Slashdot Mirror
Bilingualism Delays Onset of Dementia
Dee writes with word of a Canadian study indicating that lifelong bilingualism delays the onset of dementia by 4 years. The scientists were reportedly "dazzled" by the results. From the article: "The researchers determined that the mean age of onset of dementia symptoms in the monolingual group was 71.4 years, while the bilingual group was 75.5 years. This difference remained even after considering the possible effect of cultural differences, immigration, formal education, employment and even gender as influencers in the results. "
Normally, we would test a difference in means between two populations by a t-test. If the sample size is large enough, then even a difference that is only a fraction of a standard deviation can be statistically significant. F-tests are used in ANOVA tables, and yes, they do assume normal distributions, as well as homoskedasticity (same variance). Assuming they performed a linear regression, then one can perform a Type I F-test (added-in-order test) or Type II F-test (added-last test). One can also talk about an overall F-test, testing whether any of the effects in the model are nonzero. However, as I indicated in another post, the study only had 184 patients from a single treatment center. There is selection bias, since the study only sampled patients who were already suffering from memory loss. How many other bilingual immigrants with memory loss are lurking in the general population, who aren't going to memory loss clinics due to lack of knowledge? Also, what method did they use to adjust for "cultural differences, immigration, formal education, employment and even gender" with only 184 patients? The study only proves that bilingual patients who arrived at a particular treatment center were, on average, 4 years older than monolingual patients. It does NOT provide a causal link between bilingualism and cognitive reserve.