A family quiz is not fair simply because everyone receives the same questions. If every question depends on one decade or hobby, some people have little chance to contribute. A mixture of recognition, reasoning and knowledge gives different people a way into the game.
Use short rounds with different entry points
For example, plan one round using pictures from the shared household, one with general knowledge and one with simple estimates. Familiar objects photographed unusually close can supply the picture round. Knowledge questions should not exclusively reflect the question writer’s interests.
Estimation questions need a checkable answer, such as the number of pieces you have counted in a prepared box. Avoid vague wording that invites arguments about counting. Write down beforehand what counts as one piece.
Check the question before its answer
Read each question aloud and ask whether more than one answer would be reasonable. “Which animal is fastest?” is ambiguous without a context. A fair question states the necessary qualification or is replaced with something clearer.
Keep answers and any references separate from the question sheets. Decide beforehand which spelling variants count and whether partial answers receive points. For a relaxed evening, a transparent scoring system is better than a complicated points table.
Try two or three questions on someone who will not be playing. If they interpret the wording differently from your intention, revise it rather than blaming the group later.
Make teams collaborative
Mix experience and interests where participants agree. Allow brief discussion and collect written answers so that the quickest voice does not always win. One person can read and another write, with roles changing between rounds.
If a question proves unclear during play, make one calm decision that applies equally to everyone, such as removing it from scoring. The quiz is meant to create shared entertainment. Handling a mistake transparently contributes more than insisting on a poorly worded answer.
Make one question checkable on the spot
One example of an original estimation task is the number of identical paper cards in a closed stack. Count them twice beforehand and record the answer. After submissions, the group can count together. Unlike an unclear internet statistic, the task has a visible basis. If teams are equally close, use the tie rule agreed beforehand rather than inventing it after seeing the results. The counting can itself become a brief shared reveal.