3. AFST USES PROXY VARIABLES TO
DETERMINE CHILD ABUSE/
MALTREATMENT
COMMUNITY REFERAL
– frequency of calls to
hotline on same child
w/in two years
FOSTER PLACEMENT –
frequency a child has
been removed from
home
4. DATA SOURCES
• OVER ONE BILLION RECORDS FROM:
• public agencies and programs including child welfare agencies;
Head Start; housing authorities; and the county jail
• means-tested welfare programs like TANF, SSI, SNAP, and
county medical assistance
• interactions with the juvenile justice system and the CYF office
5. “OVERSAMPLING THE POOR”
SAMPLING: Collecting data from a ‘representative’ group within a
specific population.
OVERSAMPLING: Collecting myriad data points that may actually
overlap from a representative group, and possibly overrepresenting
that group.
6. BIAS IN =
BIAS OUT
• IN: Mandated reporters and other
community members call the child abuse
hotline about Black and biracial families 3.5x
more often than they call about white
families.
• IN: By collecting call and referral data, the
AFST is collecting data at the most racially
biased and disproportionate point
• OUT: ‘DATA’ supports targeting low-income
families, and Black/ bi-racial families.