08-06-2020, 07:36 AM
Actually, to me, "getting stuck" sounds encouraging. That suggests the information is in the system somewhere. "getting lost" or "getting dropped" would sound much more worrisome. Fixing the problem may be simple. What will be hard is making sure the "stuck" information (and only it) is processed.
When those numbers come through, there will be a big hit to California (and the country's numbers). If California has been undercounting by half for two weeks, that's a lot of cases. BUT, I only suspect "by half" for SCC and San Mateo County. I don't have any sufficient evidence to say the numbers are down by half in L.A. County. It doesn't seem likely.
I would also be worried that a similar problem could have been ongoing. The current situation hints that there may not be adequate safeguards to be sure that the data that goes in is actually being processed. I hope they're reviewing that as well to make sure it wasn't happening (perhaps to a lesser extent) prior to July 20th or so.
I don't know when they first found out there was something other than a delay. I commented about the > 50% drop in a comment to SCC PHD on Friday July 31 to which they responded on Aug 2:
At that time, the responder (not part of the data team, but one who responds to comments from the public) clearly didn't understand the situation as more than a delay. (As indicated in other posts, I thought the 50% drop was due to the decision not to process samples of people in Tier 4.)
The data calculation error I reported was that SCC (not the state) was calculating 7-day percent positive numbers by averaging the percent positive of each of the 7 days (while the state specified calculating number of positives over 7 days divided by number of tests over 7 days). As a result, SCC's graph of %-positive values were slightly too high (I'd say about 5% too high: say 4.0% reported vs 3.8%). (The low volume days contributed more than they should.)
The sudden inclusion of the "stuck" data is going to impact the state's monitoring list dramatically.
As I expected, Santa Clara dropped out of the monitoring list in the list released on July 29. (But, surprisingly to me, San Mateo County was added, for having too many new cases. Santa Clara, San Francisco, Alameda, and Sonoma Counties dropped out the same day, and they were the only ones that dropped out.)
That list wasn't updated again until August 2, at which point those counties again were on the list (except Sonoma, which remained off). The list as of August 5 hasn't changed since August 2. Previously, the list was changing daily. That may be a hint that the state realized something was wrong by July 30. (I'm surprised the Governor didn't reference it in his public comments earlier this week.)
The dropping of the several Bay Area counties suggests to me that the "stuck" data may be because of some regional difference. Maybe the problem only affects the Bay Area counties. (But I did report that Orange County had a notice about the problem.) Or it may be that the counties using certain software are being affected.
And, by the way, if you haven't looked, the list indicates that the hospitalization numbers (showing "N/A" now) haven't been updated since July 23, due to changes in federal reporting requirements. So, the most critical numbers (number of beds available) aren't being reported!!
When those numbers come through, there will be a big hit to California (and the country's numbers). If California has been undercounting by half for two weeks, that's a lot of cases. BUT, I only suspect "by half" for SCC and San Mateo County. I don't have any sufficient evidence to say the numbers are down by half in L.A. County. It doesn't seem likely.
I would also be worried that a similar problem could have been ongoing. The current situation hints that there may not be adequate safeguards to be sure that the data that goes in is actually being processed. I hope they're reviewing that as well to make sure it wasn't happening (perhaps to a lesser extent) prior to July 20th or so.
I don't know when they first found out there was something other than a delay. I commented about the > 50% drop in a comment to SCC PHD on Friday July 31 to which they responded on Aug 2:
Quote:The current testing data set for the past 2 weeks do not reflect the actual number of tests conducted. There is currently a backlog of tests that have yet to be recorded and are not represented on the testing dashboard. Please continue to check back as we update the testing figures and fix the positivity rate that you discovered to be incorrectly computed.
At that time, the responder (not part of the data team, but one who responds to comments from the public) clearly didn't understand the situation as more than a delay. (As indicated in other posts, I thought the 50% drop was due to the decision not to process samples of people in Tier 4.)
The data calculation error I reported was that SCC (not the state) was calculating 7-day percent positive numbers by averaging the percent positive of each of the 7 days (while the state specified calculating number of positives over 7 days divided by number of tests over 7 days). As a result, SCC's graph of %-positive values were slightly too high (I'd say about 5% too high: say 4.0% reported vs 3.8%). (The low volume days contributed more than they should.)
The sudden inclusion of the "stuck" data is going to impact the state's monitoring list dramatically.
As I expected, Santa Clara dropped out of the monitoring list in the list released on July 29. (But, surprisingly to me, San Mateo County was added, for having too many new cases. Santa Clara, San Francisco, Alameda, and Sonoma Counties dropped out the same day, and they were the only ones that dropped out.)
That list wasn't updated again until August 2, at which point those counties again were on the list (except Sonoma, which remained off). The list as of August 5 hasn't changed since August 2. Previously, the list was changing daily. That may be a hint that the state realized something was wrong by July 30. (I'm surprised the Governor didn't reference it in his public comments earlier this week.)
The dropping of the several Bay Area counties suggests to me that the "stuck" data may be because of some regional difference. Maybe the problem only affects the Bay Area counties. (But I did report that Orange County had a notice about the problem.) Or it may be that the counties using certain software are being affected.
And, by the way, if you haven't looked, the list indicates that the hospitalization numbers (showing "N/A" now) haven't been updated since July 23, due to changes in federal reporting requirements. So, the most critical numbers (number of beds available) aren't being reported!!