Statistics: Problems with observational data

From MathWiki

Problems with observational data

Thanks to Dr. M. Konstantareas (Guelph University) for the following:

An example of a change in interpretation in brain morphology of autistic children.

Hashimoto T, Tayama M, Murakawa K, Yoshimoto T, Miyazaki M, Harada M, Kuroda Y.(1995). Development of the brainstem and cerebellum in

autistic patients. Journal of Autism and Developmental Disorders, 25, 1-18.


Studies of magnetic resonance images have revealed morphological disorders of the brainstem and cerebellum in autistic children and adults. When we studied development of the brainstem and cerebellum in autistic patients, we found that although the brainstem and cerebellum significantly increased in size with age in both autistic patients and controls, these structures were significantly smaller in autistic patients than in controls. The speed of development of the pons, the cerebellar vermis I-V and the cerebellar vermis VI-VII was significantly more rapid in autistic patients than in the controls. However, the speed of development of the other brain structures in the posterior fossa did not differ between autistic patients and controls. The regression intercepts of the brainstem and cerebellum as well as those of their components were significantly smaller in autistic patients than in controls. Results suggest that brainstem and vermian abnormalities in autism were due to an early insult and hypoplasia rather than to a progressive degenerative process.


Hardan, A.Y., Minshew, N.J., Harenski, K, & Keshavan, M.S. (2001). Posterior

fossa magnetic resonance imaging in Autism. Journal of the American Academy of Child and Adolescent Psychiatry, 40, 666-672.

Quoted from conclusion:

In the present study, there were no differences in the volume of the brainstem or in the midsagittal areas of the midbrain, pons, and medulla oblongata. This finding is consistent with those of several previous studies (Garber and Ritvo, 1992; Hsu et al., 1991; Piven et al. 1992) but contrasts with others that have reported abnormalities in these structures (Gaffney et al., 1988; Hashimoto et al., 1995). The entire brainstem and the pons were found to be smaller in a study of 13 high-functioning autistic subjects when compared with 35 medical controls (Gaffney et al., 1988). In a larger study, Hashimoto et al. (1995) studied the midbrain, pons, and the medulla as a function of age in 102 predominantly retarded subjects with autism and found that these structures were smaller at all ages compared with those of controls. When the sample included only non-mentally retarded children with autism, differences were found in the midbrain and medulla oblongata but not in the pons (Hashimoto et al., 1993). However, th e major limitations of the Hashimoto et al. (1993, 1995) studies are the inclusion of children with autism who had other causes of brain damage and the comparison with typically developing children. Furthermore, the size of the whole brain was not reported, which makes it difficult to determine whether the findings truly reflect posterior structures that are small relative to the rest of the brain. Therefore, the observed abnormalities in brainstem structures may be a function of the choice of control groups, the inclusion of nonhealthy participants and failure to adjust for potential confounds such as IQ and brain size.