[thomasfromcdnjs]

I don't really get these excess death calculations.

Surely just looking at the past 5 years is not enough.

Wouldn't you at least have to take into account generations? (ageing populations) (So maybe 60 years of data?)

The linked graphs could easily just represent an older generation dying off. (Even the humps could just be regular flu seasons for an ageing population)

Then you have to choose a range for your generation 50-80 year olds etc

If Covid kills a large proportion over 80, and your range for generations is 50-79, the data would be completely off yet again.

There must be a lot more facets to take into consideration.

I feel like those graphs quite statistically weak. What am I missing?

It would be nice to put my money where my mouth is and actually just do my own study. But can't think of a country who would have reliable stats going back 60 years on death tallies.

(Semi relevant paper on looking at ageing population stats -> https://www.karger.com/Article/FullText/494025)

[altacc]

Over 60 years you would have to account for the many other factors that also affected mortality, for example better healthcare. That would increase the statistical error and reduce the validity of the results. And it isn't really necessary as long as your baseline contains enough data to be able to prove statistically differences from your comparison groups. Generations are taken into account by splitting the results by age and not looking back too many years, as even with flu, cold winters, etc... life expectancy still changes slowly.

[benmarten]

Tend to agree, check out my sheet here: https://docs.google.com/spreadsheets/d/1dvfMYnPrb_Ln9qvEZVuL...

The mortality is as bad as in 2005 roughly!

[thomasfromcdnjs]

Can you elaborate a bit?

I find the +/- sentiments interesting, it wasn't any intention of mine to approve/disprove anything, I was just sitting at home thinking were the original charts meaningful.

[tumetab1]

> I feel like those graphs quite statistically weak. What am I missing?

I think you're just missing the definition of "excess death", it's literally how much more deaths were registered against an average of the last years.

It's just a basic indicator that indicates that something has happened and further investigation is required to clarify the reasons.

If the "excess deaths" was raised because the "baby boomers" generation was dying off a further investigation would be required to identify that.

[thomasfromcdnjs]

Thanks for the reply, stats aren't my strong point, I am sincere with my questioning.

My wording sucks, I don't think my personal judgement of "excess deaths" is "correct", — nothing like that.

It just kind of feels like, as an example, people trying to predict financial markets, throw in a few factors, and a few sentiments and thus, X does equal B^(Y * B)(blah % what)(huh ^ bleh).

The "further investigation" is the "investigation" I was originally begging the question for.

Not trying to antagonize you, I just don't really get why the originally linked article is "True", until further things are investigated.

Science is one phenomena that proves itself, that's great.

Stats seems very malleable, I've met some people who are good at statistics, I've personally visually plotted graphs too, and it all kinda seems bias. (I probably did them wrong)

Can any statisticians chime in? Is this a reasonable dataset?