Tuesday, August 21, 2007

Does Child Obesity Cause School Absences? Correlation versus Causation?

This paper cited below presents an interesting fact based on some 5th grade kids. The heavy ones miss more school. So, to build up human capital do we ban fries and oreos? Or do we search for a 3rd factor (such as parental and child patience and self control) that determines both a child's weight and devotion to school?

The good news here is that statistical analysis offers an easy test of this claim. The authors could have surveyed 1,000 kids by following Victor Fuch's strategy in an old NBER paper. He played games with individuals to learn about whether they prefer $100 now or $300 2 years from now. While I"m simplifying, he used this game to identify impatient people (those who took the $100) and patient people (those who waited for the payoff). His simple game took an "unobservable" (self control) and generated an observable measure of it. He then documented that controlling for age, education and race, more impatient people were more likely to smoke.

I tell you this tale because these fat kid authors should have done the same thing. Suppose they had followed Fuchs and estimated this "patience" variable. They then could have augmented their basic statistical model and estimated:

absences from school = controls + b1*weight + b2*patience + U

and then test whether b1>0, without controlling for patience and self control
their estimates are biased toward finding that b1>0.

the key counter-factual is "if a kid randomly gained weight would he miss more school?" possible but i think this is classic selection issue over what we learn
about a kid's "unobservables" from observing his/her weight.

August 21, 2007
Vital Signs
Patterns: Weight May Influence School Attendance

The more overweight a child, the more likely he or she is to be absent from school, a new report suggests. Researchers studied 1,069 fourth- to sixth-grade students in nine schools in Philadelphia. They recorded height, weight, sex, race and days absent for each. The study appears in the August issue of Obesity.

The scientists classified each child in one of four weight categories by body mass index: underweight, normal, overweight and obese. On average, underweight children were absent 7.5 days, normal weight children 10.1 days, overweight children 10.9 days and the obese 12.2 days. Even after adjusting for race, ethnicity, age, sex and school attended, being overweight remained a significant predictor of absences.

Statistical analysis showed that weight, sex, age, school and race accounted for 11 percent of the variance in absences, meaning unknown factors are involved. The authors acknowledge that it is unclear whether the increased absences significantly affect overweight students’ performance.

Andrew B. Geier, the lead author and a doctoral candidate in psychology at the University of Pennsylvania, doubts that sickness among overweight children causes absences. “Even in fourth grade,” he said, “I believe that psychosocial factors, not physical ones, are keeping overweight kids from going to school.”