On Predicting and Preventing Divorce
I typically do all I can to avoid pop-psychology self-help stuff. This is because I invariably find it to be the intellectual equivalent of fingernails scraping across a blackboard. Self-help books abuse my sensibilities so deeply that I seethe with vexation and irritation for hours after putting them aside.
As an example, take John Bradshaw's bestselling Healing the Shame That Binds You, which recommends the toxic-shame-afflicted person to come up with affirmations like, "I, Ang, am often loving and kind", and then proceed to:
This is, of course, not to say that there is never anything of value in the self-help/personal growth industry. It's just that, when I take a look at that stuff, I feel like I'm wading through a sewer to find a nickel here and a quarter there. This is why I was pleasantly surprised by the low "scorn-factor" which John Gottman's 2000 book, The Seven Principles for Making Marriage Work, elicited in me. Not all was good, of course. There's quite a bit of triumphalist self-promotion in Seven Principles. For instance, Gottman starts out by saying:
The article describes Gottman's methodology. Situate couples (who have been separated for 8 hours) in a relaxed environment and videotape 30 minutes of conversation, divided evenly between random talk about their day and a focused discussion of a problem area of continuing disagreement. Then, code the conversations numerically in terms of the number of positive and negative vocalizations and facial/kinesic acts made by the husband and the wife. This generates scores on negative affect (criticism, defensiveness, contempt, and stonewalling) and positive affect (affection-caring, humor, interest-curiosity, and joy-enthusiasm). They then followed up with the couples 4 years and 14 years afterwards. After running statistical models on these data, Gottman found that high scores on the negative-affect model predicted early divorce (within the first 7 years of marriage), while low scores on the positive-affect model predicted late divorce (after the first child reaches 14 years). Overall, the model predicted divorce with 93% accuracy.
Pretty impressive results. The high accuracy suggests not only that the studied factors are significant factors for divorce, but also that other factors not studied (e.g., having common interests, conflict avoidance, personality problems, etc.) need not be considered for successful prediction. But unfortunately, a bit of additional searching turned up "The Hazards of Predicting Divorce Without Cross-validation", an article that shows that Gottman's work fails cross-validation tests. This indicates that Gottman got his 93% accuracy by over-fitting the data he had, so his models are of greatly reduced value for predicting divorce in situations outside the data set gathered in the study. To be fair, Gottman published Seven Principles a year before the article on cross-validation failure was published, so his exuberant optimism about his work was not dishonest. But this does show how fragile statistical investigations of human behavior can be. (I hear that Gottman presents a more sober appraisal of his work in his newest book, Ten Lessons to Transform Your Marriage.)
Generally speaking, it seems that scientific work in this and related fields makes the significant assumption that our minds can be investigated successfully through the observation and cataloguing of our behavior. I'm not sure how promising I should judge this approach to be. But it, and the neuroscientific approach, appear to be the only scientific methods we have to investigating human psychology.
Notwithstanding all this, Gottman poses an interesting model for marital success. Negative affect in conflict correlates with early divorce. So couples should avoid the criticism, defensiveness, contempt, and stonewalling that characterizes early divorcers. (Gottman goes as far as to say that there is no such thing as "constructive criticism" in a marriage.) Lack of positive affect correlates with late divorce. So couples should work to foster the attentiveness, transparency, fondness, mutual admiration, shared meaning, and openness to mutual influence which characterizes long-lasting marriages.
For instance, one sign that you have a healthy marriage, on Gottman's model, is that you don't exhibit stress indicators (elevated blood pressure, increased heart rate, etc.) when you discuss difficult areas of conflict with your spouse. A sign you have an unhealthy marriage is that you live parallel lives with minimal significant involvement with one another.
This is a plausible model. But I should point out that this is based on research which has exhibited only limited predictive success (limited by the cross-validation fault), and even if Gottman has gotten the predictive indicators of divorce right, it is debatable whether techniques designed to address those indicators will reduce divorce. Gottman spent 15 years studying divorce to get the data needed to develop his models. We'll need to wait at least that long to determine whether his therapeutic techniques are effective.
As an example, take John Bradshaw's bestselling Healing the Shame That Binds You, which recommends the toxic-shame-afflicted person to come up with affirmations like, "I, Ang, am often loving and kind", and then proceed to:
To my mind, such morbidly ridiculous behavior should induce far more shame on a person than any that it may (dubiously) manage to alleviate.
- Write each affirmation ten or twenty times.
- Say and write each affirmation to yourself in the first, second and third person.
- Continue working with the affirmations daily until they become totally integrated with your consciousness
- Record your affirmations and play them back when you can.
- Look into the mirror and say the affirmations to yourself out loud.
- Use visualizations with your affirmations.
This is, of course, not to say that there is never anything of value in the self-help/personal growth industry. It's just that, when I take a look at that stuff, I feel like I'm wading through a sewer to find a nickel here and a quarter there. This is why I was pleasantly surprised by the low "scorn-factor" which John Gottman's 2000 book, The Seven Principles for Making Marriage Work, elicited in me. Not all was good, of course. There's quite a bit of triumphalist self-promotion in Seven Principles. For instance, Gottman starts out by saying:
When I first began researching marriage in 1972, you could probably have held all of the "good" scientific data on marriage in one hand.... To address this paucity of good research, my colleagues and I have supplemented traditional approaches to studying marriage with many innovative, more extensive methods.... So far my colleagues and I are the only researchers to conduct such an exhaustive observation and analysis of married couples. Our data offer the first real glimpse of the inner workings---the anatomy---of marriage. The results of these studies, not my own opinions, form the basis of my Seven Principles for making marriage work [pp. 7-8].Such bold claims prompt skepticism, but I was willing to give Gottman some slack on this. After all, he does have his claim to fame: a >90%-accurate formula for predicting divorce from a few minutes of observation of a couple's conversation. This empirical success was enough to prompt me to look up his published research, The Timing of Divorce: Predicting When a Couple Will Divorce over a 14-Year Period, in the peer-reviewed Journal of Marriage and Family.
The article describes Gottman's methodology. Situate couples (who have been separated for 8 hours) in a relaxed environment and videotape 30 minutes of conversation, divided evenly between random talk about their day and a focused discussion of a problem area of continuing disagreement. Then, code the conversations numerically in terms of the number of positive and negative vocalizations and facial/kinesic acts made by the husband and the wife. This generates scores on negative affect (criticism, defensiveness, contempt, and stonewalling) and positive affect (affection-caring, humor, interest-curiosity, and joy-enthusiasm). They then followed up with the couples 4 years and 14 years afterwards. After running statistical models on these data, Gottman found that high scores on the negative-affect model predicted early divorce (within the first 7 years of marriage), while low scores on the positive-affect model predicted late divorce (after the first child reaches 14 years). Overall, the model predicted divorce with 93% accuracy.
Pretty impressive results. The high accuracy suggests not only that the studied factors are significant factors for divorce, but also that other factors not studied (e.g., having common interests, conflict avoidance, personality problems, etc.) need not be considered for successful prediction. But unfortunately, a bit of additional searching turned up "The Hazards of Predicting Divorce Without Cross-validation", an article that shows that Gottman's work fails cross-validation tests. This indicates that Gottman got his 93% accuracy by over-fitting the data he had, so his models are of greatly reduced value for predicting divorce in situations outside the data set gathered in the study. To be fair, Gottman published Seven Principles a year before the article on cross-validation failure was published, so his exuberant optimism about his work was not dishonest. But this does show how fragile statistical investigations of human behavior can be. (I hear that Gottman presents a more sober appraisal of his work in his newest book, Ten Lessons to Transform Your Marriage.)
Generally speaking, it seems that scientific work in this and related fields makes the significant assumption that our minds can be investigated successfully through the observation and cataloguing of our behavior. I'm not sure how promising I should judge this approach to be. But it, and the neuroscientific approach, appear to be the only scientific methods we have to investigating human psychology.
Notwithstanding all this, Gottman poses an interesting model for marital success. Negative affect in conflict correlates with early divorce. So couples should avoid the criticism, defensiveness, contempt, and stonewalling that characterizes early divorcers. (Gottman goes as far as to say that there is no such thing as "constructive criticism" in a marriage.) Lack of positive affect correlates with late divorce. So couples should work to foster the attentiveness, transparency, fondness, mutual admiration, shared meaning, and openness to mutual influence which characterizes long-lasting marriages.
For instance, one sign that you have a healthy marriage, on Gottman's model, is that you don't exhibit stress indicators (elevated blood pressure, increased heart rate, etc.) when you discuss difficult areas of conflict with your spouse. A sign you have an unhealthy marriage is that you live parallel lives with minimal significant involvement with one another.
This is a plausible model. But I should point out that this is based on research which has exhibited only limited predictive success (limited by the cross-validation fault), and even if Gottman has gotten the predictive indicators of divorce right, it is debatable whether techniques designed to address those indicators will reduce divorce. Gottman spent 15 years studying divorce to get the data needed to develop his models. We'll need to wait at least that long to determine whether his therapeutic techniques are effective.
posted by Ang at 1:26 PM
0 Comments:
Post a Comment
<< Home