Profile picture
Paul Novosad @paulnovosad
, 21 tweets, 5 min read Read on Twitter
🚨 New working paper 🚨 on mortality change in the U.S.. In a graph: Rising mortality among non-Hispanic whites is almost entirely driven by the least educated 10%.
With @charlesrafkin. More discoveries below 👇🏼. Thread 1/N
Mortality among the least educated non-Hispanic whites is rising substantially at all ages. 2/N
Among the least educated non-Hispanic white women, mortality rates are rising from cancer, heart disease, deaths of despair (suicide, poisoning, cirrhosis) and other causes. 3/N
Why is this new? Because it’s very hard to measure mortality change in comparable education groups. Take high school dropouts — they were 16% of the female middle-age population in 1992 and 9% in 2015. 4/N
Even if the mortality *distribution* doesn’t change at all, mortality could still rise among HS dropouts. As the U.S. becomes better educated, dropouts represent a shrinking portion of the population — so they now have lower socioeconomic status than they did in the past. 5/N
This would be a selection bias, not a substantive mortality change — also called the Will Rogers paradox. This paradox explains why transferring the worst Dartmouth football player to Cornell will raise the average team quality of both schools. Think about it! 6/N
This selection problem is why Case and Deaton's work did not look specifically at dropouts — they couldn’t, precisely because of these sorts of concerns about selection bias. 7/N
The trick is to measure mortality change in constant percentile bins — e.g. to follow the least educated 10%, instead of high school dropouts. 8/N
This is still hard, because education is reported coarsely in the mortality data. But we came up with a way to do this in a separate paper with @thesamasher. 9/N dartmouth.edu/~novosad/asher…
With this solved, we can study mortality change in a constant education percentile group. Now we can make an apples-to-apples comparison: we can track mortality among the bottom 10% over time. 10/N
How much of the rise in mortality among the less educated is attributable to overdose and poisoning, suicide, and alcoholic liver disease, the so-called “deaths of despair”? 11/N
Among the least educated, deaths of despair explain all of mortality change only for the youngest. The least educated middle-age white women are dying more from many different causes. 12/N
Now let’s look at white women in education percentiles 10 to 45 — this is about a high school education in 2015. 👇🏼 13/N
Mortality is still rising among younger cohorts in percentiles 10 to 45, but not among older cohorts. Deaths from despair explain almost all mortality change in this part of the education distribution. 14/N
White women in the top 30% of the education distribution are doing totally great. 15/N
Education is a big predictor of mortality change among non-Hispanic blacks as well. This graph shows mortality change for each group of education percentiles among black women. 16/N
And non-Hispanic black men. Good news: mortality is plummeting among black groups outside of the bottom 10%. Bad news: gains are stalled among the bottom 10% for both men and women. 17/N
More details in the paper. Thank you for reading! 18/18
Thread Appendix 1: Here’s the raw data of white female mortality age 50-54 for each education level. The X axis shows the changing education rank, which is where the selection bias comes from.
Thread Appendix 2: We don’t cover Hispanics in this paper (like earlier work), because migration makes it hard to track mortality accurately. Hispanics tend to have much lower mortality than either non-Hispanic blacks or whites.
I forgot the link to the paper! Here it is. dartmouth.edu/~novosad/novos…. If only tweets were editable!!
Missing some Tweet in this thread?
You can try to force a refresh.

Like this thread? Get email updates or save it to PDF!

Subscribe to Paul Novosad
Profile picture

Get real-time email alerts when new unrolls are available from this author!

This content may be removed anytime!

Twitter may remove this content at anytime, convert it as a PDF, save and print for later use!

Try unrolling a thread yourself!

how to unroll video

1) Follow Thread Reader App on Twitter so you can easily mention us!

2) Go to a Twitter thread (series of Tweets by the same owner) and mention us with a keyword "unroll" @threadreaderapp unroll

You can practice here first or read more on our help page!

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just three indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member and get exclusive features!

Premium member ($30.00/year)

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!