memes. IG: dataforprogress
How was your analysis different from the original? Anywhere I can read more?
We re-created this NYT plot. The chart does not actually show a trendline and the relationship is mostly explained by Social Security.pic.twitter.com/MtFpWcIYUo
-
-
-
We have no idea what that the NYT trendline is supposed to be.
-
We have a blog in the works with a fun multilevel model. Everything is a bit slower cause of the holidays.
-
All that glmer()s is gold.
End of conversation
New conversation -
-
-
Cannot figure this out. Did they use a fitting technique that massively overweighted outliers? Bizarre.
-
I thought maybe the trendline was just for Trump counties but that would be weird.
- 1 more reply
New conversation -
-
-
Is social security not government assistance? Awesome and hilarious though thank you
-
It is - we’re just showing the extent to which the relationship is driven by older voters being more Republican.
End of conversation
New conversation -
-
-
What happens if you add the Clinton/Trump indicator as a group to geom_smooth?
-
this is the best I could do to reproduce that linepic.twitter.com/nNA5zhQtjw
-
In other words a regression of X on Y rather than Y on X...?
-
Not quite that simple, looks like. You might flip variables by accident, but you’d have to go well out of your way to do that (maybe easier in R than Stata, but still).
-
This is actually a really easy to do in R / ggplot2. You can fit a linear trend on a 2D scatter and then in one more line of code "flip" the coordinates to get a plot similar to the one above where trend line is minimizing horizontal squared error. Quick demo attachedpic.twitter.com/jgTUerK91k
-
Thanks. Figured it was easier in R.
End of conversation
New conversation -
-
-
-
Well done, and you didn’t have to mess around with any of that least-squares stuff
End of conversation
New conversation -
-
-
It looks like the NYT fit the trend line only through the counties Trump won, instead of the entire sample.
-
There look to be far more negative than positive residuals if we look just at the red dots...I don’t think that is it.
-
The outliers and the reds below the 50% (that are disappearing into the blue) make it difficult for me to judge that.
-
That extreme of a trend line is not explained in any way by that distribution of data. Maybe if they fit a line based on the min amount of government aid and max. Lol. Ignoring every other data point
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.