● A lawsuit alleges that a child who wandered out of his mother’s home in West Virginia was placed – for that reason alone -- in foster care. He wandered out of the foster home and died. I have a blog post about why this tragedy is a failure not only of West Virginia foster care but also of West Virginia journalism, with links to news accounts about the lawsuit from WTAP-TV, The Parkersburg News and Sentinel and WCHS-TV.
● Acting on recommendations from a task force aiming for less mandated reporting and more mandated supporting, The Imprint reports, California lawmakers have introduced two well-intentioned bills with that intent. But it’s not clear if either bill will have the intended effect.
One bill would require counties to set up an alternative response mechanism for what are known in California as “general neglect” cases. The intent appears to be that the response to such reports come from community agencies, not the family police. That’s good, and sets this apart from many other “alternative response” programs, but the language is vague concerning how much involvement the family police still will have. That’s problematic. In New York City, for example, family advocates say the city family police agency’s alternative response can actually be more oppressive than a conventional investigation.
The second bill would set up a pilot program in Los Angeles to require mandatory reporters to get training in determining when there are better alternatives to a report. But in addition to the training, there also would be an interactive “decision support tool” that Los Angeles County agency would be responsible for creating.
But here’s the catch: According to the official summary of the bill
The bill would require the decision support tool to, among other things, make a recommendation on whether or not to report. The bill would, during the time the pilot program is in effect, deem a mandated reporter to have satisfied their reporting duties if the reporter completed the training, used the decision support tool, and complied with the recommended action. [Emphasis added.]
Caseworkers have been using these sorts of “decision support tools” for years – and there have been widespread concerns about racial bias in those tools – including in Los Angeles.
● The Imprint also reports on the legislative agenda of family advocates in New York State, which may be more promising.
● In still another failure
for predictive analytics, an experiment in Japan to use AI to point out child
abusers has failed dramatically. The
failure was in finding the abusers; the study only looked for what are called
false negatives. The researchers didn’t even ask about false positives – that is,
wrongly accusing innocent families. You
can read about it in Unseen Japan. Their story
is based on this one from the Yomiuri Shimbun
(Google Translate works pretty well.)
