Covariates Sample Clauses

Covariates. Using Directed Acyclic Graphs (DAGs) and the previous literature, potential confounders were identified as mother’s age at conception of the index child, father’s age at conception of the index child, mother’s race, mother’s smoking status during pregnancy, mother’s education, mother’s marital status, mother’s birthplace, mother’s desire to get pregnant, procedures used to conceive the index child, prenatal vitamin use in the first month of pregnancy of the index child, family home ownership, child’s sex, and preterm birth. Information on these potential confounders was collected during the primary caretaker’s interview.

Covariates. Age, number of prenatal care visits, education, and race/ethnicity were included as potential effect modifiers or confounders based on information in the literature. Medicaid enrollment was used as a proxy measure for income. In Georgia, pregnant women are eligible for Medicaid coverage if they fall below 220% of the federal poverty level (9). All covariates are recorded on the birth certificate, and are self-reported. The Georgia birth certificate has payment categories of: Medicaid managed care, Champis, Medicaid, private insurance, other government assistance, self-pay/uninsured, and unknown. Unknown payment observations were set to missing and payment options were simplified into non-Medicaid or Medicaid, with Medicaid including women who paid using Medicaid or Medicaid managed care. Payment information was missing for 8.4% (n = 280) of the sample. Maternal age is recorded on the birth certificate as a continuous variable. Maternal age was separated into six categories for Tables 1 and 2: 18 – 19, 20 – 24, 25 – 29, 30 – 34, 35 – 39, and 40+. Maternal age was used as a continuous variable in the logistic regression models. Age information was missing for 6.4% (n = 214) of the sample. Prenatal care adequacy is measured on the birth certificate using the Kotelchuck Index, which is calculated based on the gestational age when prenatal care began and on the gestational age at delivery. The Kotelchuck Index is based on the assumption that the earlier prenatal care begins the better, and the index has four different levels: inadequate (received less than 50% of expected prenatal visits), intermediate (50% - 79%), adequate (80% - 109%), and adequate plus (110% or more) (22). Unknown prenatal care information was set to missing. Prenatal care information was missing for 28.1% (n = 935) of the sample. Maternal education is self-reported as the mother’s highest grade completed at the time of birth. Unknown education information was set to missing, and education was divided into four categories: less than 9th grade, 9th through 11th grade, High School/GED, and some college or higher. Education information was missing for 3.4% (n = 113) of the sample. Maternal race is recorded on the birth certificate as six categories: American Indian or Alaska native, Asian, black or African-American, multiracial, native Hawaiian or other, and white. Maternal ethnicity is recorded as Hispanic or non-Hispanic. For the purposes of this study, race and ethnicity were combined into a...

Covariates. Both confounding and effect modification, or interaction of a confounder and the exposure of interest, are suspected. Confounders that will be assessed are as follows: education, time in the U.S., region of birth, employment status, U.S. region of residence, gender, health insurance, marital status, age, and smoking status. Interaction is suspected for gender and health insurance. Education is based on a survey question that asks respondents how many years of education they have received in total, whether in or outside of the U.S. This indicator will be kept linear and measured in years. Time in the U.S. is a self-constructed variable that is linear and measured in years. This variable measures the length of time between the year one first migrated to the U.S. and the year that one was surveyed. While many immigrants in the survey traveled outside of the U.S. one or many times, the U.S. is considered as an exposure that impacts health, and these subsequent trips are not taken out of the amount of time one spends in the U.S. The region of birth variable is split into four disparate regions: 1) Europe and Central Asia, North America [Non-Latin], 2) Latin America and the Caribbean 3) Africa and the Middle East 4) East Asia, South Asia, the Pacific, and Oceania. This variable is derived from the survey question that asks respondents their country of birth. In modeling, the group including Europe and North America will be used as the reference group, as it is suspected to be most similar to the U.S. Because of limitations in calculating household income and socioeconomic status, one’s employment type will be used as a proxy for socioeconomic status. The survey asks respondents if they are employed full-time, part-time, or unemployed. If they are unemployed, the survey further asks if one is a homemaker, laid off, retired, disabled, or under some other an unspecified employment circumstance. For this analysis, employment is broken into the following categories: 1) employed 2) unemployed and seeking work 3) unemployed due to being laid off, disabled, on leave, or retired 4) unemployed homemakers 5) unemployed under other circumstances. The employed group is the largest and will be used as the referent group in modeling. One’s U.S. region of residence is based on a pre-populated INS variable that provided the state or region where one’s green card was sent. While it’s possible that someone moved between getting their green card and being interviewed, it’s the only...

Covariates. We included the following socioeconomic and demographic variables to adjust for confounding of the relationship between the schooling variables and prior year IPV: residence (urban versus rural), age measured continuously in years, relationship status (married, living together, previously partnered), work status (worked within the past year versus not working), number of children living at home measured continuously, witnessing father beat mother (yes, no), and a continuous score for household wealth. The DHS household wealth index is the standardized score (mean 0, standard deviation of 1) derived from the first principal component of a principal components analysis of recoded items measuring whether or not the household had a specified set of assets and amenities (▇▇▇▇▇▇▇▇, 2004). Age of partner was not asked of all women in the previously partnered category and therefore was not included in the analyses. All descriptive and inferential analyses were conducted using the PASW version 18.0 statistical package for PC (PASW Statistics 18, 2009) and multivariate logistic regression was conducted using SAS version 9.2 (SAS Institute, 2009). Univariate analyses were conducted for all covariates, outcomes, and variables from which analytic covariates were derived to assess their completeness, distributions, and relative frequencies. Bivariate associations of all covariates were estimated to assess potential co- linearity among these variables. Variables statistically associated with our outcome in the bivariate analysis were included in the multivariate model. Because the outcome variable, prior year IPV (y), for each individual is binary, multivariate logistic regression analysis was used with strata at the regional level and clustering at the primary sampling unit (PSU) to estimate the adjusted associations of explanatory variables on the log odds of having experienced IPV in the prior year. The SAS survey logistic procedure fits linear logistic regression models for survey data using the maximum likelihood method (Binder, 1983; ▇▇▇▇▇▇▇▇▇, 1989). The procedure incorporates complex survey sample designs, including designs with stratification, clustering, and unequal weighting for statistical inferences (SAS Institute I, 2004). The model used to estimate the adjusted effects of covariates on the adjusted log odds of having experienced IPV in the prior year was:

Covariates. Through previous research findings, sex, marital status, education, income, age at time of survey, age at diabetes diagnosis, health insurance status, and race/ethnicity were included as covariates of interest in the initial full model. Race/ethnicity was categorized into three groups, Non-Hispanic Black, Other (including American Indian or Alaskan Native, Non-Hispanic Asian, Native Hawaiian or Pacific Islander, Hispanic, Non- Hispanic Multiracial, and Hispanic), and Non-Hispanic White (reference). Education was categorized into three groups: less than high school (reference), high school graduate, and some college or more. Additionally, income level was grouped into four categories: <25k (reference), 25k-35k, 35k-50k, and 50k+. Lastly, age was categorized into 18-44, 45-64, and 65+ years (reference).

Covariates. Health-relatedcovariates:Multimorbiditywas evaluatedin two ways. The firstof them was by means of self-reportof medicaldiagnosisof some conditionsassessedin the individualquestionnaire:arthritis,stroke,angina,diabetes,chronic lung disease,asthma, hypertensionor cataracts.The second was by applyingfour additionalalgorithmsto determine Statistical analysis Ethics statement Results

Covariates. A robust set of relevant covariates was constructed for inclusion in our analytic models using baseline survey data. Covariates were selected for their potential relation to our seven outcomes of interest including noncustodial fathers’ demographic characteristics (i.e., age, race, and ethnicity), marital and cohabitation status, and socioeconomic status (i.e., highest level of education and earnings in the 30 days before baseline survey administration). Additional covariates include incidence of major severe depression, a father’s relationship quality with and involvement of their own biological father,4 NCPs having children with more than one partner, gatekeeping by any custodial parent or a custodial parent’s family or friends, and NCPs residence with a non-biological child of a romantic partner. Child-level covariates include the number of nonresidential minor children a father has, average nonresident child age, average relationship quality with nonresident children at baseline,5 and nonresident child sex. Finally, the CSPED treatment group and a father’s state of residence are included. The full set of covariates is listed in Table 1. To assess the associations between incarceration among NCPs and their relationships with and financial investments in their nonresident children, we conducted a robust set of multivariate regression analyses. Our main estimation strategies are ordinary least squares and logistic regression modeling. All regression models are weighted with CSPED survey weights. 4This variable is coded as 0=not at all involved, 1=somewhat involved and fair or poor relationship, 2=somewhat involved and excellent/very good/good relationship, 3=very involved and fair/poor relationship, and 4=very involved and excellent/very good/good relationship. We chose this approach to be consistent with previous literature that has used the same strategy (e.g., ▇▇▇▇▇▇▇ et al., 2018; ▇▇▇ et al., 2023). 5Relationship quality with children is both an outcome in its own right and potentially very important to the other outcomes of interest, so we include it as a covariate in those models (but exclude it from the models where relationship quality is the outcome). Our analytic approach attempts to quantify the possible associations between different measures of a father’s history of incarceration and a variety of relational and financial outcomes. Our analysis focuses on the relationship between incarceration and our outcomes of interest, controlling for a va...

Covariates. Age Gender RIDAGEYR RIAGENDR Best age in years of the sample person at time of HH screening. Individuals 85 and over are topcoded at 85 years of age. Gender of the sample person Categorized into pre- school and primary school versus teen and college age Coded the same as 2 levels 2 levels 0-12 years / 13-19 years Male / Female Race/Ethnicity RIDRETH1 Reported race or ethnicity NHANES Coded the same as 5 levels Non-Hispanic White / Non- Body mass index RIDAGEMN Best age in months at date of NHANES Variables entered into 3 levels Hispanic Black / Mexican American / Other Hispanic / Other Race - Including Multi Racial <5th percentile / ≥5th RIAGENDR screening for individuals under 85 years of age. Gender of the sample person CDC's SAS macro for children's BMI percentile but <85th percentile / ≥85th percentile BMXHT Standing Height (cm) Poverty income BMXWT BMXHEAD INDFMPIR Weight (kg) Head Circumference (cm) Poverty income ratio (PIR) - a PIR was assigned 3 3 levels ≤1 / >1 but less than 5 / ≥5 ratio Attached Home Age of Home Presence of HOQ011 HOQ040 HOQ240 ratio of family income to poverty threshold Is your home . . .{Mobile home or trailer / One family house detached / One family house attached / Apartment / Dormitory}? When was this {mobile home/house/building} originally built? In the past 12 months, have categories for poor, medium and high Categorized into houses that are standalone residences and those they share walls, utilities, etc. Original 6 levels were dichotomized If subjects answered yes 2 levels 2 levels 2 levels Attached / Detached Before 1959 / 1960 or later Yes / No Cockroaches Presence of HOQ230 you seen any cockroaches in your home? In the past 12 months, has to this question they were considered to have cockroaches in their home If subjects answered yes 2 levels Yes / No Mold/Mildew your home had a mildew odor or musty smell? to this question they were considered to have mold/mildew in their home

Covariates. Age was included as a continuous variable. Other categorical sociodemographic variables include gender (male, female), marital status (recatagorized into a binary “married/not married” variable), urbinicity (counties with a USDA Rural Urban designation of metro were recategorized as “urban”, counties with a nonmetro designation were recategorized as “rural”), education (less than high school, high school graduate, some college, college graduate or more), race (Hispanic, Non-Hispanic White, African American or Black, and Other), income category ($20,000 to < $35,000, $35,000 to < $50,000, $50,000 to < $75,000, $75,000 or More), and having a regular healthcare provider (yes, no) All analysis was conducted in SAS-callable SUDAAN, version 11.0.1, to account for the multistage sampling design and sample weights. A full-sample weight was used to calculate population and subpopulation estimates. 50 replicate weights, calculated using the delete one jackknife (JK1) replication method, were used to calculate standard errors. We conducted our analysis in two phases - first, an analysis of the entire sample and second, an analysis stratified by where respondents first search for health information (internet or other). Descriptive statistics were examined for all outcome, predictive, and demographic variables. Bivariate analysis, including chi square tests and t-tests, were conducted to determine if there were any significant independent associations between the three outcome variables and predictor variables. Weighted logistic regression was conducted separately for each outcome variable with each variable measuring internet health activity. Unadjusted analysis was conducted to generate odds ratios (OR) and 95% confidence intervals (95% CI). Adjusted odds ratios (aOR) with 95% CI were calculated by including sociodemographic covariates in the model to control for any confounding by sociodemographic characteristics. All analyses were conducted at a significance level of alpha = 0.05.