The existing household surveys in the Philippines have been using the barangay as a sampling unit. However, the absence of quantitative indicators for categorizing barangays is serving as a major constraint in the development and improvement of survey designs, and the present urban-rural definition is deemed inadequate. In an attempt to create a new system that can clearly pinpoint areas lagging behind in development which may properly guide economic planners and policy makers, this study developed a statistics-based system of grouping barangays according to the level of agricultural development. This was done by listing a wide array of 46 statistical data/indicators for the 26 sample barangays from Batangas (derived from the Quarterly Corn Survey). The identified indicators/variables underwent various multivariate data analysis techniques to reduce its dimensions, such as Correlation, Cluster and Principal Component Analysis (PCA), given that the number of indicators/variables exceeded the number of barangays under investigation (46 > 26). The clustering resulted to 20 groups/clusters of variables, wherein for each group, the variable(s) was chosen subjectively according to its capability in significantly influencing agricultural development (e.g., average farm size, number of chicken and duck per household, number of corn mills). This enabled the reduction of the number of variables from 46 to 23, and was eventually used to classify the 26 barangays according to their level of agricultural development. The Discriminant Analysis was also employed to determine the most important variable to distinguish the barangays falling into levels of agricultural development. The results of the study showed that the case of five (5) clusters using two (2) groups yielded a p-value of 0.9113, as opposed to the case of four (4) clusters and six (6) clusters both using two (2) groups with p-values equal to 0.1993 and 0.1538, respectively.

Keywords:

  Request for Download