Maryland State Innovation Team

Maryland Child Poverty Correlates Data

Data Sources

• American Community Survey (ACS) 5-Year Estimates (2023): Socioeconomic, housing, and demographic variables via the tidycensus and censusapi R packages.
• Maryland Family Network Childcare Center Locations: Custom geocoded dataset.
• Small Business Lending Data: Federal Financial Institution Examination Council (tract-level lending, 2014–2023).
• Maryland Census Tract Names: OpenStreetMap.

Methodology

• Variable Selection: Variables are chosen to capture correlates of child poverty, including unemployment, underemployment, self-employment, industry, education, family structure, language, internet access, childcare access, and small business lending.
• Data Acquisition: Data is pulled using R packages (tidycensus, censusapi, sf, etc.) and merged by census tract GEOID.
• Derived Metrics: Most variables are calculated as rates (e.g., child poverty rate, unemployment rate, percent without internet) by dividing relevant ACS estimates.
• Special Calculations:
  • Child Poverty Rate: Sum of children under 18 in poverty divided by total children under 18.
  • Unemployment/Underemployment Rates: Calculated for various age/gender groups, including labor force non-participation.
  • Self-Employment & Industry: Rates for self-employment and employment in agriculture, manufacturing, and transportation/warehousing/utilities.
  • Education: Percent of families in poverty with less than high school education.
  • Family Structure: Rates for married-couple, single-dad, and single-mom families in poverty, including those under 25.
  • Grandparent Responsibility: Rates for grandparents responsible for grandchildren, including duration and absence of parent.
  • Language: Percent of households in poverty speaking a non-English language at home.
  • Internet Access: Percent of households without any access to the internet.
  • Childcare Access: Count of childcare centers per tract (spatial join), and average childcare cost as a percent of median family income.
  • Small Business Lending: Dollars per capita from tract-level lending data.
• Data Cleaning: All NaN or infinite rates are set to zero.
• Merging: All data is merged by tract GEOID, with tract names appended.

Key Variables

• child_pov_rate: Child poverty rate
• laborforce_nonparticipation_rate: Labor force non-participation rate (16+)
• civilian_unemployment_rate: Unemployment rate (civilian labor force)
• male_*_unemployment_rate, female_*_unemployment_rate: Unemployment rates by age/gender
• male_*_laborforce_nonparticipation_rate, female_*_laborforce_nonparticipation_rate: Labor force non-participation by age/gender
• mean_usual_hours_worked: Mean usual hours worked (16–64)
• male_underemployment_rate, female_underemployment_rate: Underemployment rates by gender
• self_employment_rate: Self-employment rate
• industry_agriforestfishhuntmine_rate, industry_manufacturing_rate, industry_transportwarehouseutility_rate: Industry employment rates
• poverty_education_lthighschool_rate: Percent of families in poverty with less than high school education
• poverty_family_marriedcouple_rate, poverty_family_singledad_rate, poverty_family_singlemom_rate: Family structure rates in poverty
• poverty_family_under25_rate: Percent of families in poverty with householder under 25
• poverty_grandparents_responsible_rate: Grandparents responsible for grandchildren in poverty
• grandparents_responsible_noparentpresent_rate: Grandparents responsible with no parent present
• grandparents_responsible_length_gt1year_rate: Grandparents responsible for >1 year
• poverty_language_nonenglish_rate: Percent of families in poverty speaking languages other than English at home
• internet_no_internet_rate: Percent of households without any internet
• childcare_centers: Count of childcare centers in tract
• childcare_cost_percent_median_family_income: Average childcare cost as percent of median family income
• small_business_loan_dollars_per_capita: Small business loan dollars per capita
• (Tract name fields appended)

Notes

• All rates are calculated at the census tract level for Maryland.
• All NaN or infinite values are set to zero.