Close Menu
    Daten

    Amerikaner, die ihren Partner online kennengelernt haben: Vorsicht beim Glätten [OC]

    Share.
    View 28 Comments

    28 Kommentare

    2. hbarSquared on

      But has it gone down since? Can’t really claim it’s an artifact unless you can show behavior has returned to baseline.

    3. thetreecycle on

      I see the point but can’t we include data from the past few years? I feel as though it would be relevant.

    4. NerdMachine on

      How convenient that the data ends 4 years ago and leaves out more recent data that might contradict OP’s thesis.

    6. Rasutoerikusa on

      Why did you stop the „without covid“ line before the „incl covid“ line? That doesn’t really make any sense since the two graphs stop at a different point in time. Can’t really compare those in any way. Of course they will differ a lot, if their endpoint is completely different. I belive your point is valid though, but this doesn’t prove it a meaningful way

    7. Monster_Dumps_2026 on

      This is basically a chart showing the death of „3rd Places“. The dating age population are drinking significantly less and in turn having significantly less social engagements

    8. LocusHammer on

      My wife was my literal first match on coffee meets bagel? haha. Great app. But this was in 2016. I deleted the app a week after we matched.

    9. viking_linuxbrother on

      Love to corrections. Data is a viewpoint and perspective is important.

    10. CLPond on

      COVID does represent an outlier and wonky data generally in most contexts, but why would that be the case for a lagging indicator such as where you met your partner? In 2020/2021, most newly married couples didn’t meet during COVID

    13. MobySick on

      My husband & I met on line in 1998. I was 40 & he was 43. We’re celebrating our 25th wedding anniversary this year.

    14. HopesBurnBright on

      Sure, but Covid did happen so it is true. Plus that is clearly increasing, so even if this is an actual mistake, just give it another 15 or so years and it will be over 50%.

    15. cryptotope on

      The problem with your interpretation is that you don’t seem to have access to post-COVID data in your underlying dataset.

      Yes, there was a sharp jump in the share of couples that met online during the pandemic. But the smoothed line you’ve drawn still lies below the actual reported data points for the pandemic years–if anything, the ’smoothing‘ softens the abrupt discontinuity rather than overstating it.

      The question that remains unanswered – and which is unanswerable with the data presented – is whether the abrupt pandemic-era change is wholly the result of temporary factors (lockdowns, etc.), or whether it represents a more permanent state of affairs for which the pandemic was merely a *catalyst* for the transition. The pre-COVID trend was certainly headed that way, after all; maybe the pandemic just accelerated an inevitable change.

      For comparison, look at data on how people pay for things in the United States: [Figure 3.](https://www.frbservices.org/news/research/2025-findings-from-the-diary-of-consumer-payment-choice) In 2019, 13% of payments were made remotely. In 2020, that jumped to 20%. People were doing a *lot* more online shopping and contactless purchasing during lockdowns. The share retreated a bit to 18% in 2021–but people never really went back to the old way, and the remote-purchase share of transactions has climbed every year since.

    16. f8Negative on

      I’ve been to a lot of weddings in the past 5 years…but the divorces/separations are also keeping up.

    19. NordWitcher on

      I don’t know if it means much but shortly after Covid I decided to try online dating for the very first time after a break up. honestly wasn’t expecting much but was surprised at how many matches I had. I’m over 3 months I had about 72 matches until I met my now ex. Since then I’ve gone back to it and it’s a lot harder meeting or matching up with people on there. 

    20. krefik on

      Regardless of over-smoothing, I am skeptical about the source dataset anyways – I have neither the access to the original data, nor time to hunt it down and analyze, but according to the webpage in the bottom of the chart, not insignificant part of the respondents has met online pre-1980 – for 1961 it was on par with the couples who met at the church. Either the data was not cleaned at all, or transformed in a really clumsy way.

    21. MattieShoes on

      What kind of smoothing is this?

      Normal would be a moving average, but a moving average isn’t going to show 59%.

    22. Ok-Perspective-1624 on

      Is this satire or am I just too autistic to understand the English here? I know what smoothing is in relation to a line fit to data, but what do you mean „careful with smoothing“ like it is a warning to the dating pool?

    24. davidswelt on

      So what is the generalizable learning here? That for time-series data, the last few data points can have a disproportionate and unreasonable impact on the shape of the curve, making extrapolation into the future unreliable? Because future data points don’t exist, they don’t add errors. Statisticians must have considered this, right?

      Or is this about black swan events?

    25. stoneimp on

      LOESS isn’t a model, it just tries to hint at structure. To be honest, neither ought to be used to describe data because it isn’t globally modeling anything, and local modeling is basically just over fitting a global model but trying to hide it. LOESS is like doing a polynomial of degree 50 and then saying you found a trend.

    26. KingsFan96 on

      I met my wife in 1998 online……..in an AOL chat room. Going to see the movie „You’ve Got Mail“ was one of our first few dates.

    27. Soft_Welcome_5621 on

      Ok but Covid did happen… sorry why spend time thinking about alternate realities?

    Leave A Reply