Core Dataset Project: Child Welfare Service Histories. Method


The basic method of this study is the aggregation of the investigative and service events for a particular child into a sequence of events that they experience across the system. We accumulate events so that, for each child, we have the sequence of events that he or she experienced.

We follow the child from first investigation through last event of his or her first episode of service in the system. A complete episode would include the following events:

  • Birth of the child
  • Investigation records to which the type of allegations and the results of investigations are attached. Investigations occur after a report of abuse or neglect to a state central registry (a “hotline”). Not all reports are investigated and those reports that are not investigated are usually not recorded. Often the results of investigations are recorded soon after the report, but they must be concluded within 21 days in Michigan and 180 in Illinois. An investigator can remove the children from the home immediately (protective custody) and place them in an out-of-home placement before the investigation is complete. (Juvenile court must rule on the appropriateness of the removal within 48 hours.) We use the date of the report as the date of the investigation. Since we are interested in whether a child is part of a substantiated report, we assign the outcome of the family investigation to the child.
  • Case openings and closings for the child occur when the child is going to be placed in an out-of-home placement (the vast majority) or when the child is going to be the focus of some intensive in-home service.
  • Out-of-home care placements include foster family care, kinship care, specialized foster care, group homes, shelter care, residential care and other out-of-home placements (e.g. hospitalization) that the child may experience while in the custody of the state.

Of course, all children may not experience each of the events. We follow each child through all of his or her events until she or he has experienced a complete episode.

All of the events are merged into one file and the characteristics of child, caretaker or family are attached to these events. From the series of events that each child has experienced, we create a sequence of events. We then count how many children experience each type of sequence and cross-classify these by independent variables (region, year, type of initial allegation). The children are classified into cohorts by the year of their first contact with the system. In our database for this study, we include children from the 1990-1994 cohorts and follow their experiences through the end of 1994.

When tracking children from 1990 to 1994, unless otherwise noted, it is important to note that for each subsequent cohort, there is one less year of follow-up time that we have in our database. For children who had their first contact in 1990, there is follow-up through 1994. For children who had their first contact in 1994, we only look at that year’s subsequent events for that child. To adjust for this, we have also analyzed our data for all of the events that children experience only one year after their first contact with the system.

Because unsubstantiated records are expunged from the system, in compliance with state law, meaning that all identifying information on the records of the involved individuals is deleted, if a child does not have a substantiated investigation previous to an unsubstantiated one, that child may be double counted over a period of time. If a child experiences two unsubstantiated reports within a short period of time (usually less than a month), the child will likely be double counted. Also, if a perpetrator requests that the record not be expunged, because he or she wants to prove harassment on the part of a reporter, a child’s unsubstantiated reports will also be “trackable” over a longer period of time. The privacy concern is that states do not track perpetrators who have not been involved in substantiated cases.

Child welfare case opening refers to how many children have case openings of their own. This is the least comparable event in our report, since case opening is very much an administrative function and because families can also have case openings under which children can receive services. In Illinois, we are able to track family case openings for each child, while in Michigan, there is no source for such information. Thus, this will be an instance where separate analyses will have to be done for each state.

There are cases which come into the child welfare services domain without first experiencing an abuse or neglect investigation. These are cases in which for some other reason, the child and parents cannot live together. These reasons include the child’s behavior (children too young to be incarcerated), dependency (those cases in which the child may be an orphan or the parent in institutionalized) and, in some cases, status offenders. Since the practice around children entering child welfare services without an abuse or neglect report is highly variable from one county to another, not to mention one state to another, an argument could be made to drop these children from the study altogether. However, since these children are receiving service from the child welfare system and they do represent a sizable portion of the child welfare caseload, we include them.

Given that some children can have upwards of 20 either substantiated or unsubstantiated investigations and 5 case openings and 5 placement spells, the possible combinations that we have put together are in the thousands. We must make certain assumptions about types of cases in order to reduce our analytic units to a reasonable number.

We have taken the string of investigations that children experience and reduced them to three types. The first grouping is only one substantiated investigation with any number of unsubstantiated investigations. The second grouping is 2 or more substantiated investigations and any number of unsubstantiated investigations. The third is any number of unsubstantiated investigations. These investigations all have to occur before a case opening, which must occur before an out-of-home placement. Investigations that happen after a case opening or out-of-home placement constitute another type of sequence or maltreatment while in foster care.

In Michigan, the 1992 and 1993 cohorts’ placement figures will increase significantly since we could not obtain 1994 placement data from the state of Michigan. Children from the 1992 and 1993 cohorts could be placed at any time in the future. In 1992, there was a change in policy in Michigan, where the state had to inform the perpetrators of alleged child abuse or neglect. While a decrease in the number of substantiated reports is seen at that time, it is unclear what proportion is due to the follow-up period being short or due to the change in policy. With additional follow-up, the reason will become clearer.

View full report


"chapin1.pdf" (pdf, 71.08Kb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®