House Rules: Spreadsheet Data

2018 MIDTERM ELECTION

Time: D H M S

Over the past month or so, several researchers, reporters and other data nerds have inquired about whether I'd be willing to open up the QHP and Medicaid spreadsheet permissions to let people export them to Excel. It was always my intent to do this once the 3/31 deadline was past. However, a few things complicate this.

  • First of all, due to the extension period, we're not even close to having all of the data in (and yes, I know that technically "all" of the data will never be in, since Medicaid/SHOP/etc. are year round with no deadline, and then we're into the 2nd open enrollment period, etc etc.)
  • Second, I never expected this site to take off the way it did--I assumed that it would have served its purpose, and was planning on pulling the plug on it later this week anyway. Now, that's obviously not a viable scenario...too many people have asked me to keep it going, and I've committed to doing so at least until the March HHS and CMS reports have been released (which likely won't be until early May...heck, the February CMS report still isn't out as of this writing!)
  • The third reason is the one which worries me the most, however. You have to understand that the spreadsheets are not nearly as precise and organized as you may think. Yes, they're pretty stable now, but they've evolved over time, and the earlier weeks (up through around Week 6) are effectively useless (ironically, just like HC.gov was through Week 6; this is not a coincidence).

Even the later weeks, where it is fairly solid and stable, aren't quite precise; I didn't always switch from the worksheet for one "week" to the next on the same day every week (sometimes it was Saturday, sometimes Sunday, etc), and I changed my mind frequently about which "week" to include given data points on anyway. Plus, of course, the date that I entered the data was usually a day or more later than the date of the article it came from, which in turn was often several days or even weeks after the date that the data within the article was good through.

So, I have a bit of a dilemma here...how do I make sure that people who download the spreadsheets don't completely misinterpret or misuse it, and then try to claim that there's a "fault" or a "flaw" in my methodology? (OK, I know the answer to that one: I can't.)

Anyway, the only way I can see around this is to download it myself, and then offer it on a case-by-case basis.

Thoughts welcome.

UPDATE: I should note that I will be making a PDF version available to anyone as soon as the HHS/CMS reports are in, however. Can't screw around with it, but at least it'll be printable :)