maybe this
yoinked from data is ugly
https://www.reddit.com/r/dataisugly/comments/1ci8ny7/the_music_genreyear_19001980_x_axis/
wow
why are the population lines (US F/US M) jagged? does the horiz axis also have a year/time component for that data (while the year/time component doesnât apply to the genres)?
I have noticed busloads of geriatric women at all the Megadeath concerts I go to, and it does seem like they are not musicians themselves.
Horiz axis appears to be over time, which bears no relationship to genres.
In the comments, I found this lovely (and true) bit:
Why is it quite interesting? The average age of death from people in music genres in more recent decades is going to be lower⌠because⌠well⌠theyâre more recent decades.
This is the same error those stupid psychologists made 45 years ago when attributing left-handedness in males to early mortality. And some actuaries, people who study mortality, had to correct them.
Coren has also collected studies showing that the overall proportion
of left and right-handed people has not changed in the US during this century.
He has even surveyed art work in galleries and on the walls of caves to
see how many people are shown holding tools or weapons in their left hands.
Based on all these data, he concludes that the percentage of left-handers
has stayed almost constant through history. The reason there is such a striking
absence of elderly left-handers, Coren says, is because they must die early.
âWeâre not worthy!â to say anything about Alice.
That seems like the expected result assuming Fox News replaced an intern with AI.
So, having talked w/ someone involved in this sort of thing (not at Fox, but elsewhere)
obviously, they use graphic design software to put these together, like the Smart Art in Powerpoint (but itâs not Powerpoint, nor Smart Art) â and just type the numbers on there.
There is no connection between the numbers and the visuals. They just manually pick stuff or manually adjust angles.
In this case, itâs not even appropriate to do donut/pie charts, but something like clustered column charts for the info.
That looks like a standard donut chart from Powerpoint (I used one recently) and the donut reflects the numbers in the table behind it. There is just no requirement that the input numbers are %s that add up to 100%. The chart should do the math for youâŚand the splits look reasonable. Note: those would be two independent charts in this case.
Now is that what Fox used? No idea.
100% agree it is the wrong chart though.
I think I like these hexbin maps for categorical data and when size, shape, and orientation of the regions donât matter. Here is one I did in R: This represents percent margin in popular votes in one party over the other party.
The hexbins have their usefulness.
However, Iâm not here to discuss the mapping, I am here to discuss the bias in the selections.
I have found other sources that do not label some states as you have them labeled, which seems to imply your source has bias (or mine do)
AZ, PA, and WI are called âtoss upâ or âtoo close to callâ on 270towin.com and CNN.com
GA, MI, and NV are called pink not light blue on CNN.com
Also, NE and ME spread their electoral college (EC) votes by district, so they are not nearly as important as other states in the EC. Interestingly, the effect of this two-states-that-split seem to cancel each other out, so collectively they are even of lesser importance.
If everything else goes to form, my cross product of your map data and the constitutional EC votes by state seem to imply that those 6 states I mention should decide the election.
Please note that I dislike both candidates, probably wonât vote in the election, and I have to fight the urge to vomit any time politics enters the conversation. I do, however, respect & have interest in the election as a predictive modeling exercise.
Hi Deep. My source is based on historical 2020 popular votes, not any kind of forecast. My goal was to find a data source that would allow me to create four bins for four colors (recall that I have color issues, so four is enough for me!) and then produce a state hexbin in R, not to attempt to predict the next election.
Maybe Deep is referring to the Republican version of 2020 where Trump won the election.
Please do not mistake me for a Trump supporter. Nothing could be further from the truth. I had incorrectly assumed that the data was a prediction for the upcoming election. I had not read the title properly.
I recall your aversion to cast things in color, JT. Somewhat ironically coincidental, as an undergrad at the University of Illinois I studied under Dr. Ken Appel who was somewhat famous in math circles for the Four color theorem - Wikipedia
Sorry, that was not my intention. I did read that you were staying neutral. I wouldnât be surprised though if there existed an entire suite of alternative charts trying to back up the view that the election was stolen.
Hereâs a more funny (to me) map of the 2020 election that I think is more appropriately categorized as âAwfulâ. Itâs from wikipedia!
I think it sort of looks like a diagram of a beating heart, or a bat winged monster caught mid attack, or strawberry sauce on blue ice cream, or a manifestation of why my mother really didnât love me.
ugh