Honestly it depends on what the goals of the test are.
Ask 10,000 people in <x position> the same generic question. So ask 10,000 Sales Managers if they like dressing up for work , look for a correlation, compare the correlation to performance, get your answer.
Out of 10,000 sales managers, 3,000 said they enjoy dressing up for work. Of those three thousand managers 2,600 are top performers and get high marks from their coworkers. From this you can make a very slight up-tick towards someone answering true to them having the possibility of being a top performer. This would just be one question over hundreds of similar questions so it isn't going to be a deal-breaker.
Of course, ask 10,000 programmers and "false" might be the correct answer. That's what makes these questions powerful. There is no correct answer. There's only a relationship.
Ask 10,000 people in <x position> the same generic question. So ask 10,000 Sales Managers if they like dressing up for work , look for a correlation, compare the correlation to performance, get your answer.
Out of 10,000 sales managers, 3,000 said they enjoy dressing up for work. Of those three thousand managers 2,600 are top performers and get high marks from their coworkers. From this you can make a very slight up-tick towards someone answering true to them having the possibility of being a top performer. This would just be one question over hundreds of similar questions so it isn't going to be a deal-breaker.
Of course, ask 10,000 programmers and "false" might be the correct answer. That's what makes these questions powerful. There is no correct answer. There's only a relationship.