central locations: Quezon City, Cebu City, Iloilo City, Zamboanga City and Davao City. The interviewers who covered Luzon were trained in Quezon City. Those trained in Iloilo City covered Ilonggo-speaking regions while those trained in Cebu City covered all of Cebuano-speaking areas (Central and Eastern Visayas). Interviewers trained in Zamboanga City covered Zamboanga Peninsula areas while interviewers trained in Davao City covered other Mindanao areas.
Training activities in Visayas and Mindanao were conducted online during which the objectives of the study, the questionnaire, interview techniques and interval sampling in the field were discussed.
Supervisors reporting to the field manager monitored the study full-time. They observed interviewers (14% of total were observed by supervisors), followed-up and conducted surprise checks on the field interviewers. They also ensured that field logistics were received promptly and administered properly.
b. Spot Checking
Spot checking was done at various stages of fieldwork. The first one took place after about 30% of interviews were completed. The second spot-checking was conducted after 60% completion and the last one, immediately after 90% completion of interviewing.
During spot-checking, 25% of the unsupervised interviews were re-interviewed/back-checked. If serious errors persisted after spot-checking, the original interviews were invalidated and respondents were re-interviewed. An error is considered serious if dishonesty in recording is apparent or if there was a serious misinterpretation of the study in that it resulted in wrong information.
If some questionnaires were found incomplete or had inconsistent answers, the interviewer was asked to go back to the respondent, so that the questionnaire could be completed and corrected.
c. Number of Calls and Substitution
Respondents sampled who were not available during first attempt were booked for an appointment within the day. Two (2) valid call backs were made within the same day and if probability respondent is still not available, a substitute who possessed the same qualities (in terms of gender, age bracket, working status and socio-economic class) as the original respondent was interviewed. The substitute respondent was taken from another household beyond the covered intervals in the sample barangay.
d. Field Editing
After each interview, the interviewer was asked to go over his/her own work and check for consistency. All accomplished questionnaires were submitted to the assigned group supervisor who, in turn, reviewed at least 50% of total interviews. If there were persistent errors, 100% of questionnaires would be reviewed.
3. Data Encoding
An office editor conducted a final consistency check on all questionnaires prior to coding. Interview sheets were edited/checked twice by office editors before the information were encoded. A data entry computer program verified and checked the consistency of the encoded data before data tables were generated.
To ensure quality of data, 73% double encoding was conducted.
4. Preliminary Data Processing
Preliminary runs are conducted on the data wherein some frequency data tables are generated so as to check the quality of the encoding done. In particular, the encoded data are checked for accuracy of variable labels, variable values, completeness of data entries, and consistency of entries.
E. WEIGHTING PROCEDURE
Design weights corresponding to each probability respondent sampled was applied.
Base weights were computed to compensate for the unequal selection probabilities in the sample design. The associated base weight for the respondent (R ) in the lth household in the kth barangay in the αth city/municipality in a given region is given by
Βαlk= 1/pαlk ,
where pαlk is the unconditional joint probability of selecting the respondent into the sample. The term pαlk is the product of the
unconditional probability of randomly selecting the αth city/municipality in a given region (pα);
conditional probability of randomly selecting the kth barangay from the αth city/municipality (pαk);
conditional probability of randomly selecting the lth household from the kth barangay (pαkl) ;
conditional probability of randomly selecting respondent R from a roster of eligible respondents in his/her household (pαklR).
That is to say,
pαlk = pα * pαk * pαkl * pαklR
The cities and municipality in the National Capital Region and Cebu City and Davao City are chosen with probability 1.
The base weight for R’s household is given by
Βαlk,HH = pαklR * Βαlk.
The base weights were then adjusted to projected population counts.