Gustavo Angeles

SOCIOECONOMIC STATUS MEASUREMENT WITH DISCRETE PROXY VARIABLES: IS PRINCIPAL COMPONENT ANALYSIS A RELIABLE ANSWER?

The last several years have seen a growth in the number of publications in economics that use principal component analysis (PCA) in the area of welfare studies. This paper explores the ways discrete data can be incorporated into PCA. The effects of discreteness of the observed variables on the PCA are reviewed. The statistical properties of the popular Filmer and Pritchett (2001) procedure are analyzed. The concepts of polychoric and polyserial correlations are introduced with appropriate references to the existing literature demonstrating their statistical properties. A large simulation study is carried out to compare various implementations of discrete data PCA. The simulation results show that the currently used method of running PCA on a set of dummy variables as proposed by Filmer and Pritchett (2001) can be improved upon by using procedures appropriate for discrete data, such as retaining the ordinal variables without breaking them into a set of dummy variables or using polychoric correlations. An empirical example using Bangladesh 2000 Demographic and Health Survey data helps in explaining the differences between procedures.

Purposive program placement and the estimation of family planning program effects in Tanzania

Abstract Most studies evaluating the impact of family planning on fertility treat the presence of family planning clinics as being “randomly” assigned among the areas included in the study. They tend to ignore the possibility that the distribution of services may be related to the fertility level observed in a particular area. In some cases the distribution of services may respond to a conscious effort by public authorities or funding agencies to target areas with observed higher fertility. Even in absence of program planning, the factors determining service placement might be related to the determinants of high, or low, fertility in a particular area. If that is the case, and one fails to account for the endogeneity of family planning services, then the estimated impact of family planning programs will be biased. This article presents a modeling approach to address this issue. The model extends the simultaneous equation framework by integrating an individual-level model of timing and spacing of children ...

The Impact of Community-Level Variables on Individual-Level Outcomes Theoretical Results and Applications

The authors study alternative estimators of the impacts of higher level variables in multilevel models. This is important since many of the important variables in social science research are higher level factors having impacts on many lower level outcomes such as school achievement and contraceptive use. While the large sample properties of alternative estimators for these models are well known, there is little evidence about the relative performance of these estimators in the sample sizes typical in social science research. The authors attempt to fill this gap by presenting evidence about point estimation and standard error estimation for both two-and three-level models. A major conclusion of the article is that readily available commercial software can be used to obtain both reliable point estimates and coefficient standard errors in models with two or more levels as long as appropriate corrections are made for possible error correlations at the highest level.

The Effects of Education and Family Planning Programs on Fertility in Indonesia

Numerous studies indicate that female education is a major determinant of fertility and that the estimated effects are large relative to other variables, including family planning program variables. There are, however, two serious deficiencies in the research relating educational attainment to fertility that could give rise to invalid inferences. First, many public programs, including health and family planning programs, may influence a woman’s decisions about education, and these indirect programmatic effects might be large. Second, nearly all existing studies of the impacts of education on fertility assume that a womans educational attainment is exogenous. Education could be serving as a proxy for such unobservable determinants as ability, motivation, and parental background, as these factors most likely are important determinants of a womans educational attainment. We use the 1993 Indonesian Family Life Survey to compare the estimated impacts of education on fertility from a simple model that assumes the exogeneity of education and an unobserved factor model that allows for endogeneity of schooling. The model imposing questionable exogeneity assumptions appears to overpredict the fertility‐reducing effects of female education, better schools, and higher government health expenditures and to underpredict the importance of family planning programs for reducing fertility and for inducing young women to remain in school.

Prevalence and Correlates of Physical Spousal Violence Against Women in Slum and Nonslum Areas of Urban Bangladesh

This study explores the prevalence and correlates of past-year physical violence against women in slum and nonslum areas of urban Bangladesh. The authors use multivariate logistic regression to analyze data from the 2006 Urban Health Survey, a population-based survey of 9,122 currently married women aged between 15 and 49 who were selected using a multistage cluster sampling design. The prevalence of reported past-year physical spousal violence is 31%. Prevalence of past-year physical spousal violence is higher in slums (35%) than in nonslums (20%). Slapping/arm-twisting and pushing/shaking/ throwing something at the women are the most commonly reported acts of physical abuse. Multivariate analysis shows that the risk of physical spousal abuse is lower among older women, women with post—primary education, and those belonging to rich households and women whose husbands considered their opinion in decision making. Women are at higher risk of abuse if they had many children, believe that married woman should work if the husband is not making enough money, and approve wife-beating norms. This study serves to confirm the commonness of physical spousal abuse in urban Bangladesh, demonstrating the seriousness of this multifaceted phenomenon as a social and public health issue. The present findings suggest the need for comprehensive prevention and intervention strategies that capitalize on the interplay of individual and sociocultural factors that cause physical spousal violence. Our study adds to a growing literature documenting domestic violence against women in urban areas of developing south Asian nations.

The 2005 census and mapping of slums in Bangladesh: design, select results and application.

BackgroundThe concentration of poverty and adverse environmental circumstances within slums, particularly those in the cities of developing countries, are an increasingly important concern for both public health policy initiatives and related programs in other sectors. However, there is a dearth of information on the population-level implications of slum life for human health. This manuscript describes the 2005 Census and Mapping of Slums (CMS), which used geographic information systems (GIS) tools and digital satellite imagery combined with more traditional fieldwork methodologies, to obtain detailed, up-to-date and new information about slum life in all slums of six major cities in Bangladesh (including Dhaka).ResultsThe CMS found that Bangladeshi slums are very diverse: there are wide intra- and inter-city variations in population size, density, the percent of urban populations living in slums, and sanitation conditions. Findings also show that common beliefs about slums may be outdated; of note, tenure insecurity was found to be an issue in only a small minority of slums.ConclusionThe methodology used in the 2005 Bangladesh CMS provides a useful approach to mapping slums that could be applied to urban areas in other low income societies. This methodology may become an increasingly important analytic tool to inform policy, as cities in developing countries are forecasted to continue increasing their share of total global population in the coming years, with slum populations more than doubling in size during the same period.

Cost and results of information systems for health and poverty indicators in the United Republic of Tanzania

OBJECTIVE To examine the costs of complementary information generation activities in a resource-constrained setting and compare the costs and outputs of information subsystems that generate the statistics on poverty, health and survival required for monitoring, evaluation and reporting on health programmes in the United Republic of Tanzania. METHODS Nine systems used by four government agencies or ministries were assessed. Costs were calculated from budgets and expenditure data made available by information system managers. System coverage, quality assurance and information production were reviewed using questionnaires and interviews. Information production was characterized in terms of 38 key sociodemographic indicators required for national programme monitoring. FINDINGS In 2002-03 approximately US

Kernel density estimation as a technique for assessing availability of health services in Nicaragua

0.53 was spent per Tanzanian citizen on the nine information subsystems that generated information on 37 of the 38 selected indicators. The census and reporting system for routine health service statistics had the largest participating populations and highest total costs. Nationally representative household surveys and demographic surveillance systems (which are not based on nationally representative samples) produced more than half the indicators and used the most rigorous quality assurance. Five systems produced fewer than 13 indicators and had comparatively high costs per participant. CONCLUSION Policy-makers and programme planners should be aware of the many trade-offs with respect to system costs, coverage, production, representativeness and quality control when making investment choices for monitoring and evaluation. In future, formal cost-effectiveness studies of complementary information systems would help guide investments in the monitoring, evaluation and planning needed to demonstrate the impact of poverty-reduction and health programmes.

An equity analysis of performance-based financing in Rwanda: are services reaching the poorest women?

Typically, accessibility ratios have been calculated through a simple mathematical division of the number of people in an area by the number of facilities (or staff) in that area. This approach does not take into account the service area of the facility or its proximity to population centers, and is often performed using aggregate numbers for an administrative region. This paper describes an approach to calculating accessibility ratios such as population to facility ratios or population to staff ratios using Kernel density estimation (KDE) within a geographic information system. KDE disperses discrete phenomena across continuous space and is unrestrained by administrative boundaries. Therefore it provides a better representation of the spread of people and services across the landscape. Two types of accessibility ratios are calculated on a national level for Nicaragua: population-per-facility and population-per-staff; the merits of using KDE over traditional approaches are discussed.

Impact of Oportunidades on skilled attendance at delivery in rural areas.

Maternal health services continue to favour the wealthiest in lower and middle income countries. Debate about the potential of performance-based financing (PBF) to address these disparities continues. As PBF is adopted by countries, it is critical to understand the equity effects for maternal services. The aim of this study is to examine the effects of PBF on equity in maternal health service use when no specific provisions target the poorest in the population. In Rwanda, PBF was designed to increase health service use, which was universally low. Paired districts were randomly assigned to intervention and control for PBF implementation. Using Rwandas Demographic Health Survey data from 2005 (pre-intervention) and 2007-8 (post-intervention), a cluster-level panel dataset of 7899 women 15-49 years of age from intervention (4477) and control districts (3422) was created. The impact of PBF on reported use of facility deliveries, antenatal care (ANC) and modern contraceptive use was estimated using a difference-in-differences model with community fixed effects. Interaction terms between wealth quintiles and PBF were estimated to identify the differential effect of PBF among poorer women. The probability of a facility delivery increased by 10 percentage points in the intervention when compared with the control districts (P = 0.014), while no significant effects were noted for ANC visits or modern contraceptive use. Service use increased for intervention and control populations and across all wealth quintiles from 2005 to 2007, with no evidence that PBF was a pro-poor or a pro-rich strategy. Insurance remained a positive predictor of service use. This research suggests that if service use is uniformly low then a PBF programme that incentivizes select services, such as facility deliveries, may improve service use overall. However, if the equity gap is extreme, then a PBF programme without equity targets will do little to alleviate disparities.