Last week, on Facebook, a friend of mine linked to a Yahoo blog post whose title caught my eye: "Spanking Linked to Mental Illness, Says Study." My husband and I have a 14-month-old boy, whom we've never planned to spank. Still, we have years of discipline ahead of us, and I was curious about the findings.

So I read the story. Then I read the study it was based on. Then I got mad.

Despite the Yahoo headline, and many others like it, the study, published in Pediatrics in early July, does not actually link spanking to mental illness. In fact, the study has nothing to do with spanking at all. Canadian researchers asked 34,000 adults how often they had been pushed, grabbed, shoved, slapped, or hit by their parents or other adults when they were children. The authors explain that they were trying to assess the long-term effects of regular harsh physical punishment, which, they write, "some may consider more severe than 'customary' physical punishment (i.e., spanking)." Ultimately, the researchers reported that adults who have mental problems are more likely to say they were pushed, grabbed, shoved, slapped, or hit by their parents than healthy adults are.

Yes, OK. Abuse is bad. But now I wanted to know: What about spanking? According to a 2011 study, more than half of all American parents spank their toddlers; some studies have put the number closer to 60 percent. But the American Academy of Pediatrics opposes it, and the practice is illegal in 32 countries, including Spain, Israel, and all of Scandinavia. So what's the deal -- are slaps on the tush OK if your children deserve it, or will it screw them up for life?

After digging into the literature on the topic and talking to a handful of experts, my best shot at a conclusion is this: It depends. If you spank your kids frequently, harshly, or after you've lost your temper, then your kids may end up worse off because of it. If, on the other hand (no pun intended), you rely primarily on nonphysical disciplinary tools like time-outs, but you (lightly) spank your kids with the palm of your hand several times when they don't comply with these tactics, reasoning calmly but firmly with them as you do -- then spanking might make your children better behaved, and it probably won't do them any harm.

I wrote my best shot at a conclusion because, despite a recent Huffington Post piece calling the "debate" settled, the research on spanking is messy and controversial. The same study might be cited by different experts as evidence that spanking is safe and that it's dangerous. That's in part because the effects of spanking are hard to evaluate: You can't assess its effects like you would a drug, in a randomized controlled clinical trial, because it's unethical for researchers to instruct a random group of parents to spank their kids and a second group not to. Yet in the name of science, that type of study would be ideal, because it would allow scientists to compare what happens to both groups of kids -- and conclude, with some certainty, that any differences arising between them were due to the spanking.

But for physical punishment, the best researchers can do is compare what happens to kids who are punished with those who are not. And the thing is, kids who incite spankings may be more difficult or delinquent to begin with than kids who don't. Ideally, studies should control for a child's behavior prior to spanking, to ensure that findings only reveal how many more problems kids develop after they are spanked -- but researchers don't always do this, in part because they often collect information about people at a single point in time. Researchers might ask adults if they were hit as kids and, at the same time, assess these adults for mental illnesses, looking for associations between the two -- such as in this 1999 study. (The recent study on harsh punishment was conducted in this manner, too.) But when these types of studies report that kids who were spanked have more behavioral problems later in life than kids who were not, it's impossible to pin the cause on spanking.

To prove this point, in 2010, researchers at Oklahoma State University investigated whether nonphysical punishments are also associated with delinquent behavior later in life. They found that psychotherapy, grounding, and sending children to their rooms all make kids more antisocial. The researchers don't actually believe that psychotherapy causes behavioral problems -- they just wanted to show that these kinds of studies identify associations that aren't necessarily causally linked the way you would expect. Discipline is associated with behavioral problems in part because discipline is caused by behavioral problems.

There's also the possibility that parents who spank are different in important ways from parents who don't. A 2010 study published in Pediatrics found that spankers are more likely than non-spankers to be young, depressed, under a lot of stress, and substance abusers -- so it's hard to know whether some of the purported effects of spanking might actually be caused by poor parenting skills in general. And what about genetics? Substance-abusing or depressed parents who spank may have given their kids genetic predispositions to mental health problems, so, again, it's not necessarily the spanking that's the cause of future problems. Some studies try to control for these types of confounding variables, but not all.

Then there are what some researchers call "lumping problems" -- when studies are reported as being about spanking but aren't really about spanking. A 2002 meta-analysis scrutinized data from 88 studies and reported that although physical punishment makes kids more immediately compliant, it also makes them more aggressive in the short and long term and more likely to engage in criminal behavior as adults. This analysis has been cited as some of the strongest, most comprehensive evidence that spanking is dangerous, but it actually looked at far more than just spanking: In 16 percent of the studies, parents were hitting their kids with objects, not with the palms of their hands. It also included punishments such as pinching and slapping. It doesn't surprise me that kids who are hit with objects or pinched or slapped end up more screwed up than kids who aren't. But it's misleading to conclude that these kids were worse off because they were spanked.

Finally, there's the issue of frequency, severity, and timing. Some research, such as a 2007 study reporting that spanking induces behavioral problems, lumps kids who were ever spanked into a single category and compares them with kids who were never spanked. But these subjects may have been spanked once in their lives (like me) or once a day for 10 years straight. Some may have been hit hard on bare buttocks, while others were spanked lightly on top of clothes. And all of this matters in terms of the impact of the spankings.

The 2010 study in Pediatrics, for instance, reported that 3-year-olds who were spanked were more likely to exhibit aggression when they were 5, even controlling for the kids' prior behaviors and the mothers' other risk factors (like depression). But the study only found these ill effects if the kids were spanked more than twice a month. Another study, published in April 2012, found that kids who were spanked once a month or less with hands (no objects) were no more likely to subsequently develop delinquent behavior than were kids who were never spanked at all. Kids spanked more frequently or with objects, however, did develop problems. Finally, research suggests that preschool kids are the least likely to get scarred from spanking but that striking infants or older kids could cause problems.

So spanking, it seems, can be safe. But is it effective? In a 2005 meta-analysis, researchers analyzed 26 studies that compared the effects of spanking with those of other disciplinary tactics. They found that when spanking was used as a backup, only on 2-to-6-year-olds, and wasn't done in anger, it helped to increase kids' compliance in both the short and long term compared with 10 other disciplinary tactics, including scolding, reasoning, and removing privileges. There are also four studies (published in 1981, 1983, 1988, and 1990) that did use randomized, controlled clinical trial designs to assess the positive effects of spanking in preschool kids with conduct disorders, reporting in all four cases that spanking when kids did not adhere to time-outs improved their subsequent behavior.

With those results, why wouldn't we all start occasionally spanking our kids? Well, there's evidence to suggest that spanking begets spanking, or worse. The same 2012 study that found that light spanking was OK also reported that parents who lightly and infrequently spanked were 50 percent more likely to spank more harshly and frequently the following year. And a study published in 2008 reported that when moms began spanking more often, they were also more likely to become physically abusive -- suggesting that spanking might be a kind of gateway drug to more horrible things. So, yes, spanking may well be safe. Until it's not.

- - -

Melinda Wenner Moyer is a science writer living in Brooklyn, N.Y.