================================================================================ README: Replication Files "Financial Strain Among Mexican-American Adult Children Caregivers of the Oldest-Old" Anna Bokun, Jacqueline Angel, Flavia Cristina Drumond Andrade, Sunshine Rote, Philip Cantu Demographic Research, Submission #7124 ================================================================================ Last updated: April 30, 2026 Software: R (developed in RStudio "Cranberry Hibiscus") -------------------------------------------------------------------------------- OVERVIEW -------------------------------------------------------------------------------- This archive contains all programs and data necessary to replicate the analyses reported in the paper. The study uses data from the Hispanic Established Populations for the Epidemiologic Study of the Elderly (HEPESE), Wave 7 (2010-2011), and supplementary data from the American Community Survey (ACS) via IPUMS USA. Scripts should be run in the order listed below. -------------------------------------------------------------------------------- DATA SOURCES -------------------------------------------------------------------------------- 1. HEPESE Wave 7 (Primary data source) - Source: Inter-university Consortium for Political and Social Research (ICPSR) Series 546, Study 36537 https://www.icpsr.umich.edu/web/NACDA/series/546 - Files used: 36537-0001-Data.rda (care recipients, Wave 7) 36537-0002-Data.rda (caregivers/informants, Wave 7) - Access: Publicly available; free registration with ICPSR required - License: Please adhere to ICPSR terms of use - Records used: N = 659 adult child caregivers (sons, daughters, children-in-law) linked to their oldest-old care recipients (age 80+) - Key variables: caregiver demographics, co-residence, financial strain items (TT57I_1 through TT57SPECI), caregiving intensity (ADL/IADL hours), care recipient health, dementia status, insurance, and income 2. IPUMS USA ACS (Supplementary data for Table A2 external validity check) - Citation: Steven Ruggles, Sarah Flood, Matthew Sobek, Daniel Backman, Grace Cooper, Julia A. Rivera Drew, Stephanie Richards, Renae Rodgers, Jonathan Schroeder, and Kari C.W. Williams. IPUMS USA: Version 16.0 [dataset]. Minneapolis, MN: IPUMS, 2025. https://doi.org/10.18128/D010.V16.0 - Source: IPUMS USA, Minnesota Population Center https://usa.ipums.org - Access: Free registration with IPUMS required - License: Please adhere to IPUMS terms of use - Extract file: 2010-2011, restricted to AZ, CA, CO, NM, TX - Records used: N = 2,696 Mexican-origin adult children (18+) co-residing with parents (80+); N = 6,038 non-Hispanic White adult children (18+) co-residing with parents (80+) - Key variables: AGE, SEX, HISPAN, RACHSING, EDUC, HHINCOME, MARST, MOMLOC, POPLOC, NUMPREC, LANGUAGE, PERWT, CLUSTER, STRATA -------------------------------------------------------------------------------- DATA FILES (included in /data/) -------------------------------------------------------------------------------- ACC.csv Analytic sample of adult child caregivers (N = 659), merged from HEPESE Wave 7 caregiver and care recipient files, with all derived measures generated in 01_acc_clean.R. ACC_small.csv Subset of ACC restricted to variables used in the imputation model (see impute_acc.R). ACC_imputed.csv First completed dataset from the MICE imputation (imputation #1 of 5), containing only the imputed variables. ACC_full_imputed.csv Full analytic dataset (N = 659) with imputed values merged back in. This is the primary dataset used for all descriptive tables and regression analyses. mice_method_assignment.csv Lookup table documenting the imputation method assigned to each variable (logreg, polyreg, or pmm) for transparency and reproducibility. -------------------------------------------------------------------------------- CODE FILES (in /code/, run in this order) -------------------------------------------------------------------------------- 01_acc_clean.R Imports HEPESE Wave 7 caregiver (36537-0002) and care recipient (36537-0001) files, merges them by household ID (Q_NO), restricts to adult child caregivers (N = 659), and constructs the dependent variable (cg_finc_strain), primary independent variable (cg_coreside), and robustness measures (cg_finc_strain2, finc_strain_level). Output: ACC.csv impute_acc.R Checks missingness across analytic variables, subsets to variables with <=10% missing and runs Multiple Imputation by Chained Equations (MICE, m=5, seed=123) using smart method assignment (logreg/polyreg/pmm). Merges imputed values back into the full analytic dataset. Includes convergence and density plot diagnostics. Input: ACC.csv Output: ACC_small.csv, ACC_imputed.csv, ACC_full_imputed.csv, mice_method_assignment.csv table1_acc_imputed.R Produces Table 1: caregiver demographic, health, and caregiving characteristics stratified by financial strain status, with chi-square and ANOVA p-values. Saves output to table1_acc.docx. Input: ACC_full_imputed.csv Output: table1_acc.docx table2_acc_econ_imputed.R Produces Table 2: caregiver financial strain components and economic characteristics stratified by co-residence status, with chi-square p-values. Input: ACC_full_imputed.csv Output: table2_acc.docx table3_parents_imputed.R Constructs parent-level variables (dementia, health, insurance, income sources, nativity) and produces Table 3: aging parent demographic, health, and economic characteristics stratified by caregiver co-residence status, with chi-square and ANOVA p-values. Input: ACC_full_imputed.csv Output: table3_acc.docx acc_logits.R Constructs caregiver-level covariates (sex, age groups, marital status, education, income, health measures) and estimates logistic regression models on the full sample (N = 659) examining the association between co-residence and financial strain. Corresponds to Table 4 in the paper. Note: variables for caregiving intensity and head of household status appear in the script but were dropped from the final models after revisions and should be ignored. Packages: tidyverse, gtsummary, ggtext Input: ACC_full_imputed.csv Output: regression tables psm1.R Runs propensity score matching (1:1 nearest-neighbor, caliper = 0.2 SD) using the MatchIt package. Includes variable type checks, a targeted MICE imputation for matching variables, and balance diagnostics (love plot, SMD table via cobalt). Corresponds to the matching procedure described in the Methods section and Appendix. Input: ACC_full_imputed.csv Output: matched_data object (used by psm2_logits.R) NOTE: psm1.R and psm2_logits.R must be run in the same R session, as psm2_logits.R depends on the matched_data and ps_model objects created here. psm2_logits.R Estimates logistic regression models on the propensity score-matched sample (N = 368): (1) baseline model with co-residence only, and (2) doubly-robust model with co-residence and covariates. Merges both into a single formatted table. Corresponds to Table 5 in the paper. Note: variables for caregiving intensity and head of household status appear in the script but were dropped from the final models after revisions and should be ignored. Input: matched_data object (from psm1.R, same session) Output: acc_regs_psm.docx tablea2_acs_hepese_compare.R Loads the IPUMS ACS extract, constructs unique household and person identifiers, filters to Mexican-origin households (HISPAN==1), identifies households where an adult child (18+) co-resides with a parent (80+), constructs covariates (education, income, marital status), and produces the weighted survey summary table for the Mexican-American ACS column of Table A2. Input: usa_00014.xml (IPUMS ACS extract) Output: acs_mex.csv, adult_children_acs_mex.csv, Table A2 (Mexican-American column) NOTE: Must be run before tablea2_acs_white.R, as the latter depends on the acs object loaded here. tablea2_acs_white.R Filters the ACS data to non-Hispanic White respondents (RACHSING==1), applies the same household structure filters and covariate construction, and produces the weighted survey summary table for the White comparison column of Table A2. Input: acs object (from tablea2_acs_hepese_compare.R, same session) Output: Table A2 (White column, displayed in console) -------------------------------------------------------------------------------- R PACKAGES REQUIRED -------------------------------------------------------------------------------- tidyverse, mice, lattice, gtsummary, flextable, smd, gt, ggtext, MatchIt, cobalt, sandwich, lmtest, ggplot2, ipumsr, survey, srvyr Install all at once: install.packages(c("tidyverse", "mice", "lattice", "gtsummary", "flextable", "smd", "gt", "ggtext", "MatchIt", "cobalt", "sandwich", "lmtest", "ggplot2", "ipumsr", "survey", "srvyr")) -------------------------------------------------------------------------------- WORKING DIRECTORY NOTE -------------------------------------------------------------------------------- Scripts reference absolute file paths set during the original analysis (Google Drive paths on the author's machine). Before running, update all setwd() calls and file path strings in each script to reflect your local directory structure. Search for "/Users/annabokun/Google Drive/" to find all instances requiring replacement. -------------------------------------------------------------------------------- CORRESPONDENCE -------------------------------------------------------------------------------- Anna Bokun Population Division, NYC Department of City Planning abokun@planning.nyc.gov ================================================================================