Misunderstanding the Numbers

Posted by

But so many people don’t understand. That is the problem. It allows for people to mislead – either intentionally or because they don’t really understand either. Yes, it is time for our weekly dive into the ‘rona numbers. Overall, this week was a mixed bag. I would say slight trending upwards in cases but the data aren’t entirely clear on that fact.

Saw a scare headline on a news site talking about a “surge” in the Northeast. Started with the line that they thought they had it under control but now cases were “surging” again. The reporter based this on a week over week change which did look scary but didn’t account for the increased number of testing. The only thing the reporter did say (buried in the text) was the the percentage of positive results was still around 1.5%.

How many people read the article? How many just looked at the headline and said “here we go again”. This is lazy journalism – click bait headline designed to mislead by a reporter who likely doesn’t really understand the numbers. If the number of cases goes up at the same rate as the number of tests, that isn’t a “surge”. If the background rate of positive cases remains the same, that isn’t a “surge”.

Let’s start with the overall figures. It is true that the number of newly reported positives was the highest since the week of August 9th. Of course, it is also true that there 6.8 million tests reported last week which again set an all time high. The percentage of positive tests was 4.9% which was higher than last week but has basically been flat since the week of August 30th. Again, I wonder if we don’t have some floor that we’ll never get below due to false positives.

The number of newly reported deaths was down from last week but since it only dropped by 10, I’d say it was really flat. I also decided to track hospitalizations as another metric. If the case rise is real, it will be reflected in new hospitalizations. I went back and added all that data retrospectively. We’ve been between 8,161 and 9,592 since the week of August 30th. The peak in the summer was 21,432 and the peak in April was 27,260. So we are down from that but essentially flat.

Overall, the trend suggests flat to slightly increasing. Some states are seeing bigger increases which we’ll get to below but the data don’t suggest a “surge”. At least not yet. But we do have 15 states that are currently at a peak in terms of my 10 day average of reported cases.

So let’s get to it. In the first group, we start with Arizona with 7 points. Then we get New York, DC, Connecticut, New Jersey and Florida with 8 points. After that, we have California and Massachusetts with 9 points and we close the group with Louisiana, Georgia and Maine with 10 points.

The scores are a little higher for the states in the Northeast because I base my statistics on case count without adjusting for number of tests. A lot of these states ramped up testing and have “found” more positives. But the actual number of cases relative to population remains small and the growth rate is around 2% so they still fall within this group.

Here is a graph of New York so you can put the “surge” into context with what it looked like in the spring. You see the graph go negative in May – states make all kinds of adjustments to their data so you get stuff like this. Yes, there is an uptick at the end but I’d hardly call it a “surge”.

Moving on to our next group, we start with Hawaii at 11 points. They are followed by Vermont, Maryland, Texas and Washington with 12 points. Then we get Rhode Island and South Carolina with 13 points followed by Nevada with 14 points. And we finish the group with New Hampshire, Delaware and Alabama with 15 points.

New Hampshire and Vermont are like the other states in the Northeast where they showed an increase in cases but they still have a low number of cases per population. In this group, the rate of increase is 3-5%. To pick the oddest example in the group, here is the graph for Rhode Island. This is one of those states that doesn’t report new cases on the weekends so you get the points at zero. Plus they made an adjustment back in late September which probably kept the average down. But it is also another place where testing has shot up so much of the increase is due to that as the percentage of positive results was still under 2% as of last week. (Won’t get new state numbers on testing until later today)

Moving on to the middle of the pack, we start with Mississippi and Virginia with 16 points. They are followed by Pennsylvania, Oregon and Michigan with 17 points. Behind them are Ohio and Colorado with 18 points and we close the group with New Mexico and West Virginia with 19 points.

The big movers in this group were Pennsylvania and New Mexico and both were driven by rise in cases. In this group, the rate of growth over the past week ranged from 4.5% (Mississippi) to 8.8% (West Virginia). Colorado and Michigan hit new peak values in terms of the average number of newly reported positive cases. Colorado we discussed previously.

Let’s look at Michigan. The one point over 5000 represents a dump of “probable” cases they did back in May. Parts of Michigan were hit harder in March where I don’t have data but the trend over the last couple of days is slightly up. Again, not an exponential surge but a small upward tick. Michigan essentially doubled the number of tests starting the week of July 19th and that roughly tracks to where the case count jumped. Have reason to believe that some of the recent increase is real and not due to testing but the jump in positives was from 2.5% to 2.8% so we aren’t talking about a huge increase in actual cases.

I want to pick one more state from this group and that’s Pennsylvania which moved quite a bit in the wrong direction. Here’s that graph. Don’t have testing figures for the last week and there is some distortion by the one big spike which represents two days of testing. But they have seen an increase in positives – over the last month it was 6.8% vs an average of around 5% over the summer. So this one looks like a real increase that needs to be tracked.

The next group starts with Tennessee, Illinois, Kansas and North Carolina with 20 points. They are followed by Indiana and Minnesota with 21 points and the group closes with Kentucky, Iowa and Arkansas with 24 points. A couple in the Midwest and a couple of the more northern southern states (if that makes any sense). Tennessee is the only place that really moved in the wrong direction. The rest were basically flat from last week.

In this group, the rate of change ranges from 5.5% (Kansas) to 10.3% (Kentucky). Indiana, Minnesota and Kentucky are all at new peak values. I need to get the figures for last week from Kentucky but a lot of what was going on there is an artifact of increased testing. Starting the week of September 20th, they more than doubled their previous rate of testing but the percentage of positive results actually went down. I suspect this is increased testing of school age kids but I can’t verify that. But it doesn’t give me a lot to be that concerned about.

In fact, for most of the states in this group, the increase is more driven by testing as the percentage of positive results has been mostly flat. The exceptions are Iowa, Kansas and, arguably, Indiana. Kansas only reports on a couple of days so their graph is ugly. I’ll show Iowa instead.

Lots of daily variability in the reporting but the 10 day average is clearly settling at a higher place. The one day spike created what looks like a false peak back in September. Still the graph is flat and not really showing an increase in cases. If this is just a plateau, it may settle here for a bit.

Now the last group where things could be better. We start with Alaska, Oklahoma, Nebraska and Idaho with 25 points. They are followed by Missouri and Utah with 26 points and then Wisconsin with 27 points. Then we get North Dakota and Wyoming with 28 points and South Dakota with 29 points. In last place for the second week in a row is Montana with 30 points. A lot of the upper Rockies and plains.

This is the group where cases are rising at the fastest rate. We range from 8.7% (Oklahoma) to 26% (Montana). With the exception of Wyoming, all of this group is currently at a peak in terms of the average number of newly reported cases. Wyoming’s peak was on Friday and they are less than 1 person from that peak so we could consider them at peak as well.

Of the group, Oklahoma appears to be the only one where testing is driving the increase. Montana did see a big jump in testing but the case positivity rate is increasing with that so testing doesn’t fully explain the increase.

I think I’ve shown a lot of these before. Here’s Utah. Still trending upwards but the rate of change seems to have slowed.

And, to close things out, here is the graph for Wyoming. Unlike Utah, it looks to still be increasing. We have to watch over the next week to see if we get a replication of the the two high values from last week. That would help to confirm an increasing pattern. In terms of total cases per million people Wyoming has ranked near the bottom so this is likely the virus just catching up in places where it hasn’t hit very hard before.

So that’s where things stand right now. Will have to pay some attention to the testing data in places in the Northeast to see how much of the “surge” is really testing. Places like Montana and Wyoming are still in a fairly rapid growth phase and it is hard to predict how long those will last.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.