Last year, researchers at the University of Rochester Medical Center tested the claim that putting a child on the autism spectrum on a gluten- and casein-free diet would improve symptoms. They came up empty-handed.

The scientists enrolled 22 two- to five-year-old children with an autism spectrum diagnosis and had them undergo 18 weeks of an intervention that included gluten-free, casein-free (GFCF) diets and participation in an intensive behavioral therapy. After the intervention, the researchers gave the children gluten, casein, or placebo snacks to see how they would respond. Parents, teachers, and research assistants (who didn’t know which element had been added back into the child’s diet) did not note any difference in the children’s behavior.

This isn’t the first time research has failed to prove that diets improve the lives or health of children with autism. In fact, earlier in 2010, the journal Pediatrics published a report by a panel of 27 experts stating that there was no scientific weight to the diets-help-autism claim.

But parents tell a different story. In overwhelming numbers, families say that eliminating certain elements from their autistic child’s food intake has drastically reduced his or her symptoms.

In fact, a recent study found that 27 percent of children with autism adhere to a special diet (one expert from the Pediatrics study said that the majority of families at least try a diet). The most common approach is to stop feeding a child with ASD any foods containing gluten, a protein found in wheat and other grains, and casein, a protein in milk. The idea is that children with autism may have a sensitivity to these proteins, leading to an inflammatory response from the immune system. Tummy troubles and other gastrointestinal symptoms (common in children with autism) make for discomfort and irritability, and a kid who struggles with words might tantrum or withdraw as a result.

Vaccines Don’t Cause Autism: So why do so many parents think they do?

Autism and ADHD: Is there a genetic overlap?

Top 30 Facebook Pages for Autism: Most supportive, most informative and more

The other line of thinking is that children with autism don’t have the proper enzymes to break down gluten and casein’s large peptides (components of proteins) or have an extra-permeable gastrointestinal tract (a “leaky gut”) that allows inappropriate amounts of the protein parts to enter the bloodstream. These elements could interfere with brain chemistry and cause an upswing in behavioral problems and other symptoms of autism.

Parents swear that re-vamping their children’s diets works. On the official blog of Autism Speaks, for example, parents repeatedly say that after implementing the GFCF diet, their children are more present, more focused, and less moody. Many parents note an improvement in sleep and a host of physical symptoms like fevers and irregular bowel movements. And it doesn’t end with gluten and casein – some also eliminate sugar, high fructose corn syrup, dairy, and food additives.

So why the disconnect between science and parents’ stories?

One possible reason is that there have been so few large randomized studies (such as those in which kids are put on strict diets and tracked by an unbiased observer for an extended time). Many of the studies so far have small sample sizes – last year’s GFCF study, for example, enrolled under 25 children, and it excluded those with gastrointestinal issues (a subset of kids that might benefit most from the diet).

Another possibility is that changing an autistic child’s diet does have a positive effect, but it actually helps all kids have more energy and be more focused, not just kids with autism.

My guess, though, is that science doesn’t match parent reports in part because autism is not one disorder. Hundreds of genes and an unknown list of environmental influences wrap together to produce the outward symptoms of what we know as autism – and in different combinations for different children. So far we don’t understand how to divide these children up in a meaningful way, we only know how to group them together and call it a “spectrum.” In reality, autism is probably multiple disorders, each with its own profile of genes, behavioral struggles, and physical symptoms.

So it’s probably true that diet does make a difference for a certain kids (not in “curing” autism, but in improving life), but when you embark on a broad study of autism, the nuances of the individual kids disappear.

At the end of the day, it is important for science to study the effects of gluten and casein on behavior; if we have evidence of it benefitting at least a subset of children, doctors can make better medical recommendations. In the meantime, many parents seem to take the view that what matters is what works for their child. If they observe a gluten-free, casein- free diet helping, that’s all the evidence they need.