Adding an extra test would still help. They can’t perfectly measure skill, and if men are better than women, then a given score is more likely to err low if it’s for a man then a women.
Also, directness means nothing. If you have a bad enough test, it’s entirely possible that checking only there gender will be more accurate then checking only the test. All that matters is the correlation between their skill and the result of the test.
Adding an extra test would still help. They can’t perfectly measure skill, and if men are better than women, then a given score is more likely to err low if it’s for a man then a women.
Also, directness means nothing. If you have a bad enough test, it’s entirely possible that checking only there gender will be more accurate then checking only the test. All that matters is the correlation between their skill and the result of the test.