The good old days, part 2 (DCF abuse investigations over time)

Over the weekend, I posted some charts showing the number of kids under DCF supervision since 2004. Today we’re going to look at a slightly different measure: the number of abuse investigations between September 2006 and September 2015. Here goes.

1. This shows the total number of investigations from September 2006 to September 2015. Buried in DCF’s spreadsheets is a caveat that these numbers do not include calls that were screened out as “no jurisdiction.” Therefore be careful — this is the number of investigations, not the number of calls. This measures DCF’s response to calls, not the calls themselves.
2. The statewide total is at the top and the regional totals are at the bottom. Immediately you can see the stratification that we saw in the last post, this time in three clear groups: (1) Suncoast and Central, (2) Northeast and Southeast, and (3) Northwest and Southern.
3. All of the regions are strongly correlated (P <0.01) — they all rise and fall together.  This implies that anything that’s moving the numbers is either a change in DCF policy or some statewide phenomenon.

Let’s look at the numbers again controlling for the size of each region.

1. This chart shows how many investigations per child in the region were conducted each month. The three groups we saw above have fallen away and we now see that the farther north you go, the more investigations per child are conducted.  The Northwest and Northeast regions vie for the highest rate, while the Southern and Southeast Regions are consistently lowest.
2. Without knowing the number of actual calls per region, we cannot say any more than that. It could be that people in Miami do not call the abuse hotline as much as people in Pensacola. It could be that southern callers are screened out at much higher rates than callers from the northern parts of the state. The most striking difference between these regions of the state, of course, is the mix of languages. I’ve never called the abuse hotline in Spanish or Haitian Creole, so I do not know what that experience is like.

The zig-zags from month to month imply that there are seasonal effects going on. Let’s remove those to get a better picture of the trend lines.

1. This chart shows the trend lines without all the noise from the regular ups and downs that occur month-to-month. You can see the clear dip in the 2009-2010 period. Again, this means that fewer investigations were conducted. Without knowing the number of calls  total, we cannot say more.
2. Some statewide event happened at the end of 2010, right at the end of Secretary Sheldon’s tenure. Nowhere else in the data is there such a sharp change across all regions. I suspect there was a policy shift there, probably dealing with how calls were screened out, but I have not been able to find anything on DCF’s websites documenting it.
3. More recently, something is occurring in the Northwest (and to a lesser extent in the Northeast) beginning around November 2014. Either callers are getting better at bringing maltreatment to the attention of the Department or there is some policy encouraging investigations at higher rates in these areas. The same increase is not found in the south.
4. The regional rankings are amazingly consistent over time, with the exception of the two northern regions jockeying back and forth. I would expect to see more changes in the rankings. I am not a conspiracy theorist, but it looks as if there were almost quotas keeping the numbers even.
5. Remember the sharp increase in kids under DCF’s supervision starting in 2013? That’s nowhere to be seen in these numbers. DCF is not handling more investigations; they’re handling investigations differently.
6. I was curious about the effect that media reports have on DCF investigations, so I looked up the date of the Miami Herald’s Innocents Lost series. It published in March 2014 and spurred a series of town halls and other events over the following months. There is an inflection point in the data at March 2014, but notably it does not occur in the Southern Region where the Miami Herald is located. Complicating the picture is the fact that there were already very public discussions about DCF’s investigation policies in the preceding months and reforms were already working their way through the legislature. I know of one study finding that, instead of driving changes, news reports tend to follow the same forces that spur change in the system. I can’t say any differently with this data.

Interestingly, there is no correlation between the number of investigations and the number of kids under DCF supervision.

1. I checked the numbers and there is no statistically significant correlation here. Except for the slight bump in both lines around the end of 2010, the numbers of investigations and children under DCF supervision are independent.
2. I even ran cross-correlations over time to see if more investigations in one month resulted in higher numbers of kids in care in later months. Again, there was no correlation.

That’s it for now. If you read this far, you like graphs as much as I do. Next time I’ll look at the number of verified abuse reports and the types of maltreatment.