Nonlinear Decomposition Technique Research Articles

Care choices in Europe: To Each According to His or Her Needs?

Growing long-term care (LTC) needs represent a major challenge for our aging societies. Understanding how utilization patterns of different types of care are influenced by LTC policies or changes in the population composition such as age patterns or health can provide helpful insight on how to adequately prepare for this situation. To this aim, this paper explores how individuals choose between different forms of LTC. We exploit variation between countries as well as between individuals within countries using data from the Survey of Health, Ageing, and Retirement in Europe (SHARE). Using nonlinear decomposition techniques, we break down the difference in utilization rates between countries into differences based on observed sociodemographic and need-related characteristics and differences in the impacts of these characteristics, which allows us to identify the drivers behind differences in the formal-informal care mix. Our results show that a substantial fraction of the observed country differences can be explained by differences in impacts. We argue that this is a result of the different incentives provided by the different LTC systems.

Classification of imbalanced ECG beats using re-sampling techniques and AdaBoost ensemble classifier

Computer-aided heartbeat classification has a significant role in the diagnosis of cardiac dysfunction. Electrocardiogram (ECG) provides vital information about the heartbeats. In this work, we propose a method for classifying five groups of heartbeats recommended by AAMI standard EC57:1998. Considering the nature of ECG signal, we employed a non-stationary and nonlinear decomposition technique termed as improved complete ensemble empirical mode decomposition (ICEEMD). Later, higher order statistics and sample entropy measures are computed from the intrinsic mode functions (IMFs) obtained from ICEEMD on each ECG segment. Furthermore, three data level pre-processing techniques are performed on the extracted feature set, to balance the distribution of heartbeat classes. Finally, these features fed to AdaBoost ensemble classifier for discriminating the heartbeats. Simulation results show that the proposed method provides a better solution to the class imbalance problem in heartbeat classification.

Entrepreneurship, gender gap and developing economies: the case of post-apartheid South Africa

This empirical study examines the extent to which risk aversion and entrepreneurial ability influence an individual's decision to enter into entrepreneurship. Precisely, it delineates the gender gap in self-employment, nascent and high growth entrepreneurship. In doing so, it utilizes the Global Entrepreneurship Monitor South Africa databases containing 19,469 usable cases sampled between 2009 and 2014. The study adopts a quantitative approach and it applies an estimator in the form of a probit model and a non-linear decomposition technique to test established hypotheses. The results indicate that lower levels of knowledge and skills among women explain a substantial part of the gender gap. Thus, the gender gap in nascent and high growth entrepreneurship would be reduced if women had similar characteristics as men. Also, their entry rates into self-employment would be high and there would be no gender difference. This shifts the emphasis from the significance of risk aversion for local entrepreneurship to accentuate the importance of entrepreneurial-specific skills required to successfully engage in entrepreneurship.

Decomposing Gender Disparity in Total Physical Activity among Iranian Adults.

OBJECTIVESWhile gender differences in physical activity (PA) have been reported, their origin is not well understood. The present study aimed to identify factors contributing to this disparity.METHODSThis was a population-based cross-sectional study based on the 2011 surveillance of risk factors of non-communicable diseases that was conducted among Iranian adults. Multi-staged sampling was performed to obtain the required study sample. The primary outcome was gender differences in the prevalence of sufficient physical activity (SPA). Total physical activity (TPA) was calculated as metabolic equivalents (MET) per minute during a typical week, as recommended by the World Health Organization. On this basis, achieving 600 MET-min/wk or more was defined as SPA. The nonlinear Blinder-Oaxaca decomposition technique was used to explain the disparity.RESULTSThe predicted gap was 19.50%. About one-third of the gap was due to differences in the level of observable covariates. Among them, work status contributed the most (29.61%). A substantial portion of the gap remained unexplained by such differences, of which about 40.41% was related to unobservable variables. The differential effects of standard of living, ethnicity, and smoking status made the largest contribution, accounting for 37.36, 35.47, and 28.50%, respectively.CONCLUSIONSInterventions to reduce the gender gap in PA should focus on increasing TPA among housewives and women with chronic diseases, as well as those with a higher standard of living. In addition, it is essential to explore the impact of ethnicity and smoking status on women’s TPA in order to promote health.

Caste-ethnic disparity in vaccine use among 0- to 5-year-old children in Nepal: a decomposition analysis.

To examine what factors, if any, could account for caste/ethnic disparities in vaccine utilization between upper caste and indigenous, and upper caste and low caste (Dalit) groups in Nepal. Using the Institute of Medicine (IOM) approach to calculate disparity, we estimate the disparity between upper caste and indigenous, and between upper caste and Dalit 0- to 5-year-old children in Nepal. Then we use Fairlie's non-linear regression-based decomposition technique to account for those disparities in immunization. Using nationally representative data (the Nepal Living Standard Survey II and III), we construct a pooled cross-sectional series and calculate the disparity between upper caste and indigenous (8.047%), and between upper caste and Dalit (7.215%). Both of these comparisons are significant at less than 1% significance level. Decomposition results show that a major portion of the disparity can be attributed to the difference in access to immunization services, followed by differences in household income and parental education. Nepal's national vaccine programs should increase focus on reaching geographically distant populations, and continue to develop vaccination-related education efforts.

Work Environments and Labor Markets

Purpose: Knowledge about principals’ leadership roles in charter schools’ success has become more important as the number of charter schools increases and as we have learned more about the influence of principal leadership on school effectiveness. To contribute to the limited empirical literature on the principal labor market, this study explores the reasons for the disparity of turnover rates between charter school principals and their counterparts in traditional public schools (TPSs). It focuses on the differential distributions of observable factors, including principal characteristics, principal leadership practices, school contexts, and working conditions. It also examines how the associations between these observables and the likelihood of principal turnover differ between these two types of schools. Research Methods/Approach: This study uses data on a nationally representative sample of principals from the Schools and Staffing Survey in the 2007-2008 school year and its following-year Principal Follow-up Survey. The main analytic strategies include logit models and the Fairlie nonlinear decomposition technique. Findings: A statistically significant difference in charter–TPS principal turnover rates was confirmed. The explanatory variables collectively explained about 49% of the charter–TPS turnover gap: Principal characteristics explained 3%, principal leadership quality explained 4%, school contexts explained 2%, and working conditions explained 28%. Moreover, relative to TPS peers, three factors have stronger associations with the likelihood of charter school principal turnover: principal leadership quality, barriers to the dismissal of poor-performing or incompetent teachers, and salary. Implications for Research and Practice: This is one of first few studies that empirically explore the charter school principal workforce and labor market movements. Findings are practically informative for retaining principals in charter schools.

How much does broadband infrastructure matter? Decomposing the metro–non-metro adoption gap with the help of the National Broadband Map

Although overall residential broadband adoption rates have increased dramatically over the past decade, the metropolitan–non-metropolitan gap has been consistent at 12–13 percentage points. Policy prescriptions to address this problem have focused on either increasing broadband supply (typically via funding for infrastructure) or demand (such as educational efforts about why broadband is useful) in rural areas. However, the appropriate programmatic mix remains an open question, since little empirical analysis has actually assessed the degree to which a lack of infrastructure is responsible for this ‘digital divide.’ In this article, information on broadband adoption from 2011 Current Population Survey data are meshed with detailed broadband infrastructure data from the newly available National Broadband Map. A non-linear decomposition technique is used to demonstrate that existing metro–non-metro differences in infrastructure availability comprised approximately 38% of the 2011 broadband adoption gap. This same technique also shows that 52% of the gap is due to differences in characteristics such as education and income, suggesting that future policies and programs addressing this issue should include a heavily-weighted demand component.

Why Did Self-Employment Increase so Strongly in Germany?

Germany experienced a unique rise in the level of self-employment in the first two decades following unification. Applying the non-linear Blinder-Oaxaca decomposition technique, we find that the main factors driving these changes in the overall level of self-employment are demographic developments, the shift towards service sector employment, and a larger share of population holding a tertiary degree. While these factors explain most of the development in self-employment with employees and the overall level of self-employment in West Germany, their explanatory power is much lower for the stronger increase of solo self-employment and of self-employment in former socialist East Germany.

Parkinson's Disease and Home Healthcare Use and Expenditures among Elderly Medicare Beneficiaries.

This study estimated excess home healthcare use and expenditures among elderly Medicare beneficiaries (age ≥ 65 years) with Parkinson's disease (PD) compared to those without PD and analyzed the extent to which predisposing, enabling, need factors, personal health choice, and external environment contribute to the excess home healthcare use and expenditures among individuals with PD. A retrospective, observational, cohort study design using Medicare 5% sample claims for years 2006-2007 was used for this study. Logistic regressions and Ordinary Least Squares regressions were used to assess the association of PD with home health use and expenditures, respectively. Postregression nonlinear and linear decomposition techniques were used to understand the extent to which differences in home healthcare use and expenditures among elderly Medicare beneficiaries with and without PD can be explained by individual-level factors. Elderly Medicare beneficiaries with PD had higher home health use and expenditures compared to those without PD. 27.5% and 18% of the gap in home health use and expenditures, respectively, were explained by differences in characteristics between the PD and no PD groups. A large portion of the differences in home healthcare use and expenditures remained unexplained.

Why Did Self-Employment Increase so Strongly in Germany?

Germany experienced a unique rise in the level of self-employment in the first two decades following unification. Applying the non-linear Blinder-Oaxaca decomposition technique, we find that the main factors driving these changes in the overall level of self-employment are demographic developments, the shift towards service sector employment, and a larger share of population holding a tertiary degree. While these factors explain most of the development in self-employment with employees and the overall level of self-employment in West Germany, their explanatory power is much lower for the stronger increase of solo self-employment and of self-employment in former socialist East Germany.

Wave separation: Application for arrival time detection in ultrasonic signals

A method to detect and accurately measure the arrival time of wave packets in ultrasonic signals using a nonlinear decomposition technique is presented. We specifically address the problem of extracting events that are not well separated in the time, space and frequency domains. Analysis of complex ultrasonic signals, even those composed of poorly separated echoes, provided exceptional estimates of the desired time of arrival, from the media under investigation.

Why are Women Underrepresented in Elite Colleges and Universities? A Non-Linear Decomposition Analysis

The emerging female advantage in education has received considerable attention in the popular media and recent research. We examine a persistent exception to this trend: women’s underrepresentation in America’s most competitive colleges and universities. Using nationally generalizable data spanning four decades, we evaluate evidence for three possible explanations. First, we analyze whether men’s academic profiles more closely match the admissions preferences of elite institutions. Next, we consider organizational preferences for male applicants. Finally, we test whether women self-select out of elite institutions through their application choices. Using Blinder–Oaxaca non-linear decomposition techniques and multinomial logistic regression, we find that men’s advantage in standardized test scores best explains the enrollment gap. Our analyses thus suggest that the gender enrollment gap in elite colleges and universities is a matter of access, not student choice. We discuss the implications of these results for educational equity and college admissions.

How Much Does Broadband Infrastructure Matter? Decomposing the Metro Non-Metro Adoption Gap with the Help of the National Broadband Map

Although overall residential broadband adoption rates have increased dramatically over the past decade, the metropolitan – non-metropolitan gap has been consistent at around 13 percentage points. Policy prescriptions to address this problem have focused on either increasing broadband supply (typically via funding for infrastructure) or demand (such as educational efforts about why broadband is useful) in rural areas. However, the appropriate programmatic mix remains an open question, since little empirical analysis has actually assessed the degree to which a lack of infrastructure is responsible for this ‘digital divide.’ In this article, information on broadband adoption from 2010 Current Population Survey data are meshed with detailed broadband infrastructure data from the newly available National Broadband Map. A non-linear decomposition technique is used to demonstrate that existing metro – non-metro differences in infrastructure availability comprise approximately 35% of the current broadband adoption gap.

Female Overrepresentation in Public and Nonprofit Sector Jobs

Women are overrepresented in the public and nonprofit sectors. This article aims to bring to light the reasons behind this phenomenon. The originality of the employer–employee matched data used allows us to consider a large scope of potential reasons. Using a nonlinear decomposition technique, we find that in addition to the well-known occupational segregation effect, the overrepresentation of women in the public and nonprofit sectors is associated with two common factors: greater offerings of family-friendly practices and higher attraction of men for certain fringe benefits that are more frequently provided by the for-profit sector. Sector-specific factors also exist. The higher wage advantage obtained by women compared with men working in the public sector rather than in the for-profit sector contributes to the feminization of the public sector. Similarly, the overrepresentation of women in the nonprofit sector is linked to greater access to part-time jobs and shorter workweeks there.

Explaining the increase in breastfeeding at hospital discharge in Ireland, 2004–2010

Despite a steady increase in the rate of breastfeeding in Ireland over the period 2004-2010 (from 46.8% in 2004 to 55.7% in 2010), Irish rates of breastfeeding are still low by international standards. Over this period, the characteristics of mothers and babies changed, with mothers increasingly older, having fewer children and increasingly born outside the Republic of Ireland. The purpose of this paper is to understand the extent to which changing maternal and birth characteristics explained the increase in the breastfeeding rate in Ireland over the period 2004-2010. We apply non-linear decomposition techniques to micro-data from the 2004-2010 Irish National Perinatal Reporting System to examine this issue. The technique allows us to quantify the extent to which the increase in the breastfeeding rate over the period 2004-2010 is due to changing maternal and birth characteristics. We find that between 55 and 74% of the increase over the period can be explained by changing characteristics, with the increasing share of mothers from Eastern Europe, and increasing maternal age the most important contributors. These findings suggest that the existing policy initiatives have been relatively ineffective in increasing breastfeeding rates in Ireland, i.e. most of the observed increase occurred simply because the characteristics of mothers were changing in ways that made them increasingly likely to breastfeed.

Emergency department services use among immigrant and non-immigrant groups in the United States.

Immigrants have disproportionate lack of access to healthcare and insurance. Emergency departments could serve as a healthcare substitute and increased demand can negatively affect the US emergency services system. Medical Expenditures Panel Survey (2000-2008) data was modeled to compare emergency departments (ED) use between non-citizens, foreign-born (naturalized), and US-born citizens. Group differences were assessed using non-linear decomposition techniques. Non-citizens were less likely to use ED services (8.7 %) compared to naturalized immigrants (10.6 %) and US-born Americans (14.7 %). Differences in ED use persisted after adjusting for the Behavioral Model covariates. Healthcare need and insurance partially explained the differences in ED use between the groups. Between 2000 and 2008 non-citizen immigrants used markedly less ED services compared to US citizens, regardless of their nation of origin. We also found that demographic and healthcare need profiles contributed to the divergence in use patterns between groups. A less restrictive healthcare policy environment can potentially contribute to lower population disease burden and greater efficiencies in the US health care system.

Understanding Gender Differences in Statin Use among Elderly Medicare Beneficiaries

Pharmacotherapy with statins (HMG-CoA reductase inhibitors) is the cornerstone for lipid management in individuals with or at risk of developing cardiovascular diseases. Although the clinical benefits of statins are established for both women and men, there is evidence of a gender difference in their use. The current study extends prior scientific research by estimating the extent to which individual-level variables may explain gender differences in statin use by using a post-regression non-linear decomposition technique. The objective of this study was to estimate the magnitude of gender differences in statin use among the elderly and examine individual-level variables that can explain the gender differences in statin use among elderly individuals with or at risk of cardiovascular diseases. A retrospective cross-sectional study design was adopted. Data were derived from the 2005 Medicare Current Beneficiary Survey (MCBS), a nationally representative survey of Medicare beneficiaries in the US. The analytic study sample consisted of community-dwelling elderly Medicare beneficiaries, aged 65 years or older, who had reported any of the following conditions: heart disease, hyperlipidaemia or diabetes mellitus, and who were alive during the observation year. Chi-square tests were used to evaluate the unadjusted associations between gender and statin use for each of the characteristics. Multivariate logistic regressions were used to evaluate the relationship between gender and statin use. A post-regression non-linear decomposition approach was used to understand individual-level variables that could explain gender differences in statin use. Among 5,508 elderly, 47.2% of the women and 55.5% of the men reported any statin use in 2005, which translates to an 8.3 percentage point difference in statin use. In the multivariate logistic regression on statin use, women were 21% less likely than men to use statins (adjusted odds ratio = 0.79; 95% CI 0.69, 0.89). Post-regression non-linear decomposition analysis revealed that of the 8.3 percentage point difference in statin use, 29.5% was explained by the individual-level variables. Lifestyle risk factors accounted for most of the explained portion of the gender difference in statin use. Among elderly Medicare beneficiaries, women were less likely than men to report any use of statins. Less than one third of the total gender difference in statin use was attributed to individual-level variables such as demographics, economic status, physical health status, depression and lifestyle risk factors. Further research is needed to identify whether provider and/or organizational-level factors can further explain the gender difference in statin use.

Diverging Racial and Ethnic Disparities in Access to Physician Care

To examine recent changes in racial and ethnic disparities in access to physician services in the United States, and investigate the economic factors driving the changes observed. Using nationally representative data on adults aged 25-64 from the 2000 and 2007 Medical Expenditure Panel Survey, we examine changes in two measures of access: whether the individual reported having a usual source of care, and whether he/she had any doctor visits during the past year. In each year, we calculate disparities in access between African Americans and Whites, and between Hispanics and Whites, applying the Institute of Medicine's definition of a disparity. Nonlinear regression decomposition techniques are then used to quantify how changes in personal characteristics, comparing 2000 and 2007, helped shape the changes observed. Large disparities in access to physician care were evident for both minority groups in 2000 and 2007. Disparities in no doctor visits during the past year diminished for African Americans, but disparities in both measures worsened sharply for Hispanics. Disparities in access to physician care are improving for African Americans in one dimension, but eroding for Hispanics in multiple dimensions. The most important contributing factors to the growing disparities between Hispanics and Whites are health insurance, education, and income differences.

Do fundraisers select charitable donors based on gender and race? Evidence from survey data

Recent studies document that people are much more likely to donate to charity and volunteer their time when they are asked to. Using household surveys of giving and volunteering in the United States conducted from 1992 to 2001, which contain questions on whether the respondent was personally asked to give or volunteer, this paper investigates the factors associated with the probability of receiving a charitable solicitation and presents substantial evidence that race and gender differences play key roles in the selection of potential donors. In particular, males, blacks, and Hispanics are less likely to be solicited compared with females and whites. Using non-linear decomposition techniques, I find that differences in observable characteristics of individuals explain most of the racial gap in the probability of being solicited for charitable causes, but they fail to explain the gender gap in the probability of being asked to volunteer. Furthermore, these results are robust to alternative specifications. I also discuss related policy implications and argue that the economic impact of selecting potential donors based on gender and race can be considerable.

The Diffusion of Internet Technologies to Rural Communities: A Portrait of Broadband Supply and Demand

Rural communities suffer from supply- and demand-side disadvantages when dealing with Internet access. Telecommunications companies are less likely to provide such communities with needed infrastructure because of their lower population densities; these communities also tend to have lower levels of factors known to influence the access decision, such as education and income.This study looks at the broadband Internet setting in Oklahoma over a 3-year period, examining the diffusion of infrastructure and access rates. A nonlinear decomposition technique allows for measurement of how characteristics such as the availability of infrastructure contribute to observed metropolitan—micropolitan or metropolitan—noncore digital divides.The results suggest that although differences in infrastructure are consistently only minor contributors to the divide, their importance has increased as knowledge about the Internet has diffused.