FROM PAGES 30-35 OF SINGLED OUT:
Imagine that a team of scientists working for a pharmaceutical corporation is testing a new drug called Shamster, designed to increase people's health and well-being. Unlike most drugs, this one is not meant to remedy some illness; it is supposed to make your life better, whether you are sick or not.
The scientists are optimistic about the drug, but it is still experimental. They put out the word to physicians that Shamster is available to patients who would like to try it. Circumstances permitting, the patients get to decide whether or not they want to try it, and once on it, whether or not they want to continue. Over the course of the study, the people who were offered the drug end up in one of four different groups.
Some patients sign up for Shamster and stay on the drug for the entire course of the study. Scientists call this first group the drug group.
Some people never do take Shamster. Maybe they have some condition that disqualifies them from the study, or maybe they just don't want to try it - they are happy with their lives just the way they are, and they are not looking for any magic bullets. The scientists call this second group the no drug group.
Nearly half of the people who sign up for Shamster change their minds after they have been taking it for a while. The drug is not what they had hoped or expected. They just don't like it, and they are not going to take it anymore. Scientists name this third set of people the no drug--intolerable group.
In the last group are the people who started taking Shamster but had it taken away from them. Maybe their prescribing doctor retired and they could not find another doctor with access to it. Or maybe there were distribution problems in their area of the country. One way or another, through no choice of their own, the people who wanted to take Shamster and had already been taking it could no longer continue to do so. This fourth group is called the no drug--withdrawn group.
A total of 2200 Americans from 48 different states participate in the study. They all answer the question, "Taking all things together, how happy would you say you are these days?" People who say they are "very happy" get a happiness score of 4; those who are "pretty happy" get a score of 3; people who are "not too happy" are assigned a 2; and people who say they are "not at all happy" are scored as 1. Here are the average scores for each group.
3.2 No drug
2.9 No drug - intolerable
2.9 No drug - withdrawn
The scientists and the drug company want to celebrate. They notify the media that people in the drug condition are happier than people in any of the no-drug conditions. "See how wonderful our drug is," they boast. "If only more people would take our drug, they would feel so much better."
As a sophisticated consumer, you can see right through this. Sure, the so-called drug group looks better than the other groups, but that is because the sleazy drug company has classified all of the people who found the drug intolerable as a no-drug group. But they did take the drug! They stopped taking it because it made them miserable. The company did the same thing with the other group who did not fare so well - the people who had the drug withdrawn. That group was also labeled as a no-drug group and set aside. Then, the pharmaceutical company had the audacity to use the bad outcomes of the people who could not tolerate the drug, and who had it withdrawn, as evidence for the drug's effectiveness! "See," they said, "people who are currently taking Shamster are doing better than people who are not currently taking Shamster."
You can see what a fair conclusion would be, and it is not at all what the drug company says it is. What really happened is this: Of those who took Shamster, close to half of them could not tolerate it, stopped taking it, and ended up worse off than people who never took Shamster at all. The people who started on Shamster and had it withdrawn against their wishes also ended up doing worse than the people who never took Shamster. In fact, if you average the scores of the three groups of people who did take Shamster, at least for a while, you get the number 3.0. That is lower than the score of the group of people who spent their whole life Shamster-free, 3.2.
You simply cannot say that Shamster "makes" people happier. You cannot ethically suggest that if only more people would take Shamster, they would feel so much better - and the Food and Drug Administration would never allow any company to make such a claim. Close to half the people who start taking the drug feel so much worse that they just will not continue with it. The others who start on the drug and then cannot continue with it also end up worse than if they had never taken the Shamster at all.
Suppose, though, you think that you will tolerate the drug. You have confidence that if you sign up for the drug, you will stick with it, and no one will take it away from you. You have no real scientific basis for your optimism, and you have no control over whether the drug will be taken away from you. Still, you have a good feeling about your future on Shamster, so you go ahead and take it. Suppose you do end up in the best group: That's the group of people who start on Shamster and stay on it. They have an average happiness score of 3.3. Compare that to the score of those who never do take Shamster: 3.2. Statistically speaking, you are in the group that did better. But realistically, on the basis of a difference of one-tenth of a point on a four-point scale, would you really say that Shamster transformed your life? If you had to pick just one label (out of the four original choices) to describe the happiness of the people in each of the four groups, it would be the same each time. The average score of every group is closer to the "pretty happy" label (3) than to any other.
Even if the difference between the drug group and the no drug group had been sizable, though, you still could not confidently give credit to Shamster for making the people in the drug group happier. That's because the people who signed up for Shamster were different people than the ones who declined the drug. Maybe the Shamster-takers were happier than the Shamster-free even before the drug was offered to anyone, and taking the drug did not add to their joy. If so, then Shamster did not "make" them happier - they were already happier.
If the study were better designed, people would have been assigned at random to take the drug or not take drug. They wouldn't get to choose. That way, the two groups would, on the average, have the same initial scores on happiness and everything else. That's one of the hallmarks of respectable experimental research. A more rigorous study also would have assigned pills that looked the same to all of the participants. No one would know until after the study was over whether they had actually taken Shamster or a placebo that just looked like Shamster. That's important. If the drug company had blanketed the media with ads about the wonders of Shamster, people taking it might report greater happiness just because they knew that Shamster was supposed to make them feel that way.
In my hypothetical example, though, the scientists do make their bold and reckless claim. What's more, many other teams of scientists develop an interest in Shamster, and they conduct the same sort of flawed study. Not all of the teams get results identical to the first team's, but often the patterning is similar. Eventually, there are so many Shamster studies out there that you could write a book about it. So of course, some people did. Among them were one of the scientists who conducted Shamster studies and a columnist who always did like the idea of Shamster. Together, they reviewed a number of the studies that, in their opinion, made Shamster look good. The called their book, The Case for Shamster: Why People Taking Shamster are Happier and Healthier.
If the study I just described really had been a drug study, no respectable medical journal would ever publish it. No scientists worth their salt would want their names associated with it. Anyone writing a book with such a shameless and misleading title could kiss their credibility good-bye.
Yet this is just what much of the research on married and single people looks like. The study I described is one that actually has been conducted, much as I detailed it - 2200 participants from 48 states. I will call it the One-Time Happiness Study. The numbers in the table really are the results from the 2200 participants who answered exactly the question about their happiness that I described earlier. (The actual scientists, though, were more appropriately cautious about their conclusions.)
In the hypothetical study, I changed the names of the four groups. Here again is the same table, this time with the actual group names added to it.
3.3 Drug (Currently married)
3.2 No drug (Always single)
2.9 No drug - intolerable (Divorced) 2.9 No drug - withdrawn (Widowed)
When you hear claims about the latest study showing that married people are better off than unmarried people, you are probably learning about a study like this one. There would be nothing wrong with saying, on the basis of the data in the table above, that people who are currently married are, on average, happier than people who are currently unmarried - as long as you also made it clear that the biggest difference was between the currently married and the previously married, and that on a four point scale, the always-single group differed by just a tenth of a point from the currently married group. That would be a descriptive statement and accurate as far as it goes.
The enterprise becomes problematic when descriptive statements get recast as claims about the transformative power of marriage. When Linda Waite or Maggie Gallagher summarize a book full of results such as the ones in the table, and trumpet to the media the conclusion that marriage "makes" people happy or "improves" their lives or "is good for everyone," they are flat out wrong. Marriage was not good for the many people who could not stand it and divorced. In the end, perhaps it was not so good for the people who outlived their spouse. Even the married people who were skimmed off the top (the ones who never experienced divorce or the death of a spouse) ended up just a tad happier than people who had always been single.
To be fair, it is impossible to conduct perfectly rigorous scientific research on marriage. That's because it is impossible and unethical to assign people at random to be married or to stay single. It is also impossible to keep people unaware of whether they got the marriage drug or no drug. If you are married, you know it and if you are single, you know that; there is no placebo comparison group.
It is possible, though, to be clear and accurate in conveying the conclusions that can be supported scientifically by these studies. That's where the objectivity breaks down. The unsubstantiated claims are overwhelmingly ones that make marriage look better than it really is, and singlehood look worse.[END OF EXCERPT]
Later in that chapter, I explain the value of longitudinal studies (studies that follow the same people over time - for example, as they stay single, get married, or get unmarried). Those studies are better than studies of people at one point in time (the Shamster study) or studies of different groups of married and single people at different points in time (as in the study currently in the headlines). The very best studies methodologically, in which people are randomly assigned to marital status, are impossible to do.
Even if you've read just the brief excerpt I've posted here, maybe you can already tell what is wrong with the very first sentence of the press release about the recent study. Here it is: "For years, researchers have known that adults who have swapped rings say they are healthier than their never-married peers are."
I'll be back with my full analysis of that study (and the reports about it) soon.