Table 3.

Standardized and unstandardized predicted probabilities and average treatment effects on the treated associated with having below average general health for Philadelphia adults and their children based on whether the adults witnessed or did not witness domestic violence during childhood.^a

	Standardizedb					Unstandardized
Outcome of interest	Non-witness	Witness	Effecta	95% CI	P-value	Non-witness	Witness	Effecta	95% CI	P-value
Poor adult health	0.41	0.45	0.04	−0.12, 0.19	0.65	0.37	0.50	0.13	−0.07, 0.34	0.21
Poor child health	0.11	0.26	0.15	0.02, 0.28	0.02	0.09	0.30	0.22	0.10, 0.34	0.001

Values in bold are statistically significant at p<0.05.

^aInverse probability of treatment weighting is used for all models to compute average treatment effects on the treated (Effect). Propensity scores for the IPTW models were comprised of baseline (childhood) characteristics including sex; race; growing up in Philadelphia; >2 household Adverse Childhood Experiences; >1 community Adverse Childhood Experiences; >5 total Adverse Childhood Experiences; victim of physical abuse; victim of sexual abuse; victim of verbal abuse; victim of abuse that resulted in visible injury; food insecurity; sense of community in one's neighborhood; victim of bullying; observed community violence during childhood; had a supportive adult in childhood that made you feel special; lived in foster care; felt discriminated against; and lived with someone during childhood who had: a mental health condition, abused alcohol, abused drugs, been in prison. After applying the inverse probability of treatment weighting, balance was achieved for all covariates across witnessing groups except for: respondent sex, injury from abuse as a child, and living with adult who had a mental health condition during childhood; thus, these factors were added to all response models. The survey sample was weighted to account for adult age, poverty status, sex, race, and Hispanic ethnicity distributions from the most recent Philadelphia census and American Community Survey (U.S. Census Bureau, n.d.).

^bStandardized adult models include: witnessing domestic violence during childhood, adult age, chronic disease, smoking, feeling unsafe in one's current neighborhood, and socioeconomic status. Standardized child models include: parents' witnessing status from childhood, parent health score (comprised of: adult age, chronic disease, smoking, feeling unsafe in one's current neighborhood, and socioeconomic status), and child asthma diagnosis.