The Daubert standard is used in federal courts and the courts of many states to decide whether evidence based on a scientific theory may be admitted at a trial. (Courts in other states retain the older Frye standard for making that decision.)

Shaken Baby Syndrome, in what the author calls "its most extreme form," asserts that an infant found to have a subdural hematoma (bleeding on the brain) and retinal hemorrhages (spots of bleeding in the retina) must have been violently shaken. This shaking, according to the theory behind SBS, caused the baby's head to rotate in a whiplash fashion, and that caused the bleeding. Many caretakers have been convicted of assaulting infants based on little or no evidence beyond these two medical findings-- and a doctor's opinion that they prove violent shaking.

The article reviews scientific literature on SBS and then applies the Daubert criteria to the syndrome.

The first Daubert criterion is whether the theory "can be (and has been) tested." Whether it can be tested means whether, if it were false, testing could show that. The article concludes SBS cannot be tested. According to SBS, violent shaking has occurred whenever an infant has a subdural hematoma and retinal hemorrhages, but no sign of any trauma from anything other than violent shaking. This, says the article, is "a logical circularity--" the cause and the effect are the same. This makes it impossible to prove SBS theory false.

The second Daubert criterion is whether the theory has been published in peer-reviewed journals. The article acknowledges that this has happened but contends that that process is detecting flaws in SBS theory, especially in recent years.

The third Daubert criterion is the error rate. Because cases having the two key medical findings are assumed to be examples of SBS, despite denials by caretakers, there is no error rate. The article quotes a 2001 letter to a medical journal by Brian J. Clark:

In this field [SBS research] more than others we are more readily prone to alter the facts to fit the hypothesis rather than alter the hypothesis to fit the facts. For example, if the accused admits to severely harming the child, we tend to believe because this fits our hypothesis, but if the accused offers an alternative apparently innocuous explanation, we discount the explanation because it doesn't fit the hypothesis. Published cases purporting to demonstrate less traumatic causes are attacked for their anecdotal data and for overlooking the real, more sinister explanation. This "illogical inconsistency" overlooks scientific process and forgets that the requirement for severe shaking forces is no more than a favored but unproven hypothesis.