Erratum: Contact with Child Protective Services is pervasive but unequally distributed by race and ethnicity in large US counties (Proc. Natl. Acad. Sci. U.S.A. (2021) 118 (e2106272118) DOI: 10.1073/pnas.2106272118)

The authors wish to note the following: "Shortly after the publication of our article we were contacted by another team of researchers (Emily Putnam-Hornstein, Eunhye Ahn, John Prindle, and DanielWebster) who have access to privately funded data from California that are consistently updated by the state and not publicly available. They pointed to differences between estimates generated using their privately held data and our estimates using publicly available data from the National Child Abuse and Neglect Data System (states are strongly encouraged to submit to this public data collection annually); they noted only very minor differences for the portion of our analysis that used publicly available data from the Adoption and Foster Care Analysis and Reporting System (AFCARS) data. "Upon receipt of this concerning information, we immediately began a thorough audit of our code and estimates. For three of four outcomes we estimated-confirmed maltreatment, foster care placement, and termination of parental rights; i.e., the outcomes most consequential for children and families-our estimates remain unchanged. Unfortunately, we found a coding error related to investigations that led us to incompletely remove children with prior investigations, resulting in an overestimate of first investigations. "Although some of the differences, especially for Black children in California, are nontrivial, the core conclusions from our manuscript hold in their entirety. We apologize for this honest coding error, thank the research team mentioned above for calling attention to a possible error, and are pleased that we are able to correct this error so shortly after the publication of the article." Due to this error, Fig. 1 and its legend appeared incorrectly. The corrected figure and its corrected legend appear below. In addition, some corrections to the text are necessary. In the Abstract, lines 6-9: "The total median investigation rate was 41.3%; the risk for Black, Hispanic, and White children exceeded 20%in all counties. Risks of having a CPS investigation were highest for Black children (43.2 to 72.0%)" should instead appear as: "The total median investigation rate was 34.5%; the risk for Black, Hispanic, and White children exceeded 10% in all counties. Risks of having a CPS investigation were highest for Black children (32.9 to 62.8%)." On page 1, right column, first full paragraph, lines 3-6: "The median cumulative prevalence was 41.3%. No jurisdiction had a total cumulative prevalence below 20%. One (Wayne, MI) had a total rate of over 50%" should instead appear as: "The median cumulative prevalence for these 20 jurisdictions was 34.5%. No jurisdiction had a total cumulative prevalence below 16%. One (Wayne, MI) had a total rate of over 44%." Also on page 1, right column, second full paragraph, lines 1-5: "Black children had consistently high rates of investigations, ranging from 43.2% in King County, WA, to 72% in Los Angeles County, CA. In most counties, having had a CPS investigation was a modal outcome for Black children. In 11 of the 20 counties, Black children had risks of investigation that exceeded 60%" should instead appear as: "Black children had consistently high rates of investigations, ranging from 32.9% in King County, WA, to 63.3% in Maricopa County, AZ. In most counties, having had a CPS investigation was a modal outcome for Black children. In 11 of the 20 counties, Black children had risks of investigation that exceeded 50%." In the same paragraph, lines 7-12: "Their highest rate was 24.2% in Riverside, CA, still roughly 40% below the median. Hispanic and White children tended to experience investigations in the 20 to 50% range. American Indian/ Alaska Native children experience low rates in most of these counties, but experience rates in the high 40s in three counties" should instead appear as: "Their highest rate was 19.8% in Riverside, CA, still roughly 40% below the median. White children tended to experience investigations in the 15 to 35% range. Hispanic children tended to experience investigations in the 25 to 40% range. American Indian/Alaska Native children experience low rates in most of these counties, but experience rates above 40% in two counties." The online version has been corrected.

Duke Scholars

Author Kieran Healy Sociology

Author Christopher Wildeman Sociology