The 2014 study behind the “bigger wedding, bigger divorce” headlines didn’t find a straight line, and the researchers said so plainly in their own conclusion. Andrew Francis-Tan and Hugo Mialon, economists at Emory University, surveyed 3,151 ever-married US adults about their wedding costs and modeled how spending in different brackets related to the odds of divorce.
Women whose weddings cost $20,000 or more had roughly 1.6 times the divorce hazard of women in the middle bracket, $5,000 to $10,000.
Men who spent under $1,000 had roughly half the divorce hazard of men in that same middle bracket.
Neither fact describes a straight line running from “spent little” to “spent a lot”; together they describe a pattern where the bracket in the middle behaves differently from the brackets on either side, and where the effect doesn’t even move the same way for men and women in the same marriage.
What the actual bracket data shows
Francis-Tan and Mialon published their findings in Economic Inquiry in 2015, under the title “‘A Diamond is Forever’ and Other Fairy Tales.” Using the $5,000-to-$10,000 wedding-cost bracket as the reference point, they found men who spent under $1,000 had a hazard ratio of 0.492, about half the reference group’s risk, and men who spent $1,000 to $5,000 had a smaller, still-lower hazard of 0.782. Men in the $10,000-to-$20,000 bracket landed close to the reference group, at 1.067.
Only the top bracket, $20,000 and up, showed a higher hazard than the middle for men, and even there the increase, to 1.122, was modest. For women, the pattern looked different across the same brackets: 0.697 in the bottom bracket, 0.913 and 0.970 through the middle brackets, close to flat, then a sharp jump at $20,000 and up, to 1.595. The size of that jump, and the fact that it shows up much more strongly for women than men, doesn’t survive translation into “spending more causes more divorce.”
What the engagement ring numbers add to the picture
The same study looked separately at engagement ring spending, using $500 to $2,000 as the reference bracket, and the ring results undercut a simple story even further. Men who spent $2,000 to $4,000 on a ring had a hazard ratio of 1.334, roughly 33 percent higher than the reference group and the largest ring-related effect in the study’s main full-sample model, yet men who spent more than that showed no consistent climb: the $4,000-to-$8,000 bracket (1.193) was still slightly elevated, though not statistically significant, and only the $8,000-and-up bracket (0.884) dropped below the reference group.
Women’s ring-spending results didn’t show a statistically significant effect at any bracket. A relationship that spikes at one specific bracket and then drops back down as spending increases further isn’t a dose-response curve so much as noise with one loud exception, or a bracket boundary that happens to capture something the study wasn’t designed to measure directly, such as financial strain relative to a couple’s actual means rather than the raw dollar figure.
Why the flattened version of this study spread anyway
“Spend more on your wedding, raise your odds of divorce” is a much easier sentence to repeat than “divorce hazard varies by spending bracket in a pattern that isn’t monotonic and differs by gender,” and outlets covering the study’s release didn’t just simplify the finding, some inflated it. PBS NewsHour reported that spending more than $20,000 on the wedding raised the odds of divorce by 3.5 times compared with couples who kept it between $5,000 and $10,000.
The paper’s main full-sample model puts the figure at 1.6 for women. But 3.5 isn’t invented — it’s the paper’s own result for a different, smaller subsample: recently married women, where spending $20,000 or more carries 3.5 times the hazard of the $5,000-to-$10,000 group. PBS’s actual error was reporting that subsample-specific, women-only figure as applying to “couples” of both sexes generally.
CNN had run a simpler headline months earlier, when the study first circulated in October 2014: “Want a happy marriage? Have a big, cheap wedding.”Francis-Tan and Mialon’s own conclusion is more careful than either: marriage duration, they wrote, is “either not associated or inversely associated” with wedding and ring spending, a statement that rules out a strong positive line running from cost to risk without claiming that every low-budget wedding is safer than every expensive one.
What this means for anyone quoting a study instead of reading it
Bloggers and content creators cite statistics constantly, usually secondhand, and this study is close to a textbook case of what gets lost in that chain, and occasionally invented along the way. The popular version keeps a headline claim, sometimes with a number nobody can find in the original paper, and drops the part of the paper doing the actual analytical work, the breakdown by bracket that shows the relationship isn’t linear, doesn’t apply the same way to both partners, and is driven almost entirely by one bracket at the very top of the range. Repeating “bigger wedding, bigger divorce risk” without checking the brackets, or the arithmetic, is doing exactly what a source that repeats “research shows” without naming the research does: passing along a conclusion that sounds like the study’s finding but isn’t actually the shape of the data underneath it.
What survives contact with the real numbers is narrower and less quotable: very low spending is associated with somewhat lower divorce hazard, very high spending, north of $20,000, is associated with somewhat higher hazard for women specifically, and everything in between looks close to flat. That’s a real finding, backed by a peer-reviewed sample of over 3,000 marriages. It just isn’t the tidy warning about wedding budgets, inflated numbers included, that keeps getting attached to it.
