Core Dataset Project: Child Welfare Service Histories. Data analysis



In this analysis we looked at children’s child welfare careers as a series of "patterns". We created a string variable based on a child’s events, in which each of the characters represented a certain type of contact (event) with the child welfare system. We called this string the child’s "history". The histories were then analyzed using pattern matching techniques that we developed using the AWK programming language. This methodology allowed us to examine the varying paths children take through the child welfare system.

Reducing the number of unique patterns

Because each child’s situation is unique, there are countless variations to their histories or the "paths" they take through the system. The challenge in this analysis was to find comparability among the variation. In practice, what this meant was that we were forced to emphasize certain events over others.

In this case, we knew that we wanted to look at the progression of events that led to child welfare case opening and placement. A child can have any number of investigations before a case opening, but we knew from policy and preliminary analyses that for most, a case would not be opened as a result of an unfounded investigation. Therefore we were able to make the decision to ignore unfounded investigations in our history strings when they were interspersed with indicated investigations. We then decided that there was a significant difference between having only one indicated and having multiple indicated investigations before case opening, and that we would treat these as two separate pattern categories.

Even though we were not interested in unfounded investigations when there were indicated present, we were curious about what happened to children whose investigations were all unfounded. Therefore we included these as our third pattern of entry into the system.

The last way a child could enter the system was directly through child welfare service intake, without passing through child protection. Since these are mostly dependency and "behavior" cases, we decided it would be important to know how their subsequent histories differed from children who entered with an abuse or neglect allegation.

Entry Pattern Types and Descriptions

Pattern Type Description
A Indicated Investigation, any number of unfounded
B Indicated investigations, any number of unfounded
C All unfounded/pending investigations
D Case opening, no investigations

Once the children had entered the system by any of the four pathways, our pattern matching program looked for a case opening immediately following the entry pattern. If it found one, it then looked after the opening for one or more indicated investigations, the beginning of a placement, one or more indicated investigations again, case closing and/or placement ending. The program would continue to look for these subsequent patterns until we had captured the child’s first full sequence or, the child’s history ran out of events.

Subsequent Pattern Types (the program cycles through these repeatedly)

Pattern Type Description
D Case opening
D At least one indicated investigation
F Case Closing
G/H Placement Beginning
I Foster parent involved investigation
J Placement Ending

When the program finished, we were left with a condensed” pattern for each child that could be imported into a spreadsheet program and analyzed.

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®