I think in cases like that smile-test they would evaluate it case-by case, so not a total score, but per picture... And since not everybody is going to have that same thought at the same time, averages will cancel that effect out.
But on things like personalised tests, where in the end the result would tell you something detailed like IQ testing or whatever, this could probably be an issue...
Then again, they don't generally have questions that are true or false, so I dunno if the same problem will apply...
Gonna have to ask tonight, I know a few budding psychologists. Crazy people