This chart is NOT what it suggests it is. Pass it on.

[Ms. Toad]

https://public.flourish.studio/visualisation/1712761/

Since I keep seeing posts about this - and lots of shock and horror - but very little recognition that the visualization presents a false narrative.

The causes of death, other than COVID 19 are AVERAGE daily deaths from that particular cause. What that means is that it is as if every single day that many people diedof that cause during the period of observation.

So - as if every single day 1,774 people lose their lives to heart disease.

In contrast, the COVID 19 death bar represents how many people died on a specific day (the day the line at the bottom of the visualization crosses). To find the average daily deaths (in other words, how long the bar should be to be comparable to the data represented by the other bars), you would need to divide the total number of COVID 19 deaths since some starting point by the number of days since that starting point.

The first COVID death was on 2/29. There are 40 days between 2/20 and 4/8 (inclusive). The daily average number of deaths over the period during which people have been dying of COVID 19 is 14788 (total COVID 10 deaths)/ 40 days = 369.7

That puts the endpoint on 4/8 less deadly than a stroke, and more deadly than Alzheimer's.

Less dramatic, but more honest.

[SheltieLover]

1. TY For explaining!

I appreciare you making sense of all these #s!

[mr_lebowski]

2. It's a legitimate 'point in time' comparison ...

As in, "On Day X, the death's from COVID-19 exceeded the daily averages for all other causes of death"

IOW, it's statistically very likely that COVID-19 was the #1 killer on Day X. How likely depends of course on the std deviations of the other causes

Not sure what other people are saying about it/how they interpret it, but this is clearly what's meant to be conveyed, and there's nothing dishonest about it as long as it's represented correctly.

[Ms. Toad]

3. If it was clearly labeled as such - so that people viewing it realized what they were viewing.

Newsweek reported on the visualization (without even giving a link to it) it as a representation of the average daily COVID 19 cases. So people reading Newsweek were told the conclusion was based on the average number of COVID cases.

In an earlier thread in DU, I started to respond to a question about it, and got all tangled up because I knew the numbers were not right based on what the visualization labels suggest it epresents (average v. average), or what the person asking the question understood it to be (average v average). It took me a trip back to the data to sort it out.

If you read the text on the chart, the ONLY explanation for the numbers is in the title - and identifies the data as averages, not point in time comparisons, or statistical likelihood. The implication when you put two things on the chart together for comparison they represent comparable things.

[mr_lebowski]

5. Alright. I will stipulate that it could be misinterpreted ... but I understood it quickly myself nt

[Ms. Toad]

7. I might have - if someone hadn't asked a question about it

I watched it, and thought, "Cool!" My impression, without thinking about it, was that averages were being compard to averages - since that's how you construct graphic comparisons.

Then I tried to respond to someone who was confused about the average number of COVID deaths in the visualization - and I took a second look at the numbers (with the reinforcement by the person asking the quesiton of the concept that it was averages) and said, "That's not right."

If, before the question came up, someone had asked me exactly what was being represented - I'm not sure what i would have answered had I taken a second look before the question identifying it as an average was asked.

[blm]

4. Plus, strokes and Alzheimer aren't contagious, so,

Damn straight this should be more dramatic.

[Alex4Martinez]

6. The graph displays a day by day look at the day-over-day number of deaths, not the running average.

I agree that it's misleading, but it's not inaccurate.

Moving the clicker along different dates, it's showing how many deaths on that date were added to the total.

In this sense there have truly been days that this number was higher than heart attacks and cancer.

Of course, the number will decline for Covid while the others will remain identical each day because they are averages, not daily actual counts.

[Ms. Toad]

10. Misleading in a chart becomes inaccurate

when it is misleading enough that it starts being quoted as fact (the average daily number of COVID 19 cases on 5/8 was 1970).

This fits that bill. Vertical represents different diseases, the horizontal represents the average daily number of deaths from certain health conditions. If you're running a horizontal bar for comparison, the implication is that it is also an average daily number of deaths from the disease it represents - unless you clearly label it something else. The author chose not to, likely becuase it is more dramatic using the absolute number of deaths on a particular day.

It is reasonably being interpreted as the average daily number of deaths from COVID (and republished by others - identifyint the numbers as a representation of the average number of daily deaths without a link to the chart).

[Alex4Martinez]

11. I agree, it's bad form to mix data types in one image or one video.

People will invariably come away with the wrong conclusion, or most of them.

[HarlanPepper]

8. I'm more interested in how this can be leveraged against Dump than accuracy

But I appreciate you pointing that out.

[unblock]

9. I think averaging it over 40 days is much more misleading.

If I pick a name by random, the chart tells me how likely they are to have died from covid-19, and on which day. It tells me they are more likely to have died on certain days than on others.

Average daily rate of death from the first death to today, I'm not sure what that tells me. What does the timing of the first death have to do with anything?


the death rate is not flat, it follows a curve. It was zero and eventually will be zero or close to it. But it is meaningful to compare the rate at particular points during the event to other causes of death that are more stable.