Multilevel modeling of poverty data

Date of Publication

2014

Document Type

Bachelor's Thesis

Degree Name

Bachelor of Science in Mathematics

Subject Categories

Physical Sciences and Mathematics

College

College of Science

Department/Unit

Mathematics and Statistics

Abstract/Summary

The hierarchical nature of the data is often ignored in the analysis on Philippine datasets. These databases are usually applied on standard regression models for estimation and prediction. Using multilevel modeling allows for the relationship between variables clustered within levels to occur. In this paper, the researchers used a three-level modeling on the Community-Based Monitoring System (CBMS) Pasay 2011 database. The first level is modeled for the frequency of government programs using Poisson and negative binomial regressions. The classification whether a household is poor or not poor based on the PCI criterion was modeled in the second level using binary logistic regression, conditional on whether the household has been involved in at least one government program. Lastly, the poverty gap, using the gamma distribution and the generalized beta distribution, was modeled for the third level, conditional on the whether the household is considered poor and has received at least one government program. The best models were determined with the use of model fit statistics. Moreover, results show that the number of couples and the number of members in the household who attended school are the significant predictors in modeling poverty gap, conditional on the first two levels.

Abstract Format

html

Language

English

Format

Electronic

Accession Number

CDTU019128

Shelf Location

Archives, The Learning Commons, 12F, Henry Sy Sr. Hall

This document is currently not available here.

Share

COinS