Thanks for the interesting analysis, but I wouldn't be so hasty in drawing any conclusions from the latest NSFG survey.
"One might argue that these recent declines in the NSFG on every metric should be interpreted cautiously; mostly because this is the first post-COVID NSFG."
The reason I would refrain from getting too excited over these apparent changes is that as the user's guide mentions:
'Data users are encouraged to exercise caution when making comparisons to prior years of data collection and interpreting any differences. The multimode design is very different from the prior FTF-only design, and the COVID-19 pandemic prevented experimental evaluation of the impact of the multimode design on survey estimates. Like other surveys that had to change to a new design due to the disruption of FTF interviewing, changes in estimates from before and after the pandemic are confounded by real changes in the population, measurement differences, and nonresponse, among other possible sources of error (coverage, sampling, and processing). As a result, some estimates may show larger differences from the 2017-2019 NSFG and earlier data compared to those observed between prior data releases .Users are advised to note the design-related changes, particularly with regard to mode, if making statements comparing earlier estimates to those from 2022-2023.'
I've added a section to this article showing why these changes aren't possible:
Basically, past a certain age, the virginity rates among people in the 2022-23 survey are higher than those of people five years younger in the 2017-19 survey (who would represent roughly the same cohort, making these changes impossible since you can't unlose your virginity).
Sure, this is a good observation. I overlooked it because I was mostly interested in just updating the data on my previous post and not looking into why there is a substantial increase. The change in older cohorts in particular could only be due to a change in how the survey was conducted. Next time there's a data update, I will be sure to include this note.
Some other comments if you're interested:
1. I know you're a fan of the NSFG. For whatever reason it appears to be an outlier relative to the YRBSS and GSS. However, I would be cautious on leaning on it; changes in sexlessness are expected from US social trends, like declining friendships, high schoolers going on less dates (Jean Twenge), declining adolescent birth rates (although this isn't as important). Chadopoly is obviously dumb though, nobody serious believes this.
2. It's difficult to certainly determine whether older cohorts are increasing in sexlessness.
We don't really have the data to confirm it. The GSS honestly sucks, and the YRBSS is cut off at ~18 yo, the NSFG doesn't show any real change otherwise. Worse still, whether you've had a new partner in the last year >30 is noisy (many are married, not receiving new partners), relationships are stable by then, and 98% have lost their virginity. The alternative is to measure sexual frequency, but that's not measured by NSFG.
That said, I would find it unlikely that the correlated societal trends mentioned in #1 wouldn't eventually affect the older cohorts eventually as the sexless youth ages.
'I know you're a fan of the NSFG. For whatever reason it appears to be an outlier relative to the YRBSS and GSS. However, I would be cautious on leaning on it; changes in sexlessness are expected from US social trends'
The YRBS trend from 2003-2019 isn't exactly astronomical. The virginity rate goes from 53.3% to 61.6%, and apparently saw no shift from 2003-2013.
The NSFG data also shows a small shift from 2002-19 for 15-18s, albeit slightly smaller.
In your GSS graph 'no partners' goes up about 3-4% among under 35s from 2002-2019? This is in line with the NSFG.
It might be harder to detect a significant change in the NSFG data using your method due to as you mention the smaller range of years, but the changes that are observed don't seem particularly anomalous.
I don't disagree that some changes should be expected. The claim was never that things are completely as they were decades ago, but that the GSS graph presented a misleading picture in both the scale and maleness of the changes.
It remains true that the proportional increase in sexlessness in the GSS from 2008-18 among men was far more extreme than it was in reality, and caused many misguided extrapolations. The NSFG sample size might be lower than the YRBS, but it's far higher than the GSS and allows us to more or less rule out the massive ~3x surge in the GSS as being real.
'Chadopoly is obviously dumb though, nobody serious believes this.'
Maybe to you, but I don't know where you've been if you think nobody seriously believes it. I'd go as far as to say it's pretty much mainstream at this point. Whenever threads around this kind of topic come up on reddit, there are always a flood of comments promoting it met with near unanimous agreement unless you're on a feminist subreddit or something. It's also regularly promoted by guests on podcasts such as diary of a CEO and other large platforms including CNN. Public intellectuals like Jonathan Haidt are beginning to promote it, and I have referenced several academic papers which have promoted it. It's massive.
'That said, I would find it unlikely that the correlated societal trends mentioned in #1 wouldn't eventually affect the older cohorts eventually as the sexless youth ages.'
I agree that it'd be strange if it didn't carry over at all. 'Delayed adolescence' will last longer for some than others, or even be permanent in some cases. It's not surprising to see more pronounced trends among adolescents though, which is why I don't think we can directly compare the YRBS and NSFG data and conclude that there is something wrong with the NSFG data dealing with adults.
Again, this somewhat minor critique aside, thanks for the analysis - it might be the most in depth one I've seen on this topic.
Yeah I agree with everything here. There is a 'statistically significant' shift but it's a slow multi decade trend.
As for Chadopoly; my emphasis is on serious as opposed to seriously. I would consider you serious and everyone that's loaded the data into a statistical program to see for themselves. That said - yes - all too many take it 'seriously,' including some serious grifters as you mentioned.
I appreciate the last comment. Keep up the good work on your end.
Oh I read it as 'seriously'. That makes more sense. Again though, figures who the public perceives as serious thinkers take it seriously, so I think it calls for a serious response.
An analytical tour de force, sincere kudos Uncorrelated - also glad to see Nuance Pill here, my first thought was interest about their take, given their previous takedown.
Thanks for the interesting analysis, but I wouldn't be so hasty in drawing any conclusions from the latest NSFG survey.
"One might argue that these recent declines in the NSFG on every metric should be interpreted cautiously; mostly because this is the first post-COVID NSFG."
The reason I would refrain from getting too excited over these apparent changes is that as the user's guide mentions:
'Data users are encouraged to exercise caution when making comparisons to prior years of data collection and interpreting any differences. The multimode design is very different from the prior FTF-only design, and the COVID-19 pandemic prevented experimental evaluation of the impact of the multimode design on survey estimates. Like other surveys that had to change to a new design due to the disruption of FTF interviewing, changes in estimates from before and after the pandemic are confounded by real changes in the population, measurement differences, and nonresponse, among other possible sources of error (coverage, sampling, and processing). As a result, some estimates may show larger differences from the 2017-2019 NSFG and earlier data compared to those observed between prior data releases .Users are advised to note the design-related changes, particularly with regard to mode, if making statements comparing earlier estimates to those from 2022-2023.'
I've added a section to this article showing why these changes aren't possible:
https://nuancepill.substack.com/i/153790248/impossible-changes-across-surveys
Basically, past a certain age, the virginity rates among people in the 2022-23 survey are higher than those of people five years younger in the 2017-19 survey (who would represent roughly the same cohort, making these changes impossible since you can't unlose your virginity).
Sure, this is a good observation. I overlooked it because I was mostly interested in just updating the data on my previous post and not looking into why there is a substantial increase. The change in older cohorts in particular could only be due to a change in how the survey was conducted. Next time there's a data update, I will be sure to include this note.
Some other comments if you're interested:
1. I know you're a fan of the NSFG. For whatever reason it appears to be an outlier relative to the YRBSS and GSS. However, I would be cautious on leaning on it; changes in sexlessness are expected from US social trends, like declining friendships, high schoolers going on less dates (Jean Twenge), declining adolescent birth rates (although this isn't as important). Chadopoly is obviously dumb though, nobody serious believes this.
2. It's difficult to certainly determine whether older cohorts are increasing in sexlessness.
We don't really have the data to confirm it. The GSS honestly sucks, and the YRBSS is cut off at ~18 yo, the NSFG doesn't show any real change otherwise. Worse still, whether you've had a new partner in the last year >30 is noisy (many are married, not receiving new partners), relationships are stable by then, and 98% have lost their virginity. The alternative is to measure sexual frequency, but that's not measured by NSFG.
That said, I would find it unlikely that the correlated societal trends mentioned in #1 wouldn't eventually affect the older cohorts eventually as the sexless youth ages.
Anyway, I appreciate your feedback.
'I know you're a fan of the NSFG. For whatever reason it appears to be an outlier relative to the YRBSS and GSS. However, I would be cautious on leaning on it; changes in sexlessness are expected from US social trends'
The YRBS trend from 2003-2019 isn't exactly astronomical. The virginity rate goes from 53.3% to 61.6%, and apparently saw no shift from 2003-2013.
https://yrbs-explorer.services.cdc.gov/#/graphs?questionCode=H56&topicCode=C04&location=XX&year=2023
The NSFG data also shows a small shift from 2002-19 for 15-18s, albeit slightly smaller.
In your GSS graph 'no partners' goes up about 3-4% among under 35s from 2002-2019? This is in line with the NSFG.
It might be harder to detect a significant change in the NSFG data using your method due to as you mention the smaller range of years, but the changes that are observed don't seem particularly anomalous.
I don't disagree that some changes should be expected. The claim was never that things are completely as they were decades ago, but that the GSS graph presented a misleading picture in both the scale and maleness of the changes.
It remains true that the proportional increase in sexlessness in the GSS from 2008-18 among men was far more extreme than it was in reality, and caused many misguided extrapolations. The NSFG sample size might be lower than the YRBS, but it's far higher than the GSS and allows us to more or less rule out the massive ~3x surge in the GSS as being real.
'Chadopoly is obviously dumb though, nobody serious believes this.'
Maybe to you, but I don't know where you've been if you think nobody seriously believes it. I'd go as far as to say it's pretty much mainstream at this point. Whenever threads around this kind of topic come up on reddit, there are always a flood of comments promoting it met with near unanimous agreement unless you're on a feminist subreddit or something. It's also regularly promoted by guests on podcasts such as diary of a CEO and other large platforms including CNN. Public intellectuals like Jonathan Haidt are beginning to promote it, and I have referenced several academic papers which have promoted it. It's massive.
'That said, I would find it unlikely that the correlated societal trends mentioned in #1 wouldn't eventually affect the older cohorts eventually as the sexless youth ages.'
I agree that it'd be strange if it didn't carry over at all. 'Delayed adolescence' will last longer for some than others, or even be permanent in some cases. It's not surprising to see more pronounced trends among adolescents though, which is why I don't think we can directly compare the YRBS and NSFG data and conclude that there is something wrong with the NSFG data dealing with adults.
Again, this somewhat minor critique aside, thanks for the analysis - it might be the most in depth one I've seen on this topic.
Yeah I agree with everything here. There is a 'statistically significant' shift but it's a slow multi decade trend.
As for Chadopoly; my emphasis is on serious as opposed to seriously. I would consider you serious and everyone that's loaded the data into a statistical program to see for themselves. That said - yes - all too many take it 'seriously,' including some serious grifters as you mentioned.
I appreciate the last comment. Keep up the good work on your end.
Oh I read it as 'seriously'. That makes more sense. Again though, figures who the public perceives as serious thinkers take it seriously, so I think it calls for a serious response.
Cheers.
An analytical tour de force, sincere kudos Uncorrelated - also glad to see Nuance Pill here, my first thought was interest about their take, given their previous takedown.
Fuck the internet