Correcting contamination when studying child abuse and neglect

Contamination is a significant problem when examining the effects of child abuse and neglect, according to new research from the Penn State Department of Human Development and Family Studies and Center for Healthy Aging . Contamination occurs in human-subjects research when people classified into a control or comparison group have — in reality — been exposed to the event under investigation.

For scientists who study child maltreatment, contamination can be very hard to detect and correct. Researchers could never assign a child to receive abuse and neglect the way they would assign people to different medications. So, to study abuse and neglect, scientists examine a large group of children and compare the individuals who experienced abuse and neglect to those who did not. However, determining who was abused or neglected can be difficult.

New research by John Felt, assistant research professor in the Center for Healthy Aging, found that contamination can diminish the effects of child maltreatment in research studies. Results from the study were recently published in the Journal of Child Psychology and Psychiatry.

“Contamination in child maltreatment research is highly prevalent and can obscure the results of a study,” said Felt. “To detect contamination, we recommend that researchers use two types of data sources — for example Child Protective Service records and self-reports from the children — to characterize whether a child has been maltreated.”

The researchers examined records from 634 children in the LONGSCAN study. Sixty-three participants had confirmed cases of abuse or neglect after age eight, and these were originally counted as the group who had experienced maltreatment.  

The researchers, however, reviewed self-reports of abuse and neglect from all children in the study at ages 12, 14, 16, and 18. They found that around two-thirds of the children who were counted as not experiencing maltreatment had — in fact — experienced abuse or neglect. (This does not mean that two-thirds of all children have been abused or neglected, but it does mean that more than two-thirds of the children in this sample were.)

The presence of this contamination in the study made the effects of abuse and neglect seem smaller than they actually were. After correcting for contamination, the effect of child abuse and neglect on a child’s internalizing behavior problems — such as excessive worrying or depressed mood — were as much as 54% higher than the data previously indicated. The effect of child abuse and neglect on a child’s externalizing behavior problems — such as delinquency or defiance — were as much as 18% higher than the data previously indicated.  

“This study will hopefully help researchers consider how contamination can introduce bias into child maltreatment research,” Felt said. “Addressing contamination can help improve the accuracy of analyses, which can help prevention and intervention researchers determine how to better understand and help children who have experienced maltreatment. Addressing contamination may also help explain why some associations do not replicate well from study to study.”

This work is part of more than a decade of research on contamination by Chad Shenk, professor of human development and family studies and pediatrics at Penn State and senior author on this research. Shenk is also the principal investigator on the grants that funded this work.

“Contamination is a feature, not a bug, of conducting research on child maltreatment,” Shenk said. “It presents a major concern for researchers because it makes it harder to find an effect for maltreatment when one exists and the effects that are found are often much smaller in magnitude. This paper is a part of a larger program of research in my lab testing new methods for correcting contamination bias in child maltreatment research that can also be used by other researchers outside the area of child maltreatment.”” Shenk said.  

Other Penn State faculty and students who contributed to this research include human development and family studies (HDFS) graduate students Ulziimaa Chimed-Ochir, Anneke Olson, and Yanglin Li and Assistant Professor of Human Development and Family Studies Zachary Fisher. Kenneth Shores of the University of Delaware and Nilam Ram of Stanford University are co-investigators on the grants and contributed to this work.

The National Science Foundation and the National Institute of Child Health and Human Development funded this research.

Read the full research article.  

Originally published in April 2024