Quick data science / stats question: my feature of interest is 98% missing values with a NMAR structure. Can I just assume it's missing-at-random to estimate a summary statistic? Kthxbye!https://twitter.com/zeynep/status/1324150678642692096 …
-
-
Replying to @manatcell
Maybe even worse. It’s also x NMAR and y MAR, and x+y are 98. Now let’s see that summary statistic. Illusion of weightability. (Is that a word?).
1 reply 0 retweets 5 likes -
Replying to @zeynep
I kinda suspect that polarizing elections are likely to exacerbate structured non-response, and even if you guess the direction of the effect, the magnitude is still inherently unknowable just by polling. Just when you most want them to be accurate...
2 replies 0 retweets 3 likes -
Replying to @manatcell @zeynep
With multiple subpopulations (maybe most largely unknown) you may also run into a nested Simpson’s paradox scenario that confounds direction as well
1 reply 0 retweets 1 like
Replying to @BenjaminGVincen @manatcell
You could! It’s tea leaves all the way down.
6:30 PM - 4 Nov 2020
0 replies
0 retweets
2 likes
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.
