Super interesting.
I have some qualms and questions though:
- Is it truly a random sample, using mechanical turk? Is 3,000 really enough respondents? They say mTurk itself is diverse, but do the respondents to their study match the overall characteristics? (In part I question this because the person who is doing mTurk for money may have not spent their money wisely, including on lavish weddings when they couldn't afford them...so just weighting it may not solve that issue) Also why exclude older people?
- A national, prospective, longitudinal study would be better, as they note
- I've read that people weren't getting divorced during the recession because they couldn't afford to. It's possible that people who didn't spend a lot want to get divorced, but feel they can't. I'd like to see some exploration of this. It may just be a blip though.
- I (skimmed but) didn't see how they controlled for wealth. Someone else's $50000 may be another's equivalent of $500. I'd be curious to see how this study looks when seen as a percent of income or wealth. We have a buddy that bought his now-wife a ~3 carat ruby ring...but he's got major bucks, so that probably costs less to him than my ring effectively cost us. And we're on our way to FIRE, so that ring cost is different to us than the hair on fire couple.