That is what they're designed to do, based on studies that have been done on how to tell which questions are the most or the least "g-loaded", which is an expression for how dependent a question is on mental ability rather than prior knowledge and such (based on the use of g to mean "general intelligence"). Then you can spot what the questions with the highest g-loads have in common and which questions or types of questions have the lowest g-loads, eliminate the latter, and make a test consisting of the former, resulting in a highly g-loaded test. The result: tests with the highest g-loading, rather than the ones with questions that would supposedly favor the more successful people, are the most predictive of other successful behaviors in life. And the differences between test scores that some people say are due to bias (low g-load) not only don't go away on highly g-loaded tests, but even are sometimes larger than on the less g-loaded ones. These observations indicate that g itself, rather than other supposed advantages aside from g, is the primary determiner.