Florida tears apart rainbow brick roads:

🄳🄾🄾🄼🄿🤖🅂🅃🄸🄽🄶

NEW: WSJ reports "Americans are having a record low amount of sex—even less than they did during Covid—according to a new study" from FamStudies. We find that

just 37% of adult Americans have sex weekly

🄳🄾🄾🄼🄿🤖🅂🅃🄸🄽🄶

When it comes to sexlessness among young adults, the biggest change comes post-2010.

Prior to 2010, the share of young adults, ages 18-29, who reported not having sex held steady around 15%.

But from 2010 to 2024, the share doubled, from 12% to 24% in the GSS.

This “hockey-stick” shape of young adult sexlessness "resembles the pattern identified by JonHaidt in The Anxious Generation." Haidt calls the period of 2010 to 2015 the “Great Rewiring.” "Adolescents going through puberty over this period were subject to all-pervasive digital media" which seems to have reduced their odds having sex in the real world.

🄳🄾🄾🄼🄿🤖🅂🅃🄸🄽🄶

One other key indicator of the decline in young adult human connection that grantjbailey & I document is reports of socializing w/ friends for young adults.

Socializing for young adults falls from 2010-2020, as both JonHaidt & jean_twenge would predict:

🄳🄾🄾🄼🄿🤖🅂🅃🄸🄽🄶

For adults in general, it is true that married men & women report more sex.

Almost one-half of married adults have sex weekly

Only about 1/3 of unmarried adults have sex weekly

🄳🄾🄾🄼🄿🤖🅂🅃🄸🄽🄶

Any questions?

🄳🄾🄾🄼🄿🤖🅂🅃🄸🄽🄶

But married sex has also been affected by the digital revolution.

The "sex recession" for married couples post 2010.

"It’s not surprising that more social media, Netflix, or gaming on the part of spouses translates to less intimacy."

🄳🄾🄾🄼🄿🤖🅂🅃🄸🄽🄶

This picture has been shared all around Japan Twitter lately

🄳🄾🄾🄼🄿🤖🅂🅃🄸🄽🄶

Concerning book cover, it is promoting interracial relationship! Right around time they are importing thousands of Nigerians to Japan. Strange timing

🄳🄾🄾🄼🄿🤖🅂🅃🄸🄽🄶

What if we control for which party governed the states during the time period that data is from?

That adds basically nothing. The benefit of this control is nonsignificant.

🄳🄾🄾🄼🄿🤖🅂🅃🄸🄽🄶

We can use the tidycensus R package to pull 2018-22 ACS data, which is almost the same year range for the rest of the data.

With that, we can add county sex ratios and median ages.

That adds basically nothing. No significant benefit.

🄳🄾🄾🄼🄿🤖🅂🅃🄸🄽🄶

We can use the same package to pull median household income, income inequality, poverty rates, unemployment rates, and attainment of Bachelor's degrees or great.

That adds something, but only median household income's effect is significant.

🄳🄾🄾🄼🄿🤖🅂🅃🄸🄽🄶

We can also pull county-level election results for 2020 off of GitHub, drop the state governor part, and see that...

The party supported by a given county doesn't significantly contribute.

🄳🄾🄾🄼🄿🤖🅂🅃🄸🄽🄶

And what about the split between governors and local governments?

Well, we can just add back the governor control from two plots ago, and see that

Though this model eventually has too many variables that aren't needed (like the various SES controls), I think it's still useful for seeing that partisanship fails to drive homicide rate differences across the U.S.

The far stronger driver is Black and Amerindian shares.

🄳🄾🄾🄼🄿🤖🅂🅃🄸🄽🄶

If we go back to the original model with just the shares, we still retain basically all the predictive power.

To get an idea how much these drive homicides, look at the predictor (Amerindians) net of the other shares versus the predictor (Blacks).

It's no contest.

🄳🄾🄾🄼🄿🤖🅂🅃🄸🄽🄶