Saturday, July 14, 2018

You Don't Say

It was the kind of story that statisticians and business analysts hate. The Metro (UK) website carried it with a banner headline that proclaimed “Couples who spend more on their weddings are more likely to get divorced” in bold letters above a picture of a wedding cake. The story went on to claim that the more cash spent on wedding rings, engagement rings, or the ceremony itself, the more likely the marriage would eventually crash and burn, as illustrated by pictures of celebrity couples who had managed to do just that. It’s the kind of thing that would make anyone who couldn’t afford an elaborate wedding, anyone who thinks ostentatious displays of wealth are crass, or anybody who particularly dislike weddings feel better about themselves…

Unfortunately, the original article does not give the title of the study they are citing, the name of the journal in which it appears, or where the researchers got their information. To be fair, most people who are casually reading online news stories wouldn’t care about the statistical methods employed or the degree of empirical rigor employed by the research team; most Internet readers don’t appear to consider the source at all. The Metro (UK) site isn’t exactly the BBC, but they aren’t the Daily Mail, either, so I decided to run down the original research article…

As it turns out, the paper the Metro (UK) people were talking about is called “A DIAMOND IS FOREVER” AND OTHER FAIRY TALES: THE RELATIONSHIP BETWEEN WEDDING EXPENSES AND MARRIAGE DURATION, and it appeared in the October 2015 issue of a journal called Economic Inquiry, which is published by the Western Economic Association International. It’s the work of two economics professors from Emory University, Andrew Francis-Tan and Hugo M Mialon, and as you’d probably expect, it’s not quite as sensationalist as the Metro (UK) headline, or even its own title, would have you believe…

Basically, what the professors did was ask people to fill out a Qualtrics survey that provided basic demographic information, an approximate range of what they spent on the wedding, the rings, and such, and how long they had been/were married. They then cleaned and adjusted the data to account for as many inaccuracies as possible, and ran a number of regressions and other analyses to see if any patterns appeared. If you’ve spent any time on statistical research, you already know that there are problems with self-reported data (people lie) and correlational data (correlation does not equal causation, no matter how good your math is), but the researchers in this case weren’t looking for a causal relationship in this case…

According to the report, the researchers were attempting to determine if there was any relationship between elaborate weddings and/or rings and the length or success of the marriages, specifically because the companies that supply wedding services and supplies have claimed for most of the last seventy years or so that such expenditures are critical if you want to have a successful marriage. They specifically note in their literature review that these marketing claims are a recent development, and prior to World War II advertising of this type was highly unusual, and there does not appear to have been any corresponding popular belief…

Even more to the point, what the study found was that there was no such relationship. According to the data they collected, the researchers were unable to find any support for more expensive weddings or accessories leading to longer or more successful marriages. Although they did notice a number of other interesting results, as far as this study can determine, there is no reason to believe that spending a lot of money on your wedding will help you stay married, and no reason to believe that failing to do so will doom your relationship to divorce. If you dig down into the information, this turns out to be a moderately interesting study of changing consumer economics over the last two or three generations that debunks claims made by a specific industry about the vital importance of their products and services…

Although we should probably concede that just coming out and saying that would make it much less likely that anyone would read a news article about the findings…

No comments: