What makes for a stable marriage? Part 2

About a decade ago, the gossip on everyone’s lips was that “1/2 of all marriages in the U.S. end in divorce.” That factoid was later disproven, but it left a lasting impression on the eligible bachelors and bachelorettes of America. In an effort to not become a part of that statistic, I started doing a little research on what makes for a stable marriage in America.

Last month, I ran across an interesting study on divorce titled ‘A Diamond is Forever’ and Other Fairy Tales: The Relationship between Wedding Expenses and Marriage Duration. The authors of this study polled thousands of recently married and divorced Americans (married 2008 or later) and asked them dozens of questions about their marriage: How long they were dating, how long they were engaged, etc. After running this data through a multivariate model, the authors were able to calculate the factors that best predicted whether a marriage would end in divorce.

What struck me about this study is that it highlighted about a dozen predictors that correlate with stable or unstable marriages in the U.S. By popular demand, I’ve highlighted 3 more of the biggest factors below as a follow-up to Part 1. I highly recommend checking the study out yourself (linked above) to look at all of them.

First, I’ll orient you on how to read these graphs. The authors always chose one category as the “reference point.” That means that all of the other categories are compared to that category. Below for example, “59% less likely” means that couples who had a child before their engagement were 59% less likely to ultimately end up divorced than couples who did not have a child.

Having children with your spouse

We all know someone who was on the verge of a breakup or divorce until they announced that they were having a baby with their spouse. According to this study, having a baby with your spouse can decrease your chances of divorce by as much as 76% compared to couples who do not have children. Of course, having children within wedlock — another telltale sign of a well-planned marriage — reduces your chances of divorce moreso than having children before you tie the knot.

What’s particularly interesting, though, is that even having children out of wedlock still reduces your long-term chances of divorce. It seems that shotgun weddings are more stable than we would expect them to be!

marriage-stability-children

Being the same age as your spouse

Perhaps unsurprisingly, the larger the age gap between you and your partner, the more likely your marriage will end in divorce. Only being 1-5 years away from your partner is nothing to worry about, but if you’re old enough to be your partner’s parent, then your marriage might be in trouble.

Hugh Hefner, anyone?

Note: A previous version of this article showed a chart giving specific relative percent likelihoods of divorce occurring based on number of years married. The original authors of the study have pointed out that although there is a significant correlation between wider age gaps and increased divorce, it is not possible to determine the relative percent likelihood from their study. That is left to future research.

Having the same education level as your spouse

If you’re a PhD marrying a high school dropout, your marriage may be shakier than a marriage between two college graduates. It’s particularly interesting to note that the education difference matters more for women than men: Women are 50% more likely to end up divorced when there is an education difference versus men at only 32% more likely.

marriage-stability-education-difference

Important: correlation != causation

Of course, it’s important for us to keep in mind that these are all correlations with marriage stability, and they could be telling us any number of things. For example, the “having kids with your spouse” correlation could go either way: Either people in stabler marriages are more likely to have kids in wedlock, or people in less stable (unhappy) marriages tend not to have kids. All of the explanations I wrote above are my own interpretations of the correlations, but keep an open mind when thinking about what could really be driving these correlations with marriage stability.

Dr. Randy Olson is the Lead Data Scientist at Life Epigenetics, Inc., where he is bringing advanced data science and machine learning technology to the life insurance industry.

Tagged with: , , , , ,
25 comments on “What makes for a stable marriage? Part 2
  1. Dom says:

    Am I reading the last graph right? Is it saying people who have been married for less than a year have the greatest chance of divorce?

    I would have guessed the greatest chance of divorce would be a year or two into marriage.

    • Randy Olson says:

      See the comment below. 🙂

    • mcus says:

      No it doesn’t mean that. The probability bars represent the probability of a couple that has been married x years getting divorced at some time after those x years of marriage. For example, if you’ve been married 10 years, the chance of you never getting divorced is 94%, whereas if you’ve only been married 1 year the chance of never getting divorced is only 24% (and the chance of getting divorced is 76%)

      • PL says:

        Close, but not quite. By that logic couples who are just married would be 100% likely to get divorced. The absolute likelihoods are not given. Rather, 10 year couples are .06 times as likely to get divorced as just married couples, and 1 year couples are .76 times as likely.

  2. Charlie says:

    I don’t understand how the figures can be significantly different for men than for women. Same-sex marriages aside, wouldn’t we expect the effect to be the same for either gender? Are same-sex marriages having (surprisingly large) skewing effects?

    • Randy Olson says:

      It’s likely because there’s more divorces for one gender than the other in the data set. It’s entirely possible that some of the people surveyed had multiple divorces in the time period they were surveyed on.

      Also, there aren’t any homosexual couples included in this data set.

  3. whatever says:

    Will there be a part 3? If so, I’d like to nominate the number of sexual partners for inclusion:

    http://onlinelibrary.wiley.com/doi/10.1111/j.1741-3737.2003.00444.x/abstract

  4. zkfrm says:

    I am disappointed to see this series fails to mention female partner count and divorce risk as analyzed by e.g. Teachman, 2003.

    http://onlinelibrary.wiley.com/doi/10.1111/j.1741-3737.2003.00444.x/abstract

  5. Frank says:

    I think it might be more useful to study the personal traits, history and family structures of spouse-to-be. These are the areas I did not pay enough attention to when I agreed to marry.

    Here are some specific ones I wish I had spent more time on :
    1. are the parents of your spouse-to-be married and happy ?
    2. is your spouse-to-be an anxious or angry person ?
    3. what kind of sexual activity and abortion history did your spouse-to-be have before engaging with you ?

    I looked long and hard at these after my wife tried to kill me.

  6. DJ says:

    The graphs represent the underlying data inaccurately. For example in the second graph the 95% bar is only a little larger than the 39% bar. How hard would it be to make an accurate graph?

    • Randy Olson says:

      Errr… I think you’re reading the graphs wrong. All of the bars are relative to the reference point, so the 39% bar is 39% larger than the reference point bar. Similarly, the 95% bar is 95% larger than the reference point bar.

      • DJ says:

        I think you’re drawing the graphs wrong. If the graphs are supposed to be relative to the reference point then why is there no horizontal axis throughout the graph indicating this point of reference?

        Presentation matters. Your graph is not technically wrong, but it is highly misleading, because the greatest visual emphasis (area of the bars) does not visually correspond to how you yourself state the graph should be read.

        • PL says:

          It would be more misleading to do as you suggest because the reference bar is not a zero point, it’s a 100% point. These graphs are portraying proportions, not absolute differences.

          • DJ says:

            Well, if you really mean that, why not label the graphs accordingly? The labels should then be “100%” for the reference, “103%” for the 1-year age difference, “118%” for the 5-year age difference, and so on.

            Or you could draw a narrow blank white line in each colored bar at the 100% level; this gets you the best of both worlds, making it clear what the reference point is, while also clearly indicating the 100% base.

  7. Alex says:

    If you ignore the elephant in the parlor it reveals you came at this with an agenda. The single largest predictor by an enormous margin is if the woman is a virgin or not. Why ignore this? Oh, that’s right, because the entire progressive worldview revolves around the notion that virginity doesn’t matter in marriage. If that were to get out, then promiscuity would plummet and the rest of their agenda would fall apart.

    • Eirik says:

      Firstly, not a well established fact, but a finding of a couple of studies. There are many factors with larger influences (e.g. country of residence). Secondly, did you read the last bit about how correlation does not imply causation?

      Personal note: What kind of plan do you imagine that plummeting promiscuity rates would ruin?

    • Hector_St_Clare says:

      It matters for some couples, and not for others. I’d actually be fairly turned off if I was dating a virgin who wanted to save herself till marriage.

  8. K says:

    in regards to age difference, do you have seperate stats for cases where the woman is older vs. the man is?

    • Hector_St_Clare says:

      What I’ve seen in the past is that divorce rates are higher when the woman is older, and are actually lower when the man is older.

  9. princevinco says:

    Wish to add that marriage stability equally depends on how well conflicts are managed by couples. Poor conflict management most at times causes marriages to break down.

    Of a truth, most failed marriages started with little conflicts. If the conflicts were allowed to linger unresolved, it soon grow and grow. With time it soon grow out of proportion that it become unmanageable thereby causing the marriage to breakdown.

  10. Dan Jakubik says:

    This is a difficult thing to study, considering marriage is the most complicated of all human relationships. Many factors are at work here, affecting couples differently. Each marriage is unique. Emotional maturity and stability, adequate financial resources, financial stability, honesty, respectfulness, loyalty, trust, good conflict resolution skills, ability to compromise and good communication skills are all vital.

  11. HighpointerGeocaching says:

    What if you are a guy seeking to get married, but you want to marry a younger woman, for several reasons. First, I want to have children, but most women my age are past their childbearing years. Second. I am not attracted to most women my own age. I am a healthy, outdoor enthusiast, and the number of women who are similar to me in age who like to do what I do is relatively small. Third, most women my own age who are not presently married are divorced and have children, and I prefer to marry a woman who has not been previously married and I don’t want to be a stepparent to someone else’s children. As I mentioned before, I want to be a father to my own kids, and find a great woman to give birth to them and be their mother.

    What do I do about the above, if I want to get married and have children, and desire to marry someone who is relatively younger and physically attractive? I strongly desire to get married, but I also strongly want to avoid getting divorced.

  12. NikolaiG says:

    I have had 4 long term relationships. First a marriage to a woman within 5 years my own age. We separated and didn’t see each other after 3 years and legally divorced after 5 years. Then there were two long-term relationships with women within a year of my own age. They were such disasters I was relieved when they were over. All of this said, I want to point out that I cannot identify our similarity in age as a factor that helped or hurt these relationships. The uncompromising stubbornness of my first wife was the reason for our divorce, the hyper-jealousy and general anger at all men was the reason I ended the second relationship, and the unreliability and volatile personality of the third is why that one ended.

    Then I met the woman I am today married to and, 18 years after my divorce, I married her. My wife is 20 years younger than I am and after six years, we get along much better than I did with any of the above women, and have a child to boot. It is not her age that does it. It is her baseline positive, empathetic nature, lack of any general global gender based grudge, lack of selfishness and instinctive consideration for others, that makes a very stark contrast with any of the others. In all of these relationships age difference appeared to be irrelevant.

23 Pings/Trackbacks for "What makes for a stable marriage? Part 2"
  1. […] Randy Olson reviews more of a research paper that outlines what makes for a stable marriage in the U.S.  […]

  2. […] “Data tinkerer” Randall Olson, a computer science Ph.D. candidate at Michigan State University, already produced a practical guide to how your wedding day can correlate with the longevity of your marriage, and on Thursday, he published a list of four basic things that indicate a stronger marriage. […]

  3. […] “Data tinkerer” Randal Olson, a computer science Ph.D. candidate at Michigan State University, already produced a practical guide to how your wedding day can correlate with the longevity of your marriage, and on Thursday, he published a list of four basic things that indicate a stronger marriage. […]

  4. […] “Data tinkerer” Randal Olson, a computer science Ph.D. candidate at Michigan State University, already produced a practical guide to how your wedding day can correlate with the longevity of your marriage, and on Thursday, he published a list of four basic things that indicate a stronger marriage. […]

  5. […] Olson released another set of visuals—the most intriguing of which focuses on the matter of the age gap. A one-year discrepancy in a […]

  6. […] “Data tinkerer” Randal Olson, a computer science Ph.D. candidate at Michigan State University, already produced a practical guide to how your wedding day can correlate with the longevity of your marriage, and on Thursday, he published a list of four basic things that indicate a stronger marriage. […]

  7. […] “Data tinkerer” Randal Olson, a computer science Ph.D. candidate at Michigan State University, already produced a practical guide to how your wedding day can correlate with the longevity of your marriage, and on Thursday, he published a list of four basic things that indicate a stronger marriage. […]

  8. […] “Data tinkerer” Randal Olson, a computer science Ph.D. candidate at Michigan State University, already produced a practical guide to how your wedding day can correlate with the longevity of your marriage, and on Thursday, he published a list of four basic things that indicate a stronger marriage. […]

  9. […] To read the analyses in full, click here. […]

  10. […] Randal Olson is the one who analyzed the stats from Emory, making a graph that shows couples with a 5-year gap in age are 18 percent more likely to divorce, and those with a 30-year gap in age are a whopping 172 percent more likely to divorce. But even just being one year apart puts you at a 3 percent higher divorce rate.  […]

  11. […] Randal Olson is the one who analyzed the stats from Emory, making a graph that shows couples with a five-year gap in age are 18 percent more likely to divorce, and those with a 30-year gap in age are a whopping 172 percent more likely to divorce. But even just being one year apart puts you at a 3 percent higher divorce rate. […]

  12. […] he found money, engagement ring and wedding price, children, dating length prior to engagement, but part two of his data set reveals age is one of the biggest predictors of a lasting […]

  13. […] “Data tinkerer” Randal Olson, a computer science Ph.D. candidate at Michigan State University, already produced a practical guide to how your wedding day can correlate with the longevity of your marriage, and on Thursday, he published a list of four basic things that indicate a stronger marriage. […]

  14. […] latest set of stats to catch our eye comes, again, from data researcher Randy Olson, who has taken it upon himself to uncover all of the factors that influence marriage. This time […]

  15. […] being old enough to be your spouse’s parent is a good predictor of divorce. Data tinkerer Randal Olson drew this data from a recent study, called “’A Diamond is Forever’ and Other Fairy […]

  16. […] Olson released another set of visuals—the most intriguing of which focuses on the matter of the age gap. A one-year discrepancy in a […]

  17. […] words were born out by a recent Emory University study that showed the larger the age gap, the greater the chance of a marri… A five-year age gap statistically means you are 18 percent more likely to divorce (versus just 3 […]

  18. […] words were born out by a recent Emory University study that showed the larger the age gap, the greater the chance of a marri… A five-year age gap statistically means you are 18 percent more likely to divorce (versus just 3 […]

  19. […] words were born out by a recent Emory University study that showed the larger the age gap, the greater the chance of a marri… A five-year age gap statistically means you are 18 percent more likely to divorce (versus just 3 […]

  20. […] words were born out by a recent Emory University study that showed the larger the age gap, the greater the chance of a marri… A five-year age gap statistically means you are 18 percent more likely to divorce (versus just 3 […]

  21. […] words were born out by a recent Emory University study that showed the larger the age gap, the greater the chance of a marri… A five-year age gap statistically means you are 18 percent more likely to divorce (versus just 3 […]

  22. […] disproportion were innate out by a new Emory University investigate that showed a incomparable a age gap, a larger a possibility of a … A five-year age opening statistically means we are 18 percent some-more expected to divorce (versus […]

  23. […] words were born out by a recent Emory University study that showed the larger the age gap, the greater the chance of a marri… A five-year age gap statistically means you are 18 percent more likely to divorce (versus just 3 […]