Texas: Child Maltreatment Prediction by Block

In 2015, 63,781 children in Texas were victims of maltreatment, making Texas the state with the third highest number of reports in the country.

The Texas legislature is currently looking at multiple bills that address child maltreatment including Senate Bill 11 which includes funding for traditional geo-mapping to forecast areas of likely child abuse. At the close of 2016, a group of researchers published a report on how they utilized the a different technique, Risk Terrain Modeling (RTM), to predict the location of sustained child maltreatment reports in Fort Worth with notable accuracy.

The Center for Prevention of Child Maltreatment at Cook Children’s Health Care System and Texas Christian University’s (TCU) Department of Criminal Justice used RTM to identify and weight critical risk factors. RTM is a predictive modeling technique that evaluates environmental factors to map locations where crimes will likely occur in the future. The risk factors used by Cook Children’s and TCU created a model that used data from 2013 to accurately indicate high-risk areas where almost all of sustained reports of child maltreatment occurred in 2014.

While the practice of RTM has gained attention for its use in successfully predicting crime trends, Cook Children’s and TCU are the first to use RTM as a method to map areas that are at high risk for child maltreatment. Unlike other predictive techniques, RTM utilizes environmental factors instead of relying on the analysis of past events. According to Breanna Anderson, program coordinator for The Center for Prevention of Child Maltreatment led by Cook Children’s, “RTM actually looks at risk factors surrounding what you’re trying to prevent … it’s much more future focused.” Anderson said Cook Children’s “goal with RTM is to be able to more effectively target intervention and prevention services” to high-risk areas.

The research team focused on Fort Worth, a Texas city located in a county that reported a higher rate of maltreated children in 2014 than both the state and national average, to test their model. Researchers compiled data from the Department of Family and Protective Services (DFPS), Fort Worth Police Department (FWPD), Texas Alcoholic Beverage Commission and Buxton customer analytics to identify and analyze ten likely risk factors, finding six to be the most significant in predicting areas of child maltreatment. The six risk factors (in ascending order of statistical significance) were: drug crimes, murders, runaways, aggravated assaults, domestic violence, and poverty.

Once the model had been run using data from 2013, the accuracy of the RTM was tested against the output of a traditional hotspot model. The hotspot model retrospectively analyzes past data to predict future incidents instead of looking at the current environmental factors. The RTM produces a relative risk score from one the lowest to 150 for each half-block (approximately 400 feet long) cell. The areas in the top 40 percent of the projected model were designated as elevated risk areas.

When maps of the RTM’s predicted areas of elevated risk were compared to actual reports from 2014, 98 percent of 2014 cases were found in areas the RTM had correctly identified as an elevated risk; out of a total 5,391 cases, only 133 happened in areas that were not labeled as an elevated risk. Comparatively, the hotspot model identified 91 percent and the total elevated risk area identified by RTM was 37 percent smaller than the total elevated risk area identified by the hotspot model.

Although the research done by Cook Children’s and TCU has shown the success of utilizing risk terrain modeling, the model needs to be studied in additional geographic areas to determine if the method and risk factors are applicable in other locations before adopting RTM on a larger scale. While RTM is still being tested and researched, possible uses of the model are similar to the current applications of geo-mapping efforts.

Organizations like TexProtects want to use geo-mapping to provide information on locations that “are most in need of prevention services based on the risk factors, so outreach into those areas can take place,” said Lee Nichols, the communications director at TexProtects.

While geo-mapping and RTM can be used in targeting areas in need of prevention programs, Nichols also suggested that “the exact same data as far as risk factors can help CPS see where they need … to concentrate their hiring efforts and infrastructure efforts.”

Bethany McKee is a Master of Public Policy student at USC’s Sol Price School of Public Policy with interests including education and social justice. She currently serves on the board of A Spacious Place, Inc., an Austin-area nonprofit that provides creativity classes, clubs and camps to underserved communities. She wrote this story for the Media for Policy Change course at USC.

Your support allows The Imprint to provide independent, nonpartisan daily news covering the issues faced by vulnerable children and families.

Subscribe or Donate

New York wants to use a fund for #FamilyFirst Act prep to prevent youth from aging out of #fostercare, but some counties say the money is already spent or earmarked https://bit.ly/3fU39MW #childwelfare