I have research data using the Boston Diagnostic Aphasia Examination and Boston Naming Test. I would like to be able to use a small number of summary scores such as 'spoken expressive language' instead of the many subscale scores. The reason being that analyzing the many different subscales is too risky for false positives. Are there any previously used systems for producing a reasonably small (say 3-7) set of composite scores?
This is a tricky issue, because the BDAE was designed to assist in the differential diagnosis of language disorders rather than the global identification of cognitive deficit (like the original Halstead-Reitan). There is no good neuropsychological rationale, as far as I can tell, for combining (say) fluency with naming to confrontation, even though both are components of expressive language. I suppose you could just mash everything together, but would the result measure anything coherent?
Now, I am an edition out of date here, but I thought the BDAE did yield a set of about 8 subscores at the highest level (fluency, naming, repetition, auditory comprehension, reading comprehension, oral reading, writing, and spontaneous speech). Am I way off base here? Maybe the newest edition has dropped those scores, but it they existed before, then yes, there is a precedent for deriving just 8 scores. And of course, if your initial tests show significant deficits for one group or another on one of those 8 scales, then you can dig deeper into any smaller scales that contribute to them.
You do have a potential Type I error problem here. Maybe you could just set your alpha level to .01 rather than .05? The problem with this is that you could miss a whole bunch of small but real effects. The alternative would be to address the risk of false positives in your discussion.
- Badges
- Science topic
- Similar topics
- Psychology