Philippines - Labor Force Survey (October)2007, 4th Quarter

The sampling design of the LFS uses the sampling design of the 2003 Master Sample (MS) for Household Survey that started July 2003.

Sampling Frame
As in previous household surveys, the 2003 MS used an area sample design. The Enumeration Area Reference File (EARF) of the 2000 Census of Population and Housing (CPH) was utilized as sampling frame. The EARF contains the number of households by enumeration area (EA) in each barangay.

With the conduct of the 2003 Listing of Households, the number of households of the selected sampled EA from the CPH EA lists were updated and the sample sizes that were initially generated where subsequently adjusted.

Stratification Scheme
Prior to sample selection, the PSUs in each domain were stratified as follows:
a) Proportion of strong houses (PSTRONG). PSTRONG is defined to be the percentage of housing units occupied by households that are classified as made of strong materials in terms of both the roof and outer walls, based on the data from the 2000 CPH. A roof is considered made of strong material if it is made of either galvanized iron, aluminum, concrete/clay tile, half galvanized-half concrete, or asbestos. The outer wall is considered made of strong material if it is made of concrete, brick, stone, wood, half concrete-half wood, galvanized iron, asbestos, glass.
b) A variable labeled AGRI. An initial variable is an indicator variable computed at the barangay level. That variable has the value 1 if more than 50 percent of the households in the barangay are engaged in agriculture or fisheries and 0 otherwise, based on the 2000 CPH Barangay Schedule. To obtain a measure at the PSU level, a weighted average of the barangay indicator variable was computed for all the barangays within the PSU, weighted by the total number of households, in the barangay. Thus, the 1 value of AGRI at the PSU level lies between 0 and 1.

c) Per capita income (PER CAPITA). PERCAPITA is defined as the total income of the municipality divided by the total population in that municipality. Note that the PERCAPITA value of the PSUs is the same if the PSUs are in the same municipality. The municipal income used was the 2000 municipal income sourced from the De4partment of Finance. If the 2000 municipal income was not reported to the Bureau of Local Government Finance (BLGF), 2001 income was used. If no 2000 or 2001 municipal income was reported, the median income of the municipal class of the municipality was used.

Sample Selection

The sample design is an epsem in each region (Equal Probability Selection for each Member). Given the overall sample, the number of PSUs, EAs and Housing Units (HUs) were determined so that the epsem property within region was preserved.

The PSUs were selected with probability proportional to some estimated measure of size (PPES). Since PSUs vary considerably in size, PSUs were identified as certainty and non-certainty PSUs. Each PSU which was selected with certainty (selection probability is greater than 1) was treated as a separate stratum. In each certainty selection, sample EAs were selected with PPES in each sampled PSU; and housing units (HU) were selected with equal probability in each sampled EA.

A housing unit is structurally separate and independent place of abode which by the way it has been constructed, converted, or arranged, is intended for habitation by a household. It may contain more than one household. In the 2000 CPH, about two percent of the total household population consist of two or more households.

For non-certainty selection, PSUs were sampled from a set of strata defined for each domain. In each domain, these PSUs were grouped into strata by province, Highy Urbanized Cities (HUC) or Independent Component Cities (ICCs). To further improve on the precision of the estimates, PSUs within province/HU/ICC stratum were further stratified with respect to some socio-economic variables that were associated with poverty.

Within each selected PSUs, EAs were selected with probability proportional to size and households with equal probability within selected EAs.

In each HU, all households were selected. However, for operational considerations, only 3 households will be selected with equal probability in HU consisting of more than 3 households.

Sample Size

The 2003 Master Sample consist of a sample of 2,835 PSUs of which 330 were certainty PSUs and 2,505 were non certainty PSUs. The number of households for the 2000 CPH was used as measure of size. The entire MS was divided into four sub-sample or independent replicates, such as a quarter sample contains one fourth of the PSUs found in one replicate; a half-sample contains one-half of the PSUs in two replicates.

For the purpose of the master sample, the sample EAs selected with certainty were classified as sub-sample of replicate 0, and all the non-certainty PSUs were classified in any of the sub-samples 1, 2, 3 and 4.


Deviations

Replacement of sample households is allowed only if the respondent to LFS cannot be contracted after 3 callbacks. If the sample household moved out from the sample barangay and can no longer be located then the replacement should be the household currently residing in the housing unit previously occupied by the original sample. If no household currently residing in the aforementioned housing unit then a replacement may be selected from among the neighbors who has the closest socio-economic and demographic characteristics as the original sample household. Household members who are identified as eligible respondents but cannot be interviewed should never be replaced.


Of the total households selected for October 2007 LFS, 44,097 were considered eligible households where 95.0 percent or 41,890 households responded.

Among the regions, Ilocos Region, CALABARZON and Cordillera Administrative Region had the highest response rates of 96.8 percent, 96.7 percent and 96.4 percent, respectively. Zamboanga Peninsula registered the lowest repsonse rate of 92.2 percent.

Region Response Rate

Philippines 95.0
National Capital Region 94.1
Cordillera Administrative
Region 96.4
I - Ilocos Region 96.8
II - Cagayan Valley 96.0
III - Central Luzon 95.2
IVA - CALABARZON 96.7
IVB - MIMAROPA 93.0
V - Bicol Region 96.2
VI - Western Visayas 95.7
VII - Central Visayas 95.6
VIII - Eastern Visayas 94.8
IX - Zamboanga Peninsula 92.2
X - Northern Mindanao 95.8
XI - Davao Region 94.1
XII - SOCCSKSARGEN 94.1
XIII - Caraga 92.8
Autonomous Region in
Muslim Mindanao 93.1


Estimation Procedure
Calculation of Basic Weights

Weights

Following a standard approach, the weights to be used in analyzing surveys based on the 2003 MS are developed in three stages.

First, base weights are computed to compensate for the unequal selection probabilities in the sample design.
Second, the base weights are adjusted to compensate for unit non-response.
Third , the non-response adjusted weights are further adjusted to make some weighted sample distributions conform to distributions obtained from another source.

Please refer the more detailed explanation on the three stages of the weights with the given formula based on the 2003 Master Sample provided in the External Resources.

Final Survey Weight

The final survey weight assigned to each responding unit is computed as the product of the base weight, the non-response adjustment, and the population weighting adjustment, as described above. The final weights should be used in all analyses to produce valid estimates of population parameters.