 
     |
| Total Deaths vs. COVID-19 (Page 5/11) |
|
gtjoe
|
OCT 09, 11:19 PM
|
|
| quote | Originally posted by theBDub:
This is... literally what I did. Look at what I posted... the only difference being 2020 isn’t over, but why does that matter? Deaths aren’t final for over 2 years after the conclusion of a year. Up until then, it’s all provisional, the same data I’m using week over week. |
|
deaths above expected is a long term statistic used to look at a historical event to estimate how many deaths it caused. It really only works when looking back at an event that is over. To give you an example from your spreadsheet if you look at the first 16 weeks the 2020 excess deaths are 84,253. If you extrapolate that over an entire year it would predict 1,095,289 excess deaths for the year. The current number extrapolated over the entire year would be 299,154. If the theory that it many of the deaths were only accelerated by a few months is true you wont know until later because the death rate after the pandemic is over will be lower than expected.
also of note is that the baseline is on a general upward trajectory. If you only use the first three years of your chart the deaths above expected are 268,463 and if you use only the most recent 3 years its 180,269 which is a pretty significant difference.
I do not think that covid is not a big deal, but there are a lot of people who are hyping it to be worse than it is. Its not a conspiracy, its a combination of various people looking out for their own self interests. Most of these self interest include making money. some of them involve politics.
Im not a statistician, I am only looking at the data you presented, and pointed out some things that seem to be logical.
|
|
|
|
Hudini
|
OCT 10, 12:33 AM
|
|
|
|
|
theBDub
|
OCT 10, 11:23 AM
|
|
| quote | Originally posted by gtjoe:
deaths above expected is a long term statistic used to look at a historical event to estimate how many deaths it caused. It really only works when looking back at an event that is over. To give you an example from your spreadsheet if you look at the first 16 weeks the 2020 excess deaths are 84,253. If you extrapolate that over an entire year it would predict 1,095,289 excess deaths for the year. The current number extrapolated over the entire year would be 299,154. If the theory that it many of the deaths were only accelerated by a few months is true you wont know until later because the death rate after the pandemic is over will be lower than expected.
also of note is that the baseline is on a general upward trajectory. If you only use the first three years of your chart the deaths above expected are 268,463 and if you use only the most recent 3 years its 180,269 which is a pretty significant difference.
I do not think that covid is not a big deal, but there are a lot of people who are hyping it to be worse than it is. Its not a conspiracy, its a combination of various people looking out for their own self interests. Most of these self interest include making money. some of them involve politics.
Im not a statistician, I am only looking at the data you presented, and pointed out some things that seem to be logical.
|
|
Thanks for clicking around! It’s kinda fun, right?
A few important points to make. This is provisional data, meaning not final, it gets updated weekly, including past days. 2018 is even still technically provisional (though rarely updated). Due to this being provisional, I recommend excluding a few of the last few weeks. The most recent 3 are all but useless, 4-5 get better. I think by week 6 they estimate it’s at least 80% accurate (going off memory, but the CDC has these listed if you want me to track it down). So if we toss out the last 4 weeks, it actually goes up. 274k excess deaths with 2014-2019 being the baseline for comparison.
Now, you are correct, if you play with the years, the excess deaths will change because the baseline changes. It’s not just all going up, though. It’s really predominantly due to some bad flu years in 2017 and 2018, which you can track in the bottom left chart.
I first made a static chart in Excel comparing against just 2019. It got attention because it was easy to understand, but it was misleading because I hadn’t controlled for population growth, 2019 could have been an outlier, and I didn’t toss the most recent weeks because I didn’t want to seem partisan.
I made this with those fixes, but I wanted to lead others to draw their own conclusions. It’s not up to me to decide what weeks to include, others should decide for themselves. It’s not up to me to decide if certain outlier years should be included or not, that’s up to you and how you want to explore the data. I just wanted to present it in a way that makes sense and is easy to digest.
My favorite part of this is looking at the state breakdowns. You can see New York’s clear spike and response. You can track when measures were implemented in certain states that helped or accentuated the spread. It’s super interesting to me.
But at the end of the day, if I take every liberty possible to downplay the data, choose every filter I can that lowers excess deaths, I’m still at least 170k excess deaths so far this year... which is unacceptable.
|
|
|
|
82-T/A [At Work]
|
OCT 10, 11:26 AM
|
|
| quote | Originally posted by theBDub:
It’s not an article lol it’s a Power BI report. Just click the link. You’ve spent far more time typing than the <30 seconds it would take to look at the visuals lol.
I made the report because many people believe as you do—that the deaths are exaggerated. There is a logical explanation for why you feel that way, including videos you’ve linked to me.
But when you look at the actual data of all deaths, regardless of any cause, just all deaths, it shows that the reported COVID-19 deaths are not only directionally correct, but possibly understated. In effect, you are wrong. It’s definitely confusing to sift through all of the media reports, but the deaths are truly and actually (mostly) as stated. |
|
Unfortunately, what you've done here is seek to support an answer you'd already pre-determined, and in doing so you've invalidated anything you essentially would have done here. You're not impartial, your goal is to express the significance of the death count for political purposes, and you're seeking to prove it based on correlations that only support that. There are a few things I've noticed about your dynamic report:
- You do not state percentage increase per year of your baseline... it should go up every year on a sliding scale.
- You do not attempt to break out any additional information (suicides, overdoses, etc.).
- To be statistically correct, you'd need to correlate other significant death results (car crashes, natural disasters), etc., and not just mortality from other illnesses which the CDC tracks. This is how you determine trending and any potential correlation. For example, during lockdown, I'd expect some to go down (car crashes) while others would go up (heart attacks).
- The only thing I can make any causation from is the Natural Causes designation, and I can see a spike in death from natural causes at the same time as COVID spike during lockdown. What is this saying? It could suggest that maybe COVID was undercounted, but it could also show that suicides and drug overdoses increased dramatically during lockdown, and that they were being counted as COVID deaths.
Again, it seems to me that it's politically important for you to tell us all how many people have died from COVID, that's your first problem. This isn't as scientific as you think, your judgement is clouded in your politics. The numbers will be easier to parse through at the end of the year when totals for various conditions are made definitive, but CDC's numbers are not a complete picture. They collect information from all over, but it is not a complete picture. You should include data sets, including those which extrapolate from the states themselves before CDC dumps them into random bucks. I believe CDC does of course collect deaths for car crashes (for example), but there's no such break-out. I don't expect them to list death by hot dog, but they should include asphyxiation.
Ultimately, your report is merely just a visual display of data from one data source that's been massaged. Not saying you're going to be able to get data as clean as this where you can literally just pull data from an API into BRIO or Crystal Reports or whatever this is (I don't do that stuff anymore), but for what you're trying to prove... you'd need to see either the complete data, or pull from more sources.
I've spent way more time on this than I care to any further...
|
|
|
|
theBDub
|
OCT 10, 11:30 AM
|
|
| quote | Originally posted by 82-T/A [At Work]:
Unfortunately, what you've done here is seek to support an answer you'd already pre-determined, and in doing so you've invalidated anything you essentially would have done here. You're not impartial, your goal is to express the significance of the death count for political purposes, and you're seeking to prove it based on correlations that only support that. There are a few things I've noticed about your dynamic report:
- You do not state percentage increase per year of your baseline... it should go up every year on a sliding scale.
- You do not attempt to break out any additional information (suicides, overdoses, etc.).
- To be statistically correct, you'd need to correlate other significant death results (car crashes, natural disasters), etc., and not just mortality from other illnesses which the CDC tracks. This is how you determine trending and any potential correlation. For example, during lockdown, I'd expect some to go down (car crashes) while others would go up (heart attacks).
- The only thing I can make any causation from is the Natural Causes designation, and I can see a spike in death from natural causes at the same time as COVID spike during lockdown. What is this saying? It could suggest that maybe COVID was undercounted, but it could also show that suicides and drug overdoses increased dramatically during lockdown, and that they were being counted as COVID deaths.
Again, it seems to me that it's politically important for you to tell us all how many people have died from COVID, that's your first problem. This isn't as scientific as you think, your judgement is clouded in your politics. The numbers will be easier to parse through at the end of the year when totals for various conditions are made definitive, but CDC's numbers are not a complete picture. They collect information from all over, but it is not a complete picture. You should include data sets, including those which extrapolate from the states themselves before CDC dumps them into random bucks. I believe CDC does of course collect deaths for car crashes (for example), but there's no such break-out. I don't expect them to list death by hot dog, but they should include asphyxiation.
Ultimately, your report is merely just a visual display of data from one data source that's been massaged. Not saying you're going to be able to get data as clean as this where you can literally just pull data from an API into BRIO or Crystal Reports or whatever this is (I don't do that stuff anymore), but for what you're trying to prove... you'd need to see either the complete data, or pull from more sources.
I've spent way more time on this than I care to any further...
|
|
I hear what you are saying, but if you can look at everything I posted and still say it’s due to other causes without bringing any proof, then I’m not the partisan one, Todd. Maybe consider your own biases here and how they might be influencing your opinion.
Edit: I’ll put it this way. If you’re going to suppose that it’s due to something other than (the very heavily correlated) COVID-19, then you bring the data and show me why I am wrong. If you’re willing to say it’s not COVID-19 and claim that it’s a combination of a bunch of other stuff, then prove it.[This message has been edited by theBDub (edited 10-10-2020).]
|
|
|
|
82-T/A [At Work]
|
OCT 10, 11:47 AM
|
|
| quote | Originally posted by theBDub:
I hear what you are saying, but if you can look at everything I posted and still say it’s due to other causes without bringing any proof, then I’m not the partisan one, Todd. Maybe consider your own biases here and how they might be influencing your opinion. |
|
No, I like what you did... I really do. I'm also impressed with how much better dynamic reporting has gotten. I haven't touched that stuff in over a decade, and it's wildly better, so props on that. I think CDC's influence in extrapolating the data into the designations they do, makes it difficult to determine the kinds of things that you (and myself) would be interested in, in this case.
Looking at the data, I have more questions than I do answers. That natural causes spike during lockdown either shows COVID deaths are undercounted, but it could show they're overcounted. Either way... I know COVID is bad... worse than the normal influenza virus. We'll have a clearer picture after the year is up.
|
|
|
|
theBDub
|
OCT 10, 03:37 PM
|
|
| quote | Originally posted by 82-T/A [At Work]:
No, I like what you did... I really do. I'm also impressed with how much better dynamic reporting has gotten. I haven't touched that stuff in over a decade, and it's wildly better, so props on that. I think CDC's influence in extrapolating the data into the designations they do, makes it difficult to determine the kinds of things that you (and myself) would be interested in, in this case.
Looking at the data, I have more questions than I do answers. That natural causes spike during lockdown either shows COVID deaths are undercounted, but it could show they're overcounted. Either way... I know COVID is bad... worse than the normal influenza virus. We'll have a clearer picture after the year is up. |
|
Yeah, I have thought about starting a consulting business with it. We are already at the point where the tools have surpassed the business use, it’s pretty cool, and honestly fun to tinker with.
|
|
|
Patrick
|
OCT 10, 04:14 PM
|
|
| quote | Originally posted by theBDub:
But at the end of the day, if I take every liberty possible to downplay the data, choose every filter I can that lowers excess deaths, I’m still at least 170k excess deaths so far this year... which is unacceptable.
|
|
And it's interesting the number of people who apparently find your current presence here to be "unacceptable". Even though your long term ratings are very much positive, the ratings since you've returned to the forum (nice to see you back) have taken a drastic turn for the worse. I haven't seen you involved in any tiff. You've simply been supplying data. Is this perhaps a case of the messenger being shot? 
 [This message has been edited by Patrick (edited 10-10-2020).]
|
|
|
|
maryjane
|
OCT 10, 04:50 PM
|
|
|
|
Patrick
|
OCT 10, 05:08 PM
|
|
| quote | Originally posted by maryjane:
People don't want data. They want hyperbole.
Data is difficult to disprove where as hyperbole can just be repeated over and over without having to prove anything.
|
|
Yeah, that would certainly explain the ratings of at least one member here.
|
|
|
|
