Most students can spot the comment “needs more analysis” on a marked essay. Far fewer can point to the exact sentence that earned it. That is the real difficulty: critical analysis is not a tone of voice or a matter of sounding clever. It is a set of moves you can learn, repeat and check. Below is a model paragraph written at the level expected of a second-year UK undergraduate, followed by commentary explaining what each sentence is doing and why it works.

What “Critical” Actually Asks You to Do

Description tells your reader what a source says. Evaluation tells them how much the source is worth, and what follows from that for your argument. In practice, evaluating evidence comes down to three questions you apply to every claim you use:

• What exactly is being claimed?
• What evidence is offered in support of it?
• How far does that evidence carry the claim, and where does it stop carrying it?

Answer the third question on the page rather than only in your head, and your paragraphs will read as analysis.

The Model Paragraph

The question being answered is: to what extent do recorded lectures improve student attainment? The source named here is invented so the example stays clean; swap in a real one from your reading list and the moves stay identical.

Recorded lectures are often presented as a straightforward equaliser, and Adeyemi’s (2021) survey of second-year students is regularly cited in support. The study is useful, but its design limits what it can show. Because participation was voluntary and the questionnaire was circulated by departmental email, the sample probably over-represents students who were already engaged; those who drifted away from live teaching are precisely the group whose absence from the data matters most. The measure of attainment — module marks — is also blunt, since it captures assessment performance rather than understanding, and a student who watched recordings at speed may have gained flexibility while losing depth. None of this makes the findings worthless. The consistency between participants’ stated reasons for watching recordings and the broader pattern across the module gives the argument some weight, and the open comments add texture that a purely numerical account would miss. Taken together, the evidence supports a modest claim: recordings help students who are already motivated to review material. It does not support the stronger claim that recordings improve attainment across a cohort. That distinction matters, because the first justifies investment in recording equipment, while the second would justify redesigning teaching around it.

Annotation: The Moves, One by One

It names the claim, not just the source

The opening does not say “Adeyemi wrote about recorded lectures.” It states the position being contested and attributes it. That gives the paragraph something to push against, which is the whole point.

It attacks the design, not the conclusion

The concession costs nothing and buys credibility. The critique then targets a specific mechanism — voluntary participation — and explains the consequence: the students missing from the data are the ones whose behaviour matters most. A flaw with no stated consequence is just a complaint.

It questions the measure

Module marks are a proxy for learning, not learning itself. Noticing what an indicator can and cannot capture is one of the fastest ways to lift a paragraph above summary.

It concedes something real

Evaluation that only subtracts reads as hostile, and markers notice. Here the paragraph grants the study weight for a stated reason, which makes the verdict look measured rather than predetermined.

It states the verdict and its scope

The paragraph ends with a claim the evidence can support, separates it from the stronger claim it cannot support, and says why the difference matters. That final step — the so-what — is what turns an evaluation into an argument.

For Comparison: The Descriptive Version

Adeyemi (2021) conducted a survey of second-year students. The survey was distributed by email and asked about their use of recorded lectures. The results showed that many students found recordings helpful. Adeyemi concludes that recordings are valuable for learning. This shows that recorded lectures are an important part of modern university teaching.

Nothing here is false. Nothing here is thinking, either. The reader learns what was done and what was concluded, but not whether to believe it. The last sentence is the giveaway: “This shows that” performs agreement rather than argument, and it quietly drops the sample, the method and the limits of the measure. Swap the topic and the paragraph would still work, which is exactly the problem.

A Sentence Toolkit You Can Borrow

• The study establishes X, but it cannot establish Y, because…
• This finding is strongest where… and weakest where…
• The evidence would be more persuasive if the author had…
• Read alongside other work on the same question, the claim looks narrower than it first appears.
• The author moves from [evidence] to [conclusion] without explaining…
• Even if we accept the study’s assumption that…, the conclusion only follows when…
• The evidence justifies a limited version of the claim, namely…

Five Traps That Flatten Analysis

1. The “and then” essay. One source after another, linked by “Similarly” and “Furthermore”, with no judgement between them.
2. Praise without criteria. “This is a very strong article” means nothing unless you say what makes it strong.
3. Confusing the claim with the evidence. Authors assert more than their data supports; your job is to notice the gap.
4. Ignoring who, how many and when. Sample, method and date are not background detail — they are the terms on which you decide how much weight to give a finding.
5. Evaluating so hard you forget the question. Every judgement should serve the essay’s argument, not your enjoyment of the source.

Check Your Own Draft in Ten Minutes

Highlight every sentence in one body paragraph. Colour the ones that report what a source says, and the ones that judge it. If the second colour is missing, you know what to fix.

Then take your weakest paragraph and rewrite two sentences using the toolkit above, remembering to include the reason your judgement holds. Reasons are what separate analysis from opinion.

Finally, read the paragraph aloud and ask: could this be written about any source on this topic? If the answer is yes, you have written a summary. Add the specific detail — the mechanism, the missing group, the limit of the measure — that makes the paragraph yours.

Photo: Pexels / Pixabay