Evaluating sources

The command word in the Section B evaluation question is evaluate (or sometimes assess / how convincing/effective). Cambridge defines it precisely: make a supported judgement about the quality, effectiveness or validity of the text, argument or evidence — weighing strengths and weaknesses. It is the only Section B task that asks you to step back from the text and deliver a verdict.

This matters because three weaker responses look superficially similar to evaluation but score far less:

• Summary retells what the writer says. ("The writer argues that screen time harms children.") That is content, not judgement.
• Description lists features. ("The writer uses statistics and a rhetorical question.") That is identification, not judgement.
• Analysis explains how a feature works. ("The rhetorical question draws the reader in.") That is closer, but still not a verdict on whether the argument is convincing.
• Evaluation judges effectiveness, with reasons. ("The statistic strengthens the claim because it is precise and sourced, but the single anecdote that follows is too narrow to prove a general trend — so the argument is only partly convincing.")

The test for whether a sentence is evaluation: does it contain a judgement word (convincing, effective, weak, unsupported, persuasive, reliable, exaggerated) attached to a reason? If not, you are describing.

Start with the argument — the chain of reasoning the writer builds. Ask three questions of it.

1. Is each claim SUPPORTED, or merely ASSERTED? An assertion is a claim stated as fact with no backing. Compare:

• Assertion (weak): "Everyone knows that social media is destroying childhood." (No proof — it leans on 'everyone knows', which proves nothing.)
• Supported (strong): "A 2023 study of 2,000 teenagers found that those using social media for over five hours a day reported sharply higher anxiety." (A specific, sourced claim.)

The strongest single evaluative move is to point out where a confident-sounding claim has no evidence underneath it.

2. Is the argument BALANCED, or one-sided? A writer who acknowledges and answers the opposing view ("Defenders of social media point to its educational uses, but...") is more convincing than one who ignores it. A purely one-sided text may feel forceful but is easy to distrust, because the reader senses the other side has been hidden.

3. Does the conclusion FOLLOW? Watch for overreach — a small piece of evidence used to justify a huge claim ("My cousin failed his exams after getting a phone, which proves phones ruin education"). One case cannot prove a general rule; the conclusion is wider than the evidence allows.

A balanced evaluation will usually find that the argument is stronger in some places than others — confident where it has support, weaker where it asserts or overreaches. Saying exactly where it is strong and where it is weak is what earns Band 3.

Once you have judged the reasoning, judge the evidence the writer offers for it. Rank it by reliability.

Strong evidence tends to be:

• Specific and sourced — "official figures show road deaths fell 18% after the speed limit was cut" is checkable and precise.
• Representative — drawn from a large or typical sample, so it supports a general claim.
• Relevant — it actually bears on the point being made.

Weak evidence tends to be:

• Anecdotal — a single vivid story ("I know a man who..."). It is memorable and emotionally powerful, but one case cannot prove a general trend. Anecdote is the most common weak evidence in opinion writing, precisely because it feels convincing.
• Selective (cherry-picked) — only the figures that suit the writer's case, ignoring those that do not.
• Vague — "studies show", "experts agree", "a huge number of people" — unnamed and uncheckable, so it carries little weight.

The subtle evaluative skill is spotting when strong-looking evidence does weak work. A real statistic ("90% of users said the app was easy to use") may still be useless if the sample is tiny, self-selected or unrepresentative. Equally, an anecdote is not worthless — it can legitimately illustrate a point already proved by data; it just cannot prove the point on its own.

Opinion and argumentative texts are designed to persuade — so part of evaluation is judging how they persuade and whether the methods are fair.

Emotive language — words chosen for their emotional charge ("a generation of children abandoned to their screens"). It is powerful, and a strong evaluative point is that it engages the reader; but an equally strong point is that it can substitute feeling for proof. The verdict depends on whether the emotion sits on top of real evidence or replaces it.

Generalisation — a sweeping claim about a whole group ("teenagers today have no attention span"). It is rhetorically tidy but logically weak, because it ignores exceptions and is rarely supported.

Bias — a one-sided slant that presents only the case the writer favours, often by loading the language ("greedy corporations" vs "job-creating businesses" for the same firms). Bias is not automatically fatal — all argument takes a side — but undisclosed, unbalanced bias makes a text less trustworthy.

Loaded rhetorical questions ("How much longer will we allow this?") and false either/or choices ("Either we ban phones or we lose a generation") are persuasive devices that can manipulate by closing down alternatives.

The evaluative skill is to give a two-sided verdict on technique: the device makes the writing more persuasive (a strength of effect) but may make it less reasoned (a weakness of logic). Naming both is exactly the balance Band 3 rewards.

Q8 is an extended answer of up to 120 words, so structure is everything. The reliable Band 3 shape is strength -> weakness -> judgement.

1. State a genuine STRENGTH, with a quotation and its effect. e.g. "The argument is strengthened by the precise statistic that 'road deaths fell by 18%', which gives the claim a checkable, factual basis and makes it persuasive."

2. State a genuine WEAKNESS, with a quotation and a reason. e.g. "However, it is weakened when the writer relies on a single anecdote — 'my neighbour was nearly killed' — to prove a national trend, because one case cannot establish a general pattern."

3. Reach a SUPPORTED judgement. Weigh the two and decide: is the argument convincing overall? e.g. "On balance the argument is moderately convincing: its statistical opening is strong, but its reliance on emotive anecdote in place of further data limits how far it persuades a sceptical reader."

Three rules keep this in the top band:

• Balance beats a list. At least one real strength AND one real weakness; a one-sided answer is capped lower.
• Quote, don't paraphrase. Every point of evaluation is anchored in a short quotation from the text — an unsupported claim about the text is itself unconvincing.
• Commit. End with a verdict ('moderately convincing', 'persuasive in tone but weak in evidence'). 'There are strengths and weaknesses' is not a judgement.

Keep it tight: in 120 words you have room for roughly one developed strength, one developed weakness and a one-sentence verdict — chosen for being the most significant, not the first you noticed.