Hedging: when to commit and when not to

It could be argued that the policy was ineffective.

Two very different things might be going on in that sentence. Either the evidence is genuinely mixed and you are being appropriately careful — or you have evidence it was ineffective and are declining to say so.

A marker cannot tell which from the sentence. That is the problem, and it is why hedging is the most double-edged habit in academic writing.

Why the good version exists

Hedging is not a stylistic tic; it does real work. Research findings are conditional, samples are limited, and claiming more than your evidence supports is a worse failure than claiming less.

Delete the hedge there and you have overclaimed. The caution is the scholarship.

Why the bad version is so common

Because it feels safe. An unhedged claim can be wrong; a hedged one cannot quite be. Students learn early that overclaiming gets corrected, and over-correct.

The second is shorter, more useful, and more honest about its own limits. The first says nothing and cannot be wrong, which is not the same as being right.

The test

One question, and it resolves nearly every case:

Can I say what the evidence actually supports?

If yes, say that instead. It could be argued that the policy was ineffective becomes the policy did not reduce rates in either cohort — which is a claim, is checkable, and is stronger.

If no — if the evidence genuinely does not settle it — then keep the hedge, and say why it is hedged. The best version of cautious writing names its own limit, and that is what separates it from vagueness:

That sentence hedges and informs. It could be argued that only hedges.

The opposite failure

Over-hedging is the common problem, but the correction has its own failure mode, and markers penalise it harder because it is a claim about the world rather than a weakness of style.

Overclaiming has three reliable tells:

Universals with no support. All, never, always, proves. In most fields nothing short of mathematics proves anything; evidence supports, indicates or is consistent with. A single counter-example destroys a universal claim, and a marker only needs one.

Correlation described as cause. Reading to children improves their vocabulary is a causal claim. If the study was observational, the honest sentence is children who are read to have larger vocabularies, and the gap between those two sentences is most of a research methods module.

Generalising past the sample. A study of 200 undergraduates at one university supports a claim about 200 undergraduates at one university. Extending it to students is a hedge you should not have removed.

The second is longer, and it is the one that survives a viva.

Where in the essay it belongs

Hedging is not distributed evenly through a good piece of writing, and knowing where it concentrates is a quick way to audit your own.

Two patterns are worth catching. A hedged results section usually means the writer is interpreting where they should be reporting. A hedged conclusion usually means they lost confidence in the argument they just spent 3,000 words making — and it is the last thing the marker reads.

The phrases worth looking at

Not banned — worth pausing on.

That last one is a genuinely useful test. Delete the hedge and read the sentence back. If the meaning is unchanged, the word was doing nothing and can go. If the sentence now claims more than you can support, the hedge was earning its place.

Double hedging

The clearest sign that hedging has stopped being deliberate is when it stacks.

It may possibly be the case that… · This might perhaps suggest… · There is some evidence which may indicate…

Each of these hedges twice, which does not make the claim twice as careful. It makes it unreadable, and it is a reliable tell that the writer is uncomfortable rather than being precise. One hedge per claim is the ceiling.

The first-person cousin

I think, I believe and in my opinion do the same job by a different route: they mark a claim as personal so it cannot be wrong.

The second is not more arrogant. It is a claim about the data rather than about you, which is what an essay is for. Whether the first person is acceptable at all varies by discipline — see our guide to what the formalizer looks for — but even where it is encouraged, I think in front of an evidenced claim weakens it.

Hedging in the abstract

A special case, and the one with the most disproportionate cost.

An abstract is read by people deciding whether to read the paper, and increasingly by search and recommendation systems deciding whether to surface it. A hedged abstract fails both. This study may suggest a possible relationship between X and Y gives a reader nothing to decide on, and it is often the only part of your work a busy marker reads carefully before forming an impression.

The convention is that the abstract reports what you found, and the discussion is where you qualify it. Save the caveats for where there is room to explain them properly. If a finding genuinely needs a caveat to be stated honestly, put the caveat in — but state the finding first and the caveat second, rather than burying the result inside three layers of maybe.

Finding them

Our formalizer flags every hedge in a passage with its position, so this becomes a list rather than a re-read. It deliberately does not remove them: whether each one earns its place is a judgement about your evidence, and no rule-based tool can make that call.

The readability checker is useful alongside it, because heavily hedged sentences are also long ones — the hedges are extra clauses, and they show up as sentences flagged for length as well.

If you are over a word limit, this is also the cheapest place to cut. Hedging is usually 2–4% of a draft, and removing the hedges that were not earning their place loses no argument at all.