Data and Methods. Data Our data are drawn from the Child Support Noncustodial Parent Employment Demonstration (CSPED) evaluation, a randomized controlled trial conducted in certain counties across eight states in the U.S. To participate, noncustodial parents (NCPs) were required to have established paternity, at least one open child support order, either current or likely future difficulty in making child support payments, and be facing employment difficulty even though they were legally and medically able to participate in gainful employment.1 All participants were at least 18 years of age and not incarcerated at the time of enrollment in the study. Baseline surveys were administered to all participants at the date of their enrollment, between October 2013 and September 2016 (Cancian, et al., 2019). Baseline surveys gathered background information about socioeconomic characteristics, relationships with children, employment status, receipt of public assistance, family of origin characteristics, and motivations for enrolling in the CSPED program. About one year after participants completed the baseline survey, a follow-up survey was conducted. CSPED also collected administrative data from each partner state which we use exclusively for child support payment-related information. 1A small percentage of participants (about 2%) did not owe current child support but were enrolled because they were anticipated to begin to owe. For simplicity, we describe the sample as all owing support. Sample We restrict our analytic sample to NCPs who responded to both a baseline and follow-up survey with CSPED. Our analysis sample consists of noncustodial parents from six of the eight participating states. NCPs (N = 853) from Texas and Iowa were excluded because of data limitations.2 Additionally, we exclude N = 466 individuals who did not identify as fathers, N = 101 NCPs without nonresident children, and N = 449 NCPs with missing data across any of our selected outcomes. We also restrict our child-level outcome data to NCPs’ biological minor children who did not reside with the NCP at both the baseline and follow-up surveys (or who were not yet born at baseline and did not reside with the NCP at follow-up) as measured by the NCP reporting spending fewer than 16 nights in the same place as the child in the past thirty days. This leaves a final analytic sample of N = 2,409. After completing an analysis drawing on data from all six states, we conducted a Wisconsin-only analysis using the same ap...
Data and Methods. This analysis uses matched couples’ data from DHS surveys in 10 sub-Saharan African countries to investigate spousal agreement (or disagreement) on a range of family planning issues. The analysis uses data from Benin, Burkina Faso, Chad, and Mali in West and Central Africa, and from Malawi, Namibia, Rwanda, Uganda, Zambia, and Zimbabwe in Eastern and Southern Africa. The surveys were conducted between 1999 and 2004. The data for women are based on women age 15-49, while the data for men are based on men aged 15-59 (with the exception of Malawi and Benin, where men are age 15-54 and 15-64, respectively). The men’s questionnaire is similar in structure to the women’s questionnaire but shorter. To the extent possible, questions and response categories in both questionnaires are worded identically to be comparable across countries. In this analysis, infecundity is measured by childbearing experience of the woman, that is, a woman is defined as infecund if she has had no births and no pregnancies in the past five years but has had a birth or pregnancy at some time, and has been married for the past five years but did not use a contraceptive method during that period. Wealth status of the household is measured using the wealth index. The wealth index is constructed from household asset data using principal components analysis (Xxxxxxxx and Xxxxxxx, 2004). Based on the first factor loading, the wealth index score divides the population into five quintiles. In this paper, “poor” refers to the bottom two quintiles, “middle” refers to the middle quintile, and “rich” refers to the top two quintiles.
Data and Methods. Data Approximately 247 million people live in Indonesia across 6,000 inhabited islands divided into 34 provinces, which are further split into 500 districts, 7,000 sub-districts, and over 80,000 villages[2]. This analysis used data from the 2012 Indonesia Family Life Survey, East (IFLS, E), a large-scale multi-topic household and community survey that was conducted in seven provinces in Eastern Indonesia [23, 24]. The 2012 survey was designed and implemented by the National Team for Acceleration of Poverty Alleviation (Xxx Nasional Percepatan Penanggulangan Kemiskinan or TNP2K), Poverty Reduction Support Facility, and Australian Aid by SurveyMETER with a structure based on a large ongoing longitudinal IFLS that has been collecting data since 1994 [24]. Sampling was completed in four stages to be representative of the Indonesian population living in eastern provinces. First, two provinces were selected with equal probability from Kalimantan and Sulawesi regions while the remaining five were selected without sampling. Xxxxxxxxxx Xxxxx, Xxxxxxxx Xxxxxxxx, Xxxx Tenggara Timur, Maluku, Maluku Utara, Papua Barat, and Puapa were the seven final provinces included in the survey. Second, 14 villages were drawn from each province without replacement and with equal probability. Third, administrative unit levels were identified; these areas consisted of approximately 100-150 households and then were further divided into smallest local area (SLS) units. One SLS group from each village was randomly selected. The fourth and final step involved listing all households within the selected SLS. From this list, a simple random sample without replacement was taken for 30 rural households or 20 urban households. A more detailed description of IFLS East survey methods are described elsewhere [23]. For each level, both household and community level cross-sectional surveys were conducted to collect information associated with health, education, and socioeconomic status. In this analysis, we focused exclusively on household level data. Outcome variables Three binary vaccination statuses were evaluated in this study: fully vaccinated, vaccinated with three doses of DTP-HepB vaccine (DTP-HepB3), and vaccinated with first dose of measles vaccine (MCV1). Each status was based on the child’s vaccination card, or if the card was not available, by caretaker recall. Fully vaccinated status was defined as the child receiving all recommended doses of vaccine. In Indonesia this includes: one ...
Data and Methods. 4.1 Data This study used 2010 data from the China Family Panel Studies (CFPS) (2010) to examine the relevant factors that influence Chinese household decision of gift-exchange. The Institute of Social Science Survey (ISSS) of Peking University launched this annual longitudinal survey in 2010 to collect individual-, family-, and community-level data in contemporary China. There have been multiple waves of surveys conducted, among which 2010, 2011, and 2012 are released. This paper used the 2010 survey results, which cover 25 out of 34 provincial-level administrative divisions that include 23 provinces, four Municipalities, five Autonomous Regions, and two Special Administrative Regions (for additional details see Appendix Table 1). Except for Shanghai and Gansu provinces, of which the share of community numbers is relatively higher than their population share in the entire Chinese population, the weights in other provinces are fairly reasonable compared to the Chinese population distribution. Applying the sample weights in the dataset can solve this issue. As for the age-structure, the median and mean of individual age are both around 45 years old; the average percentage of people over 60 years old is about 18%. This nationally representative survey involves 57,155 individuals that come from 14,960 households across China. 33.52% of the households come from urban residential communities (Xx Xxx Xxx), while 66.48% of them come from villages (Cun Xxx Xxx). The minority ethnicity consists of approximately 10% of the sample. As for occupations, 28.45% of the adults are employed. To analyze the potential impact of occupation on gift-exchange decisions, individuals that were marked as household representations were also categorized into ten industries (for additional details see Appendix Table2). There are several reasons for which only the 2010 dataset was used. First of all, compared to the datasets in other years, it contains a wealth of information about household level financial decision-making, particularly the part that is relevant to the present study on gift-exchange. Also, it provides enough demographic and other relevant information to be controlled in regression analyses. Moreover, the present study does not focus on the over-time changes of behaviors; therefore using one of the panel datasets will not only suffice the purpose of the study, but also avoid the issues that exist potentially in panel datasets.
Data and Methods. With an untested institution involving a complex web of actors and a potentially tangled causal network, qualitative within-case analysis can yield very meaningful results (Xxxxxxx 2006: 251). This is achieved by collecting a wide array of relevant data for each point of analysis, including bylaws, policy changes, news articles, first-hand accounts, secondary sources, participation data, government papers and announcements, and Internet statistics. These sources are used to piece together ICANN’s entire history from multiple perspectives. Consequently, the predictions of each different hypothesis can be meaningfully compared to the historical record at each point of analysis. The units of analysis are turning points in ICANN’s evolution: its formation, its involvement with the United Nations and other organizations during the World Summit for the Information Society, and its shift from U.S. oversight to an international multistakeholder model of governance. The independent variables for this study vary for each hypothesis. For Hypothesis 1, the independent variables are the power and preferences of states. For Hypothesis 2, the independent variables are the power and preferences of corporations. For Hypothesis 3, the independent variables are the power and preferences of individuals and groups in civil society. For Hypothesis 4, the independent variable is past institutional rules. The dependent variable for all hypotheses is institutional form. Because this is not a quantitative analysis, these hypotheses are not proven with statistical significance. The in- depth, theory-driven historical evaluation of each hypothesis does, however, provide strong indicators in support or refutation of each perspective, and is a solid foundation for future quantitative work. Here, I describe the different sources of evidence and how they are used in my analysis. INTERVIEW AND OBSERVATION Much of my initial work was guided by anecdotal evidence derived from observation of and conversations with ICANN participants. On July 8, 2015 I observed and wrote a record of the hearing titled “Internet Governance Progress After 53” in which Xxxx Xxxxxxx, ICANN’s CEO at the time, and NTIA Administrator Xxxxxxxx Xxxxxxxxxx testified before the House Committee on Energy and Commerce Subcommittee on Communications and Technology about ICANN’s progress transitioning IANA authority away from U.S. oversight. I was struck by the intensity of the politics surrounding such a seemingly technica...
Data and Methods. DESIGN OVERVIEW In order to test these hypotheses, I conducted a vignette experiment (see Appendix 1). The vignettes were distributed to a nationally representative sample of United States adults via Time-Sharing Experiments for the Social Sciences (TESS) and GfK (formerly Knowledge Networks). 1013 respondents were asked to review and rate a profile of one job candidate. The job candidates in the vignettes are all parents, coupled, of the same race (all implied white), and equally qualified for the position, but vary in terms of gender (male/female), sexual orientation (heterosexual/homosexual), and caregiving position (not indicated/primary/secondary). Male and female respondents were randomly assigned to one of the twelve conditions of the 2x2x3 experimental design. While I make no prediction about respondent gender on job candidate ratings, this design allows for investigating if and how male and female evaluators respond differently to these manipulations. This design allows me to tackle an issue that still plagues the U.S. workplace but from a novel perspective. While Xxxxxxx et al. (2007) focus on comparing parents to non- parents and looking at the intersection with race of worker, I focus solely on white, partnered parents to test how worker evaluations are impacted by job candidate gender, sexual orientation, and caregiving position and their interactions. Further, while there is a growing literature on gay and lesbian couples in the workplace, studies typically employ census data. This approach is very helpful for analyzing trends and shifts in earnings over time and between groups, but does not offer the benefits of an experimental design, which allows for control of potentially confounding variables. Determining whether penalties and premiums arise from differences in choices or expectations around gender performance is possible in an experimental design. In such a design, all parents can be presented as having made the same choices, save for the experimentally manipulated ones. Further, while it is increasingly relevant to today’s workplace, there is scant research at this point in time on how and whether gay and lesbian parents face workplace penalties and premiums associated with parenthood. To isolate these particular main effects and interactions of gender, sexual orientation, and caregiving, I hold other variables constant. The variables that were held constant for each vignette include race (which is manipulated as white by using U.S. bi...
Data and Methods. Given that infrastructure projects have to adapt to local conditions, there is a significant variation in the cost per acre-foot (af) of water for different projects, even within similar categories. To account for this variation, we obtained information for as many projects as we could for each alternative, then we determined unit costs for each project, and finally we calculated the ranges and distribution of costs for each category. To facilitate comparisons, all results are in unit cost terms (dollars per acre-foot per year in 2018 dollars).
Data and Methods. The empirical analysis will shed light on the question of whether the strength and/or direction of party stance with respect to the issue of European integration is a determinant of party electoral performance. The study concerns itself with national, legislative elections in 15 original EU member states dating from 1996-2015, depending on the timing of elections. The study proceeds to conduct linear regressions according to the five hypotheses, listed above. Hypotheses one, two, and three will be tested in two separate time periods in order to assess if the relationship in question is at all different when comparing elections held prior to 2008, and elections held in years 2008 and after. Time period A will concern elections up to 2007, time period B will concern elections from 2008 and beyond. Note that cases included in hypothesis three are limited to “mainstream” parties, as the hypothesis is concerned exclusively with mainstream party performance.4 Hypotheses four and five will be tested with separate case groups. Hypothesis four will be split according to party radicalism, running a linear regression using exclusively parties coded as “radical parties,” and again with parties coded as “mainstream parties.” Hypothesis five is 4 The study considers “mainstream parties” to be those coded as neither “radical left” nor “radical right” by the CHES dataset. split according to party ideology: Cases are divided as “Left-to-Center,” and as “Center-to- Right.”5
Data and Methods. I conducted a cross-sectional quantitative survey with a qualitative supplement through: 120 surveys on Gichagi residents' awareness, attitudes and utilization of community health services; and a focus group discussion with 10 Gichagi community health workers assessing their views of the program.
Data and Methods. In this study we estimate associations between CDC’s Social Vulnerability Index and daily mortality counts in census tracts that were affected by the April 2011 tornado outbreak under the following hypothesis: the SVI modifies the association between tornadoes and mortality in the April 2011 tornado outbreak. The tornado event affected census tracts in five states: Arkansas, Mississippi, Alabama, Tennessee, and Georgia. Of the five states, all but Tennessee elected to participate in the study. Datasets of mortality counts by either census tract or ZIP code (depending on the format of individual state surveillance systems) were assembled for the other four states. These datasets contain the census tract-specific (or ZIP code-specific) overall mortalities (as opposed to direct event-specific fatalities) on the date of the tornado event as well as the mortalities on the same day of the week from the prior week (baseline mortality). For all mortality data provided by ZIP code (Arkansas and Alabama), HUD-USPS crosswalk files were used to convert mortality from ZIP codes to census tracts. HUD-USPS crosswalk files were created by the U.S. Department of Housing and Urban Development (HUD) to provide a data allocation method between disparate geographic units that is based on residential addresses rather than area or population (DHUD, 2015). This conversion produces non-integer mortality estimations at baseline and on the date of the event for each census tract. The estimations were rounded to the nearest whole integer values to be used as inputs in a Poisson regression analysis. All spatial analysis and geographic data manipulation were completed in ArcGIS 10.2.2. Further data cleaning and statistical analysis were performed using SAS 9.4. Tornado tracks for the April 25-28 storm were obtained from the National Weather Service (NWS) (NOAA, 2016). The study area was defined as any census tract in the four participating states that was intersected by an NWS –designated tornado track. Overall 368 census tracts were included in the study. Table 2.1 summarizes collected mortality data by tornado magnitude.
