The question is whether this effect is stronger or weaker than the gains you get from profiling.
The middle solution: Anti-profiling. It's not about going after certain groups, it's about ignoring certain groups that are harmless.
We don't need to focus on: -70-year-old British grandmas -4-year-old children -Chinese women -etc.
Obviously this doesn't preclude someone slipping something into their bag without their knowledge, so bags would still have to be checked.
But at the end of the day, there's a lot of simple things you could do to make the whole process faster and easier for everyone, including 23-year-old Arab men with congenital overactive sweat glands.
That is still profiling.
Given four groups, A,B,C,D with a property of dangerous/not dangerous.
If I say only A,B and D are dangerous then I'm saying C is not dangerous.
If I say only C isn't dangerous then I'm saying A,B and D are dangerous.
Also, because terrorists are rare and are not exclusively members of any one profile, you are more likely to miss one if you shift focus onto one group in particular.
http://www.pnas.org/content/early/2009/02/02/0813202106.abst...