Strongest-argument questions feel subjective, which is exactly why they trip people up. They aren't. There's one rule that removes the guesswork, and once it's automatic these become some of the most reliable marks in Decision Making.
The format
You're given a yes/no question — usually a 'Should we do X to fix Y?' — and four arguments, some arguing yes and some no. Your job is to pick the single strongest one. Importantly, a strong argument is not the one you personally agree with; it's the one that reasons best, whichever side it takes.
Split the stem into the solution (X) and the problem (Y). The winning argument explains why X does, or doesn't, actually affect Y. Touch only one half and you're out.
The foolproof rule: address both the problem AND the solution
A strong argument must connect the proposed solution to the actual problem, and explain how or why it does. Break the stem into two halves — the solution (X) and the reason or problem (Y) — then ask of each option: does it explain why X actually does, or doesn't, affect Y? An argument that only mentions one half, or wanders off to a side issue, is weak no matter how confident it sounds.
Relevance comes first, everything else is a tie-breaker
Apply the rule ruthlessly: any option that fails to link solution and problem is out immediately, even if it's detailed, factual and well written. Only when two or more options are genuinely relevant do you reach for tie-breakers — specificity over vagueness, evidence over bare opinion, and whether the option offers a concrete mechanism rather than a slogan.
- Off-topic — true, but answers a different question than the one asked.
- Unsupported assertion — states an opinion with no reasoning behind it.
- Restates the stem — 'yes, because more people could do X' with no effect explained.
- Overreach — sweeping 'always' or 'no one ever' claims that don't hold up.
The most dangerous option is the detailed, confident one that is simply irrelevant. Polish is not relevance. If a well-written argument never connects the solution to the problem, it loses to a plainer one that does — eliminate it without hesitation.
'Should under-fives be banned from phone apps because they can become addicted?' 'No, children enjoy the bright colours' ignores addiction — out. 'No, some apps are educational' raises a separate benefit — out. A 'Yes' linking heavy early screen use to a concrete harm connects the ban to the stated problem, so it's the strongest.
'Should vaccination be compulsory to reduce outbreaks?' 'No, vaccines can have side effects' is a vague, unlinked worry — out. The strongest 'No' gives a mechanism: outbreaks can still occur at high coverage because no vaccine is fully effective, so the policy may not deliver. Both halves, with a reason — that's the whole test.
Read the stem before you read a single option. If you've already named X and Y in your head, the weak arguments almost flag themselves on the first pass — and you avoid being talked round by a persuasive but irrelevant one.
Make the rule subconscious
Run the same checklist on every question until it's reflexive: name the solution, name the problem, eliminate anything that fails to link them, then tie-break on quality. Drill it and these questions stop feeling like opinion and start feeling like the near-guaranteed marks they should be.