Math is hard; divorce is harder

Time for another GetStatistics post.

Check out this lede from the Daily Texan story that ran under the headline, “Conservative Christians divorce more, study says.” It’s almost an exact replica of how the mainstream media covered a porn study a few years back. Here’s how this sad example of a divorce story gets started:

Divorce is more common among conservative Christians and young people, according to a recent study.

University of Iowa sociology professor Jennifer Glass presented her study on skyrocketing divorce rates in regions highly populated with conservative Christians to an overflowing crowd in Burdine Hall on Friday.

“Politically and religiously conservative states, especially in the Deep South, exhibit higher divorce rates than politically and religiously liberal states in the Northeast and Midwest,” Glass wrote in her study.

Now, it is possible to compare the marriage rate of states that are deemed to be politically and religiously conservative with states that are deemed to be politically and religiously liberal. And hopefully a study making such a comparison would control for any number of variables (race, income, religiosity, etc.).

However, journalists must never forget this: It is never possible — unless a state is completely aligned with a particular religious or political group — to say that its divorce rate can tell you about the divorce rate of a particular subgroup.

Thus, the first line — that divorce is “more common” among conservative Christians and young people — does not match up with what the sociologist says, which is about state-level data.

The rest of the article is fine. It even explains that rates are lower in other states because of all the shacking going on. But I do think the reporter might want to search for better quotes:

Journalism graduate student Mark Coddington got married at age 22 and has been married for four years. He said his Nebraska hometown exemplifies the conservative Christian culture of Glass’s study.

“I have been around that culture, so I really understand why more and more marriages end the way they do,” Coddington said.

I’m assuming the Daily Texan is completely written and edited by student reporters. This would be a perfect example of an opportunity for journalism students to learn how to follow up with the Nebraska young marriage expert, seeking some specific details. Obviously, “And why is that?” might be a good follow-up question so we can get more additional facts. The reporter did at least speak with the director of the University Catholic Center about what marriage prep entails.

And I wouldn’t be commenting on this Daily Texan piece at all if it weren’t for the fact that I saw it picked up by so many other outlets.

Yet another bad journalism chain reaction.

Print Friendly

  • Sydney

    Making inferences about conservative Christians from facts about states with high rates of conservative Christians is especially dubious since there is good reason to think from much other sociological research that with respect to a variety of behaviours people in conservative states split into opposite extremes relative to liberals. For example, (if I remember correctly) conservative men who regularly attend church spend more time taking care of their children than liberal men while conservative men who do not regularly attend spend less time than either other group. I think something similar is true with respect to helping one’s wife, spending time with one’s wife, and so on. Maybe it’s also true of one’s likelihood to divorce! At any rate, lumping conservative men together means hiding much relevant information.

  • John Pack Lambert

    Another issue is that divorce rates are thrown out so we assume they are one thing, but they really are a different thing. Most people when they hear “divorce rate” think the figure is the percentage of marriages that end in divorce. In reality the divorce rate is the percentage of the population that is divorcing. The more married people you have the higher divorce rate you can have.

    Actually, even my earlier description is slightly misleading. It is the percentage of divorces compared to the total population. In theory one person can divorce multiple times in the same year. The likelihood of this is low, but how low I do not know.

    There is another question that I wish people would dig into. Are state granted annaulments counted as divorces? Either way they are counted it will complicate things, because while some annaulments end marriages that were total shams, others are used as cheap and quick ways to end marriages (especially any that use mental health issues to justify them, most claims of no knowledge of the mental health issues before hand, at least in my very limited experience, would not hold up if challenged). Also, if annulments are ignored there will be the factor that different states have different annulment laws. If you let people file for an annulment a whole year after marriage on mental health grounds in state x but state y does not allow such over three months after marriage, state x will probably have some annulments that would be divorces in state y. So until I know if annulments are counted I will figure that the journalists have not dug deep enough into the matter.

  • tioedong

    It might be because, unlike the more sophisticated, these folks actually marry before they live together.

    Where is the part about cohabitation rates in different states?

    Two: Notice that part about “young” marriages? Until you put that item into the statistics, you are not being honest.

    Three: Some girls in overly strict homes get pregnant and married as a way to leave home. I went to college, they get pregnant and marry. Same thing.

    Four: in other states, more women don’t bother to marry their baby’s fathers, so they don’t have to divorce.

    If you add these types of (common law marriage) relationships into your data, you are missing the real story.

  • Passing By

    Now, Mollie, if you had gone to school at U.T. Austin, this would be no surprise. We didn’t call it the Deadly Toxin for nothing. :-)

  • Hector

    Re: It might be because, unlike the more sophisticated, these folks actually marry before they live together.

    Perhaps, then, that’s an argument that more people should do things the ‘sophisticated’ way.

    If Massachusetts has a lower divorce rate than Texas, then that’s a good argument that Massachusetts is doing some things more right than Texas is. People who get married too early and too hastily have a high chance of getting divorced.

  • Julia

    We didn’t call it the Deadly Toxin for nothing.

    I give. What does that mean?

  • Passing By

    Daily Texan – Deadly Toxin. It was a play on words; it was the 70s and we were very cool 20-somethings.

    OTOH, the Texan in the 70s had the early Berkeley Breathed, pre-Bloom County. Funny, I remember his name as “Berke”.

    To understand journalism in Austin, the Texan or the American-Statesman, you have to know that it’s the bluest of blue cities in the reddest of red states, with a serious need to assert it’s blue nature. Moreover, Austin dances at the border of small town and big city, often drawing on the worst of both small and big. It’s shaped by the 800 pound gorilla of the University, giving the culture a further imprint of adolescence. The population is otherwise heavy on state government employees, many of whom chose their careers for the purpose of staying in the area. Austin is generally insufferably smug and superior, but then charming at the oddest moments. Which is true, to my observation, of the journalism there.

  • Charlie

    I read somewhere (and wish I could remember where) that the statistics on divorce among religious conservatives were skewed by the combination of two groups of people as religiously conservative. The operating definition of “evangelical” for the study was a person who had “made a decision for Christ;” that is to say, had gone forward for an altar call, prayed a prayer after hearing the Four Spiritual Laws, etc. But this group had two distinct components: 1) those who had had a conversion experience and who attended church regularly, and who were often active in the programs, etc, of their congregation. 2) Those who had had a conversion experience but who were not attending church at all, or who attended sporadically.

    The divorce rate of the first group was markedly lower than for the second, and indeed for any other group. The second group may have had a “spiritual experience” of some kind, but they had no commitment to biblical values or biblical behavior; their lives were shaped by the common culture, not by Christ.

  • Geroge


    FWIW, the technical term for the statistical error of drawing inferences about individuals from aggregate-level data is called the “ecological fallacy.”

    There are complicated ways of getting valid individual level results (imagine a statistical version of a CT scan), but they are not widely practiced.

  • Dan

    This is also reminiscent of the “Red State, Blue State” book which did the same sort of thing. But who is more at fault here, the author of the report or the journalist? It seems to me that it is the author of the report if, in fact, the report purports to draw conclusions about individuals from state level data. I have no training in statistics whatsoever but even I can see that conclusions about individuals based on state level data are useless. It is really hard for me to believe that a professor at a reputable university would do such a thing. If you wanted to get to the truth of the matter, it seems to me that it would be easy enough to do a study that compared divorce rates among, say, university educated upper-middle class Catholic couples who attended Mass together every week and university educated upper middle class couples from the same city who had once professed Catholicism but now professed atheism. As Charlie’s post suggests, there are probably several studies of this sort and I would be willing to bet that they show the religious divorce less.

  • SouthCoast

    Just out of curiosity, what percentage of the divorces are among the rising tide of West Coast transplants driven south by the economhy in the past 4 or 5 years?

  • Julia

    Passing By:

    Thanks for the explanation.
    Austin sounds like an interesting place.

  • Dandy Randy


    You didn’t even read the study, did you?

    Hint: you can get it here:

    Seems like you take a lot of swipes at the student journalist, but you don’t actually address the content or the quality of the study itself.

    Your piece reads like a case of shooting the messenger.

    It seems counter intuitive: religious conservatives maintain the sacrament of marriage as one of their central tenets, yet in communities where they number large, rates of divorce are higher than elsewhere. Those are the just the facts. The study is just trying to ask questions to figure out why this is the case.

    It’s pretty obvious to anyone who thinks about it (or just reads the study) why this should be the case for many of the reasons given in the article… but even if it were not the religious conservatives themselves who were divorcing, the question would still remain: what it is about their presence which is so corrosive to marriages around them?

    As an aside there is also overlap between high divorce rates and states which have amended their constitutions to outlaw gay marriage…

  • J

    Here’s another study which breaks down things in a way I can understand better than the Glass-Levchak study.

    Some of the findings:

    33% of all married persons have been divorced.
    34% of married Protestants have been divorced, 28% of married Catholics, and yet only 26% of married Evangelical Christians (who constituted only 8.9% of the sample). On the other hand, 32% of married born again Christians had been divorced, and 33% of married not born again Christians-dead heat, and about average for all adults. 30% of married atheists and agnostics had been divorced. Of course, how you define some of these terms gets interesting… The bottom line, I think, is that there isn’t that much variation in divorce frequency based on religion. The most profound effect on divorce seemed to be economic-39% of “downscale” (less than $20,000 annual income) but only 22% of “upscale” (over 75k annual income) had been divorced.

    The definition of Evangelical in the study was pretty strict. Quoting from the study, an Evangelical must be:

    believing they have a personal responsibility to share their religious beliefs about Christ with non-Christians; believing that Satan exists; believing that eternal salvation is possible only through grace, not works; believing that Jesus Christ lived a sinless life on earth; asserting that the Bible is accurate in all that it teaches; and describing God as the all-knowing, all-powerful, perfect deity who created the universe and still rules it today.

    Definition of born again? From the study:

    “Born again Christians” are defined as people who said they have made a personal commitment to Jesus Christ that is still important in their life today and who also indicated they believe that when they die they will go to Heaven because they had confessed their sins and had accepted Jesus Christ as their savior. Respondents are not asked to describe themselves as “born again.”

  • Aaron

    The unfortunate fact is that this is common among reporters, for obvious reasons. You take a relatively boring story on statistics and you extrapolate some “juicy” detail.

    Randy, makes the point that it is entirely accurate to link religious people with higher divorce rates in areas with higher percentages of religious people. I would think one would be better served investigating those links rather than insisting correlation.

    To put it another way, as I said when I blogged about this story and Mollie’s take: Could the same not be argued with liberal politics and violence? Areas that predominately vote liberal (urban centers) have much higher violent crime rates than other areas. So does that automatically mean that liberal voters are more violent or, as Randy put it, what is it about “their presence which is so corrosive to [peacefulness] around them?”

    One would, I hope, make the point that many other factors are at play in urban violence. Voting patterns are probably not the best indicators of a violent person. The same point should be made about divorce in religious areas. What other factors are influencing behavior? Do they carry more weight? Those are questions that should be dealt with before making assumptions about the supposed corrosive nature of religion on individuals marriages.

  • Bobby

    I did a series in 2002 on Oklahoma’s No. 2-in-the-nation divorce rate. I’ll copy the first part of the series below since I can’t find a link.

    Also, you may recall that the Southern Baptist Convention passed a resolution in June 2010 on the scandal of divorce. Among the parts of the resolution was this item:

    WHEREAS, Some studies have indicated that conservative Protestants in the United States of America are divorcing at the same rate, if not at higher rates, than the general population; and

    Alas, I think there’s a potentially great story on this issue for some enterprising reporter. But obviously the Daily Texan piece is not it, for the reasons cited by Mollie.

    Headline: Divorce rate stays steady, study shows
    Byline: Bobby Ross Jr.
    Publication Date: February 10, 2002
    Page: 1-A

    Three years after Gov. Frank Keating declared war on the state’s No. 2-in-the-nation divorce rate, the enemy shows little sign of retreating.

    Oklahoma’s number of failed marriages – about 20,000 a year – has remained fairly steady, state records show.

    For every 100 marriage licenses issued in 2001, the state granted 76 divorce petitions.

    Nevertheless, Keating and advocates of the $10 million Oklahoma Marriage Initiative point to progress that they hope will help reduce the state’s divorce rate by one-third by 2010.

    “Divorce is so imbedded in the culture, it’s going to be years before we turn it around,” Keating said.

    Spousal abuse, adultery and abandonment constitute legitimate grounds for divorce, the governor said.

    “But most marriages end because one party or the other is simply bored or decides that they want to have a new Jaguar,” he said.

    Among the progress cited:

    - About 750 clergy members statewide have signed the Oklahoma Marriage Covenant, agreeing to require a four- to six-month preparation period before presiding over a wedding.

    The covenant is important because an estimated 75 percent of first marriages occur in churches, synagogues or mosques, religious leaders say.

    Too often, Oklahoma churches have served not as promoters of lifelong marriages but as “wedding factories,” said the Rev. Kent Choate, the Baptist General Convention of Oklahoma’s family ministry specialist.

    - About 200 people from state government, the religious community, private counseling agencies and other sectors have trained to teach the Prevention and Relationship Enhancement Program, or PREP. An additional 100 to 300 people are expected to be trained by year’s end.

    PREP was developed by Scott Stanley and Howard Markman, who direct the Center for Marital and Family Studies at the University of Denver. The program, they say, is a “research-based, skills- building curriculum designed to help partners say what they need to say, get to the heart of problems, avoid standoffs and connect with each other instead of pushing each other away.”

    - Creation of a “statewide delivery system” to provide marriage education seminars and resources has begun, and the state is launching a Web site at

    - Oklahoma State University’s Bureau of Social Research has finished surveying 2,000 Oklahoma adults in an effort to explain the state’s divorce rate and build a foundation for assessing the marriage initiative’s long-term impact. A thousand adults also were questioned in Texas, Kansas and Arkansas to form a comparison group.

    While the complete survey report won’t be released until June, preliminary findings indicate most Oklahomans share Keating’s concern.

    Ninety percent of those surveyed said many couples rush into marriage, and 82 percent described a statewide initiative to promote marriage and reduce divorce as a good or very good idea.

    Sixty-nine percent called divorce a very serious national problem.

    “It’s interesting that over two-thirds of Oklahomans think divorce is a very serious problem,” said Christine Johnson, the OSU researcher overseeing the survey project. “Now, maybe we’re really poised to give Oklahomans some skills to make their relationships better.”

    Thirty-three percent of the respondents had been divorced at least once. Of the divorcees, 86 percent cited a lack of commitment as a major contributing factor.

    “That is something where the marriage initiative could make a difference,” Johnson said. “The notion of commitment is definitely… a component of the curriculum that they’ve chosen to use across the state.”

    Eyes on Oklahoma

    Experts cite the state’s low per-capita income – which ranks 43rd in the nation – and a tendency of Bible Belt couples to marry young as reasons many marriages fail.

    In recent years, various studies have ranked Oklahoma’s divorce rate among the highest nationally. A 1998 Family Research Council report said only Arkansas had a higher divorce rate, if you discount Nevada, where couples from across the nation flock for “quickie” divorces.

    The Oklahoma Marriage Initiative has put the state at the forefront of a developing national debate over government-sponsored marriage programs.

    In Washington, four think tanks – the liberal Urban Institute and Brookings Institution and the conservative Heritage Foundation and American Enterprise Institute – all have agreed that marriage should be supported and encouraged in public policy, Karen Woods said.

    “That is a 180 from the public debate just one year ago and a significantly important fact,” said Woods, the Empowerment Network’s vice president for state initiatives. She is a graduate of Seminole High School and Oklahoma State University.

    The Empowerment Network, chaired by Sen. Rick Santorum, R-Pa., and Sen. Joseph Lieberman, D-Conn., is made up of elected officials and grassroots groups that hope to promote families and improve poorer communities. Keating is among the organization’s honorary co-chairmen.

    In the White House budget plan sent to Congress last week, the Bush administration offered no new money to encourage job advancement. However, it proposed more than $100 million for experimental programs aimed at encouraging women on welfare to get married, The Associated Press reported.

    Two years ago, Keating became the first governor in the nation to set aside Temporary Assistance for Needy Families funds to strengthen marriages and reduce the divorce rate. Those funds are block grants provided to each state through the 1996 welfare reform act.

    Fortifying marriages was a major goal of welfare reform, but few states have acted on it, said Ron Haskins, senior fellow at the Brookings Institution and former staff director of the U.S. House Ways and Means welfare subcommittee.

    “Nobody has done as much as publicly and conspicuously as Oklahoma has,” Haskins said.

    Diane Sollee, founder of the Coalition for Marriage, Family and Couples Education in Washington, said, “All eyes are on Oklahoma, that’s for sure.”

    Praise for Oklahoma

    Now, it appears that President Bush would like other states to follow Oklahoma’s lead.

    “I think it’s quite exciting,” administration official Wade Horn said of the Oklahoma Marriage Initiative. “I think Governor Keating has shown real leadership and creativity on this issue, and we’re looking forward to seeing the results.”

    Horn, former president of the National Fatherhood Initiative, spoke at the Oklahoma conference on marriage hosted by the governor and First Lady Cathy Keating in March 1999. As assistant secretary for children and families in the U.S. Health and Human Services Department, he’s a key figure in efforts to broaden the focus of welfare to endeavors that foster marriage, abstinence and responsible fathers.

    “It’s not about mandating marriage. It’s not about running a federal dating service or a matchmaking business,” Horn told The Oklahoman. “It’s about helping couples who choose marriage for themselves to form and sustain healthy marriages.”

    Testifying last year before a congressional subcommittee, Jerry Regier, Keating’s former health and human services secretary, said Oklahoma spends millions on foster care, child abuse and neglect investigations, adoption, out-of-wedlock births, juvenile delinquency and many other problems. Regier characterized those problems as “primarily… the result of either families not forming through marriage in the first place or because of absent parents due to divorce.”

    A 2000 Heritage Foundation study, “The Effects of Divorce in America,” reported that state and federal governments spend $150 billion a year to subsidize and sustain single-parent families, but only $150 million – one-thousandth as much – to strengthen marriages.

    But others say welfare money should be reserved for programs that pay for job training, transportation, child care and increased wages.

    “I’m really worried about taking money away from poor people and using it for marriage education and marriage promotion,” said Dorian Solot, executive director of the Boston-based Alternatives to Marriage Project.

    Critics question whether government officials and taxpayer dollars belong in decisions so personal as marriage and divorce.

    Some also suggest that abused women may be urged to lock themselves in dangerous relationships. Advocates deny that.

    “I’m less skeptical than I was,” said Marcia Smith, executive director of the Oklahoma Coalition Against Domestic Violence and Sexual Assault. “I think Oklahoma has done a fair job of addressing that issue.”

    Pros and cons

    After Keating launched the marriage initiative, the state began calculating the incomes of both individuals in a cohabiting couple when determining welfare eligibility. That removed a financial incentive for couples to live together outside marriage.

    Also, the Legislature passed a bill lowering the price of marriage licenses for couples submitting to premarital counseling.

    But so far, lawmakers have rejected Keating’s calls to enact a covenant marriage law and outlaw no-fault divorce.

    At the same time, Keating has faced criticism from a few Democratic senators over a $400,000 contract awarded to Public Strategies Inc., the Oklahoma City company hired to manage the marriage initiative. The company’s president, Mary Myrick, is a former Republican political consultant.

    Sen. Kevin Easley, D-Broken Arrow, complained last year that Myrick’s company was paid for reading books, viewing videos and performing various public relations tasks tied to the governor’s program.

    Myrick said of Easley, “With better information, I believe he wouldn’t be a critic. He has focused on our contract and not on the marriage initiative. He actually criticized some of our expenses, which was a very, very tiny portion of what we did.”

  • Karen

    Uh Aaron,

    The statistics on violent crime are NOT higher in liberal cities. They are lower. NYC for instance has a crime rate that is a small fraction of that in Birmingham, AL; Dallas; Atlanta or Oklahoma City. Ditto Boston, although it has higher auto theft and robbery but lower murder or rape. As I recall, there was a study in the last year or so with similar conclusions about violence in the Southern conservative states as the current study on marriage. And the discussion in the marriage study on the socio-economic status, education levels, and culture of violence probably holds true for the crime study.

    The marriage study was far more nuanced than the article, breaking down divorce rates by denominational groupings, using county level rather than state rates. LDS membership was the only one religious affiliation that was correlated with lower divorce rates. I suggest that before speaking about the study, one take a look at it. The math isn’t that hard.

  • High Point

    As someone who uses statistics on a regular basis, I suspect that this issue is more complicated than the title of the study implies. Glass and Levchack appear to recognize the limitations of their study but did not have the resources to yield a more definitive conclusion.