Transparent California Data Accuracy

So.. in the course of this I have had – repeatedly and often – the fact that “Transparent California data is not accurate” thrown out.

I have no way, of course, to validate that.  I have no access to actual pay records of anyone in the District, and certainly can’t examine everyone’s W-2 to find out what they really made.

For starters, the basic idea behind Transparent California is that it publishes data on individual pay and benefits, as required to be done in the California Public Records Act, sections 6250-6270.  A copy of a Guide to the Act can be found here.  Page 49 outlines these requirements.

FAQ on Transparent CA can be found here.

As an fyi also, I have worked with the District to fix the Transparent CA numbers for benefits, which I’m told are now accurate for 2016 on Transparent CA (as of 11/16/17) and I’m using the updated, “accurate” data in my analysis.

I’m told that the issues with the pay numbers relate to a query provided through the County Board of Education, and I have emailed them a request to find out how one goes about getting that fixed also, but that just happened (11/19/17) and I have not yet gotten a response.  Will update this as that progresses.

With “the data we have right now” (11/19/17) I can, however, do some double checking – at least on the teacher pay side.  Unfortunately the administrative and classified side is not verifiable with any outside sources I know so I can’t go there, but let’s look at the teacher data…

First, I’m using 2016 data (not 2017) because that’s the latest available.  This means I’m also using 2016 data files for the rest of the documentation.

My methodology for looking at total, average, and median teacher pay in Transparent California was to first eliminate any teachers making less than than the Step1/Column 1 minimum of $38,319/year, per the 2015-2016 OTA Contract.

I do this on the assumption that it is not possible for a teacher who worked full time for a full year to make less than this.

With this filter, the total list of teacher names adds up to 846, and the average teacher “total pay” for this group is $77,845.  Average “total pay and benefits” are calculated at $101,129, which means total benefits are average $23,284 or 29.91% of total compensation.

To double check this, I turned to several other documents, including…

  • Reported teacher headcount as given by the District to the DPAC in spring 2017
  • State of California’s Department of Education “J90” data that reports their view of average pay by district (page 21.)
  • OUSD 2015-2016 actuals (dated 10/11/16.)

For easy reference these documents can be found here under Additional Data Files.

Reported teacher headcount by the district is 865, which is close to the 846 in my data set.  Based on my criteria, Transparent CA under-reports total number of teachers by 19 or -2.2%.

Reported average teacher pay in State’s data (J90) is $79,629, which is also close to my data set’s results of $77,845.  My Transparent CA data under-reports the average pay by $1,784/year, -2.2% also.

Reported pay and benefit costs (for all OUSD employees) are shown in the 2015-16 actuals (page 2-1) as

Pay – $93,729,299 + 32,052,650 = $125,781,949

Benefits – $52,738,098

Which means benefits are 29.54% of pay.  This is very close to my Transparent CA calculated number of 29.91%.  In this case my Transparent CA numbers over-state the benefit percentages by 0.37%, or an overstatement of 1.25%

Conclusion…

Although I can agree that Transparent CA’s numbers are obviously not accurate if we look “employee-by-employee”, they do appear fairly accurate when viewed in aggregate.  When validated using outside documentation, no identified error is greater than 2.2% of the total for that value.

Since I’ve done that validation ONLY for certificated pay and benefits, there is no way too prove that this same level of accuracy applies to ALL pay and benefits, however given that the data comes from the same source, using the same queries, and is reported to Transparent CA using the same mechanism, it’s hard to imagine any reason why that system would introduce errors in specific employee pay data that are somehow significantly different dependent on employee group, however I cannot say that for sure.

For the purposes of this examination, barring some form of better, verified accurate data, I’m going to choose to accept the fact that the Transparent CA data is likely at most inaccurate for all employee data by somewhere around 2% – which I do not feel is of a magnitude that invalidates this analysis.

If you see something different, let me know!   ousd@maddisonweb.com