Neuropsychology · Simplified
Which Measure Separates Them?
There is no such thing as an informative test. There are tests that are informative about a particular comparison.
The pairing
Two profiles, four measures
Pick two of the four fictional profiles. Before you see their scores, say which one of the four measures you think will do most to tell them apart.
Key terms
- Fluency
- How freely speech comes out: length of phrases, effort, rhythm.
- Comprehension
- Understanding what is said, tested without requiring speech.
- Repetition
- Saying back a word or phrase just heard.
- Naming
- Producing the word for a shown object.
Which measure will separate these two most?
Both profiles on all four measures, out of 100. The measure that separates them is the one with the largest gap between the two bars, not the one on which either of them scores lowest.
What this shows
Informative about what?
The same measure can be the most useful thing you could possibly administer for one comparison and completely uninformative for another.
Key idea: A test tells you something only when the two possibilities you are choosing between predict different scores on it. Repetition is worth almost nothing when both people you are considering repeat badly, and it is decisive when one of them repeats far worse than their fluency and comprehension would lead you to expect. That is why a battery is not a list of good tests but a set chosen for the comparisons you actually need to make, and why the lowest score on a profile is often not the interesting one. The interesting score is the one that is out of line with the others.
Four fictional profiles with four made-up numbers each. Real assessment is not four scores: measures correlate, effort and fatigue move all of them together, and a single number for something as broad as comprehension hides more than it shows. These profiles are also drawn much more sharply than real ones, and the classical categories they gesture at are contested: many patients do not fit any of them, and the labels describe patterns of performance rather than explanations of them. Nothing here is an assessment instrument or a description of a real person.
The longer version adds two further profiles, all fifteen pairings, and a running record of which measures earned their place. It is at Aphasia Profile Comparator in the main collection.