Tuesday, May 29, 2007

Day Care: Digging through the Data

Fortunately a recent day care study got lots of play in the media.

“Day-care Kids Have Problems Later in Life” (NBC)
“Study Finds Rise in Behavior After Significant time in Day Care” (New York Times)
“Poor Behavior is Linked to Time in Day Care.” (New York Times)
“A Downside to Day Care?” (Newsweek/MSNBC.com)

Unfortunately the headlines tended to be, well, at worst, alarmist, at best, misleading. The articles themselves, skewed to half-baked. So let’s take a look at what the study really found…

The $200 million project by the National Institute for Child and Human Development tracked 1, 364 children from birth and examined a long list of factors to see if a link existed between child care and children’s academic and behavioral outcomes. The research team led by Jay Belsky checked to see if quality of care, amount of time in care, and type of care predicted kid’s reading, math, vocabulary, and classroom behavior from pre-kindergarten through sixth grade. And because we know parents exert some influence on their progeny, they also looked to see if parenting quality predicted the above outcomes.

Based on the headlines what do you happened? Time in day care predicted poor results across all measures? Problems at school grew over time? No hope for day care kids? Time in Day Care More Powerful than Parents?

Quality of child care. We have a positive finding. At least for parents who have their kids in good child care arrangements. The better the child care quality, the larger the kid’s vocabulary. Must be all that social interaction. So this finding is persuasive, it holds up over time, unlike our next measure.

Amount of time spent in child care. The more time, the more likely teachers reported behavioral problems at age four and a half. But, and here’s an important qualification…BUT NOT at 6th grade. Now you might recall hearing about those unruly kindergartens a few years ago. Back in 2001 another set of researchers looked at the results for these same kids, then kindergarteners, and announced a link between day care and aggressive behavior. You can imagine if not recall the headlines, no? Yet now, six years later, any differences between the day care kids and The Others have vanished (I miss Lost already one week into summer hiatus). Thus, amount of time in day care does NOT have any long-term correlation with classroom conduct. Or, and this is also striking, with any of the six other academic or behavioral outcomes included in the study (e.g., reading, math, vocabulary, social skills, work habits, social-emotional development.)

You’ll note, of course, that I’ve used the words “correlation” and “relationship”. This study can only attest to a correlational relationship and not a causal one. Despite your socio-political opinions or dare I say suspicions, there was no Orwellian experiment that randomly assigned kids to either go to day care or stay home with mommy.

Now the most controversial finding. The headline-grabber. Kids who spent time in day care centers were more likely to engage in aggressive and otherwise unruly classroom conduct than those in other child care arrangements, including staying home with parent. Sounds bad for those day care kids. The stats in the journal article? They don’t really give us a good idea of just how bad it is for those kids, or more likely, their teachers. But look at the end of that lengthy New York Times article. Those parents who persisted would have come across this nugget of reason. Dr. Margaret Burchinal, co-author of the study, put this apparently day care kid bad behavior into perspective:

“Every year spent in such centers for at least 10 hours per week was associated with a 1 percent higher score on a standardized assessment of problem behaviors completed by teachers."


All this fuss for a 1% higher score? But what does this 1% entail – 5 extra principal office visits, 10 extra cuts in line, 20 extra comments during silent reading time? It seems quite meager given as a percentage. Still, 4 years in day care translates into 4% more problems behaviors. Might not seem like much for an individual child or parent, but for the school system, the country as a whole, maybe a problem. Before we get nervous, let’s consider Dr. Burchinal’s almost apologetic clarification in the article The Kids Are Alright by Emily Bazelton in Slate magazine:

"I'm not sure we communicated this, but the kids who had one to two years of daycare by age 4½—which was typical for our sample—had exactly the level of problem behavior you'd expect for kids of their age. Most people use center care for one or two years, and for those kids we're not seeing anything problematic."

Further demystifying the problem behavior, or really, typical behavior, Burchinal also revealed the average behavior scores in relation to time in day care.

No time in day care: 49.6
One or two years: 50.0
Three years: 51.4
Four years: 52%

Other revelations by the co-author? Only 5% of the children spent 4 years in day care. And ten percent spent 3 years. That’s 15% of the sample exhibiting misconduct to a modest degree. Or really I should say, showing less proper conduct as their level was within normal limits according to Dr. Burchinal.

So now you’re probably tiring of all the stats and such. I’ve not even mentioned the other significant set of findings. Parenting, that is, quality of parenting, significantly predicted ALL the outcomes variables from reading to work habits to classroom conduct. The higher the quality of parenting, the better the child fared both academically and behaviorally.

So what? Is day care dangerous? Hardly looks it. This was a very thorough endeavor. The researchers included many variables thought related to child care and the outcomes. It was a significant research project and one specifically designed and carried out to address child care. Not some data jettisoned from another large national survey examining any number of other issues. Solid research. Solid findings.

A note about small “effect” sizes. I use the parentheses because the study cannot show causation, thus, effects, but we do use the term “effect size” in speaking about statistical results. Anyhow, it’s useful to point out that even a significant effect may not be all that meaningful, at least for individuals. Does that 1% difference in conduct scores reflect a meaningful difference in each child’s life? Collectively, across the school, perhaps, many children with slightly more behavioral problems might make a difference. Yet for individual children it might not matter much or to their parents. I like the example my friend Stephanie uses in her stats classes. Let’s say a new shampoo makes a significant difference in curing baldness. People who use it grow 10 new hairs on their heads. We’ve used a very large sample and have found that those 10 hairs are significant, statistically speaking. Would you notice those 10 hairs on your head? Is that a meaningful difference? Not really.

What’s not in the journal article is perhaps more fascinating than the actual findings. The author/researchers omitted some pieces of information they later shared in several interviews. Namely, statistics that clarified the scope of the day care kid’s conduct issues, painting the supposed daycare effects as slight. In other words, they defended day care. So why leave these details out of the academic article? Hmmm? Journals overwhelmingly publish studies with significant findings. Statistically significant findings. Many a paper with no relationships, causal or correlational – are languishing in desk drawers, filing cabinets, and attics. It’s not in an author’s interest to downplay a significant finding – not if they want to get published. So after publication we learn more about this modest relationship.


Read an abstract of the article in Child Development. By the way, Emily Bazelon's article in Slate The Kids Are Alright



Belsky, Burchinal, McCartney, Lowe Vandell, Clarke-Stewart, and Tresch Owen. (2007). Are There Long-Term Effects of Early Child Care? Child Development 78 (2), p. 681–701.




No comments: