Hacker News Reader

Comment by hnrodey | Hacker News Reader

In the past 14 days, 99.798% of residents in San Francisco County have NOT received a positive COVID-19 test.

See for yourself at https://thecovidcomplex.com/state/california/san%20francisco...

Full disclosure, I built this website. Data is sourced from the New York Times. It’s the same dataset they use to power the interactive map that was linked. And I just discovered I have an error sharing links to counties with spaces. Go figure lol

Edit: 4 minutes since posting and a downvote. At least explain yourself.

View on HN

Didn't downvote you, but that 99.7% number is a bit meaningless unless you also state the number of people who were tested at all. Not receiving a positive result due to not being tested in the first place means nothing.

Hi, I appreciate the comment and discussion.

I disagree with the statement that the number is meaningless. Media outlets are free to singularly report the number of positive cases. I see it all over mainstream media (CNN being the worst offender), Twitter, etc.

What's the issue with simply reversing that metric to show many people are COVID negative?

Final note, I'm happy to make updates based on additional data but public datasets are slim pickings. And I'm not in to tapping individual state/county datasets.

NYT publishes daily case counts and death counts across the United States by county.

edit: COVID negative is unequivocally the incorrect term to use. That was a mental mistake on my part. The specific term I use is “did not get a positive COVID test”.

First, just because some entity does it, "The Media", doesn't mean it's good.

Second, the number of positive cases is an affirmative result from people that have taken, and failed, covid tests. Your metric takes people who have never taken a test and lumps them into "COVID negative," which is pretty self-evidently extremely misleading.

I would report test positivity rate if I had a source. I’m not trying to hide anything.

I contend that if you are not taking a test it’s because you didn’t need a test.

It’s also misleading to report only case positivity. Some areas of the country allow only people who have active symptoms to get a test thereby skewing the results.

> I contend that if you are not taking a test it’s because you didn’t need a test.

Another reason you might not take a test is that you don't know you need it.

Life is full of risk and trade-off’s. I document my grievances on the About page.

Asymptotic transmission is real (per the CDC) so I’ll accept that as fact. Ultimately if you view yourself as a high risk then it should be a personal responsibility to protect yourself. I believe my family and I to be low risk - citizens should be free to decide their tolerance for risk and act accordingly. Other families across the country and world have similar feelings about health.

I’m frustrated watching small business owners get crushed, bad their employees lose their livelihood.

There is significant damage being done with shutdowns and lockdowns that is with great chance worse than this disease itself.

I'm frustrated too.

But I think your point is strong enough to be presented without exaggeration or misleading.

This is a very misleading statistic you're providing, especially to people who may not be as data-oriented as us, and also a very broad assumption that definitely doesn't hold. Case positivity is a real number based on number of tests, it's not making inferences on things like the rates among those who haven't been tested.

I understand this is your site and you built it, and it's hard to let that go. But, I would implore you to rethink it until you get the data needed to add real insights to the conversations, not misleading stats based on what you have now.

Also, I have consciously avoided the term COVID negative. Exactly why I write “have not had a COVID positive test”.

But you have used it though.

> What's the issue with simply reversing that metric to show many people are COVID negative?

Thanks for calling me out. It was a mistake in my earlier comment and I have added an edit.

I need to do better.

> metric takes people who have never taken a test and lumps them into "COVID negative,"

True, but the statistic is still useful with respect to comparisons to other locations in the country under the same reporting constraints.

...which hnrodey didn't include in the comment.

The media reports the _known_ positive cases, i.e. those that were confirmed by tests and/or doctor diagnosis. There are also a vast majority of asymptomatic cases which are not generally tested or even contact-traced in the US.

We can't make claims about how much of the population at large is healthy because we don't actually know who is healthy and who is covid-positive/infectious/asymptomatic (and this is precisely why there are indiscriminate mask mandates in SF and elsewhere)

Because while that statement is technically mostly factual, it leads to a implication - like the one you make right here "to show many people are COVID negative" - that this metric somehow means that this many people are COVID negative, which is utterly false.

"Simply reversing" that metric gets a number that mostly shows how many people have not been tested within 14 days and thus noone has any idea on their COVID status. It's negligent to imply that this means that they are or would be negative.

Furthermore, even the original data point seems not literally true - this is based on net new cases, which would exclude people who became a net new case 15 or more days ago but actually have received a positive test recently because they are, for example, in the hospital right now and have gotten a repeated test that shows that they are still positive.

I would argue that making factually true statements for which the obvious implication actually is not true is inappropriate - and if it's not by mistake (as it seems out of your other comments) then it's intentionally misleading and should be downvoted.

Large numbers of positive results are an entirely different kind of data than large numbers of the opposite (neutral + negative).

One shows the minimum scale of the problem - if it's big, that's a problem, even if it's no bigger than the value reported. E.g. if 10% tested positive, that'd be unambiguously worrying / news-worthy.

The other shows the maximum awareness of a problem. If it's big, you have no idea if awareness is low (many neutrals) or if incidence is low (many negatives). Everyone in SF could have COVID, and that 99.798% number could still be factually correct. Or it could be that only those .202% got it. There's no way to know from that number alone.

I think it's harder to grasp the negative counts, because this is essentially the population of the city almost every day. The statistical change from day to day would be less than 1% so it's hard for humans to grasp a significant increase in cases. If the daily cases were to go up by 10x (2000 daily cases), it would still show 881,000 negative cases. If it went from 881,000 negative to 879,000 negative, should we panic?

On your site, El Paso is still at around 98% negative tests, which makes it appear like everything is under control when in reality the hospitals were mostly overrun and had to send patients outside of the county.

> What's the issue with simply reversing that metric to show many people are COVID negative?

There are three groups. Positive, negative, and untested. The untested group is the biggest, and you can't call them COVID negative, because there's really no reason. They're not (yet) either positive or negative.

I used the term "COVID negative" in an earlier post and that was a huge mistake and that mistake has occurred only on this HN thread. I added an edit to that comment to say I was wrong.

No where on thecovidcomplex.com do I use the term "COVID negative".

I had no idea it was your site.

I didn't, and cannot, downvote. However my BS detector immediately went off when I read the phrase "99.798% of residents in San Francisco County have NOT received a positive COVID-19 test." That percentage doesn't really mean anything in this context. More meaningful might be number of positive test results as compared with total number of tests I suppose.

You should be skeptical of everything you read on the internet.

I applaud people who want to confirm what I have provided. All data is available on the page and I’ve provided links to all sources.

Oh, woah, it's not about confirming what's provided or believing everything on the internet. I'm actually giving you the benefit of the doubt that the raw data is correct, and that the statistic you've provided is correct. I'm questioning why you would give that statistic given the misleading nature of it -- and no, other people doing misleading things is not a good reason (as you imply in other comments).

What proportion of residents of other California counties have not received a positive test? Other states?

It’s all there.

https://thecovidcomplex.com/states

Reply on news.ycombinator.com