totally agree here, it should be bar graph - from dataviz perspective. But I think that Ben H comment is much more valid from more basic perspective/foundation - the data itsefl is a bullshit (garbage) and after that the data suffers from censoring effect. On top of that author uses confusing choice of charts (but this is a minor issue) for musicians and general population lifespan.
I love your slides - great visuals and little to no text. Using Powerpoint to to support your lecture, not to be your lecture. However, you should try to get names right - it's Kenny (as in Rogers) not Kearney.
Yes, that's effectively correct. It's an odd term using the common meaning of "to censor," but statisticians have a second meaning of the verb: the value of measurement is only partially known at the time of assessment. Wikipedia has a good page covering different types of statistical censoring: en.wikipedia.org/wiki/Censoring_(statistics)
I just couldn’t stop thinking about all the jazz musicians during the birth of jazz who used hard drugs and often died quite young... i know the demographics and general audience of jazz has evolved a lot since then, but every genre has people that live dangerous lifestyles. I was wondering if the data is also not “weighted” or sampled in a way that reflects this? Like if you were to compare the first 20 years of rap to the first 20 years of jazz you might find that they have similar results! However, jazz has increased in popularity, which has diluted the number of jazz musicians with dangerous lifestyles...
The study seems to try to link mortality and music genres. It does this by using data samples that are inconclusive as stated in the lecture. I am asking myself, how does this get "good science" if you just merely state that "This pattern reflects, to some extent, a confound in the data", what doesn't mean something else than "most of the effect is most probably false". In my opinion, you need to exclude the data that is not conclusive or treat it differently. It's not being better science when you just state that most probably you are doing bullshit but still continuing doing so. And by the way, the source of the table and graph is cited to be from the author herself. (theconversation.com/music-to-die-for-how-genre-affects-popular-musicians-life-expectancy-36660)
You're absolutely right; they are being EXTREMELY charitable to the original author. It's completely immoral to run a study that is rendered completely pointless by confounding factors, and then just have a footnote saying so, and then go around publishing it everywhere as clickbait-bait. I would love to ask her, face-to-face, what the point of running that study was.
And this is how a lecture should be given !!! Please never stop teaching !!!
Another problem with the graph, I guess, is that it should be a bar graph not a continuos line graph.
totally agree here, it should be bar graph - from dataviz perspective. But I think that Ben H comment is much more valid from more basic perspective/foundation - the data itsefl is a bullshit (garbage) and after that the data suffers from censoring effect. On top of that author uses confusing choice of charts (but this is a minor issue) for musicians and general population lifespan.
The game was called Pitfall!. I've played it on a Commodore 64. Did I win anything? Great series BTW.
On Apple II :-)
I love your slides - great visuals and little to no text.
Using Powerpoint to to support your lecture, not to be your lecture.
However, you should try to get names right - it's Kenny (as in Rogers) not Kearney.
Does "right" censoring in this context mean "correct censoring" or "censoring on the right (not left) end of the spectrum"?
Yes, that's effectively correct. It's an odd term using the common meaning of "to censor," but statisticians have a second meaning of the verb: the value of measurement is only partially known at the time of assessment. Wikipedia has a good page covering different types of statistical censoring: en.wikipedia.org/wiki/Censoring_(statistics)
I just couldn’t stop thinking about all the jazz musicians during the birth of jazz who used hard drugs and often died quite young... i know the demographics and general audience of jazz has evolved a lot since then, but every genre has people that live dangerous lifestyles. I was wondering if the data is also not “weighted” or sampled in a way that reflects this? Like if you were to compare the first 20 years of rap to the first 20 years of jazz you might find that they have similar results! However, jazz has increased in popularity, which has diluted the number of jazz musicians with dangerous lifestyles...
The study seems to try to link mortality and music genres. It does this by using data samples that are inconclusive as stated in the lecture. I am asking myself, how does this get "good science" if you just merely state that "This pattern reflects, to some extent, a confound in the data", what doesn't mean something else than "most of the effect is most probably false". In my opinion, you need to exclude the data that is not conclusive or treat it differently. It's not being better science when you just state that most probably you are doing bullshit but still continuing doing so.
And by the way, the source of the table and graph is cited to be from the author herself. (theconversation.com/music-to-die-for-how-genre-affects-popular-musicians-life-expectancy-36660)
You're absolutely right; they are being EXTREMELY charitable to the original author. It's completely immoral to run a study that is rendered completely pointless by confounding factors, and then just have a footnote saying so, and then go around publishing it everywhere as clickbait-bait. I would love to ask her, face-to-face, what the point of running that study was.
PhD in stat. Why aren't you rich.
Post hoc ergo propter hoc!