Page 69 - Contributed Paper Session (CPS) - Volume 1

CPS1071 Ibarra Aaron R. Poliquit et al.
a. Estimation Procedure
The design-based estimator of total milk production is computed as follows:

∑
̂
= ∑ ∑ , = =1 ×

=1 =1

where: ≡ final weight of sampled barangay, ≡ total milk production
ℎ

of the sampled diary farm in i sample barangay, and ≡ total number of
ℎ
dairy farms in sampled barangay. The design-based estimator of variance
ℎ
of total milk production is computed as follows:

2
2
̂
() = (1 − ) + ∑ (1 − )
2
1
=1
where: ≡ first stage sample variance, ≡ second stage sample variance
2
2
1
2
within sample barangay.
ℎ

2.5 Measure of Precision
The precision of the estimates for totals were evaluated using the
coefficient of variation (CV) and mean absolute percentage of error (MAPE).
Estimates with coefficient of variations which are less than or equal to 10%
implies precision, and hence, reliable estimates.

3. Results
Tables 1 and 2 show the computed number of sample barangays across
different strata and CVs per province with the number of animals on the milk
line and number of female breeders as a stratification variable. The number of
animals in the milk line as stratification variable computes a larger number of
sample barangays than using the number of female breeders across all
scenarios. The reason for this is that the distribution of animals on the milk
line has a larger coefficient of variation than the distribution of female
breeders. Only one scenario was considered for simulations, where the
number of strata is 5 and the target CV is equal to 5.0%. The computed sample
sizes for this particular scenario was deemed as the right compromise between
the level of precision of estimates and the cost of operations. The total number
of sample barangays for all the eight (8) provinces for the considered scenario
is 99 sample barangays when using the number of animals on the milk line as
a stratification variable, and 88 sample barangays when using the number of
female breeders as stratification variable, respectively.

58 | I S I W S C 2 0 1 9