What is differential item functioning (DIF) and why is it important in test fairness?

What is differential item functioning (DIF) and why is it important in test fairness?

• A. DIF means items are culturally neutral.
• B. DIF is when all items are equally difficult for everyone.
• C. DIF occurs when items favor one group over another after controlling ability; indicates potential bias; requires examination.
• D. DIF is a measure of internal consistency.

Answer: (C)

DIF shows up when an item behaves differently for different groups even after you account for the test taker’s overall ability on the trait being measured. In practical terms, people with the same underlying level of the construct (like ability in a domain) do not have the same probability of answering the item correctly because of their group membership, indicating potential bias in that item. This matters for test fairness because items that favor one group can distort scores and misrepresent true ability, so such items are flagged for closer review and possible revision or removal to preserve validity. The other options miss the point: an item being culturally neutral would not exhibit systematic bias across groups; all items being equally difficult for everyone would imply no differential functioning; and DIF is not about internal consistency, which is a reliability property.

DIF shows up when an item behaves differently for different groups even after you account for the test taker’s overall ability on the trait being measured. In practical terms, people with the same underlying level of the construct (like ability in a domain) do not have the same probability of answering the item correctly because of their group membership, indicating potential bias in that item. This matters for test fairness because items that favor one group can distort scores and misrepresent true ability, so such items are flagged for closer review and possible revision or removal to preserve validity. The other options miss the point: an item being culturally neutral would not exhibit systematic bias across groups; all items being equally difficult for everyone would imply no differential functioning; and DIF is not about internal consistency, which is a reliability property.