WiderAI › Blog › Examples in Task 2: What Counts as Support (No Fake Statistics)

Examples in Task 2: What Counts as Support (No Fake Statistics)

IELTS Writing29 August 2026 · 8 min read · WiderAI Team

You do not need statistics, studies or citations to support your ideas in IELTS Writing Task 2 — and inventing them does not raise your score. What examiners reward under "support" is development: an idea that is stated, explained, and made concrete enough that a reader can see why it is true. A clear explanation followed by a plausible, ordinary example is full support at any band.

This matters because a whole industry of exam advice teaches the opposite. Candidates arrive trained to write "According to a recent study, 73% of teenagers..." — a study that does not exist, with a number chosen for flavour. The examiner cannot verify your facts and is not trying to; what they can see instantly is the shape of an invented statistic, because thousands of scripts use the same one. It adds nothing to the assessment of your ideas, it spends words that could have developed the argument, and when the fake fact is implausible it actively damages the impression of a clear, credible essay.

So drop the fear that your essay is "unsupported" without data. The task instruction asks you to give reasons for your answer and include relevant examples from your own knowledge or experience — that last phrase is the permission slip. Your own knowledge and experience means the ordinary world you observe: workplaces, schools, cities, families, technology in daily use. That is the evidence base IELTS expects.

What do examiners actually mean by support?

In the assessment of task response, the difference between the middle bands and the higher ones is largely about development. A mid-band essay presents relevant ideas but leaves them thin — stated once, then abandoned for the next idea. A higher-band essay presents fewer ideas and extends each one: it explains the mechanism, illustrates it, and connects it back to the question.

Support, in other words, is not a citation. It is the answer to the reader's silent question "why is that true?" after every claim you make. There are several ways to answer it, and all of them count:

• Explanation of mechanism: Remote work reduces commuting, which returns two or three hours a day to employees — time that typically goes to rest, family or exercise. No example yet, but the claim is already supported, because the causal chain is visible.
• A typical case: A nurse working night shifts, for instance, can schedule medical appointments and childcare far more easily when administrative tasks can be done from home. Not a named person, not a study — a recognisable instance.
• A consequence or contrast: Where cities have invested in reliable bus networks, car use falls without any ban being needed; where they have not, restrictions simply provoke resentment.
• Personal or local observation, generalised: In Uzbekistan, the spread of contactless payments has visibly changed how small shops operate. Your own country is a legitimate source of examples, and specific-sounding local knowledge reads as far more authentic than an invented global survey.

A body paragraph that runs claim → explanation → illustration → link back to the question is fully developed. Two such paragraphs outscore four paragraphs of bare claims every time.

Why do invented statistics hurt more than help?

First, they earn nothing. There is no criterion that rewards factual data, so "80% of students" scores exactly the same as "many students" — the band is decided by how well the idea is developed and expressed, not by numerical dressing.

Second, they are recognisable. Examiners read enormous numbers of scripts, and the invented statistic has a signature: a suspiciously round or suspiciously precise percentage, an unnamed "recent study", a university chosen for prestige. It signals a memorised strategy rather than genuine engagement with the question, and memorised content is exactly what examiners are trained to discount.

Third, they create risk where none existed. Write "a 2019 Harvard study proved that homework is useless" and you have committed the essay to a claim that sounds false, which undermines the calm, credible tone an argumentative essay needs. Write "many teachers observe that excessive homework demotivates younger pupils" and you have made a claim no reader resists — the argument moves forward without a hostage.

Fourth — and least discussed — fake precision often produces logical nonsense. "90% of people who exercise live longer" invites the question longer than what? and the sentence collapses under its own numbers. Vague-but-honest quantifiers (most, many, a growing number, in many countries) are not weak writing; they are accurate writing, and accuracy of claim is part of what makes prose read as proficient.

If you genuinely know a real, well-established fact — that smoking causes cancer, that global temperatures have risen — use it plainly and without fabricated numbers. Well-known truths need no citation. What you should never do is decorate them with invented figures to look scientific.

What does a well-developed example look like?

Take a common question: Some people think companies should allow employees to work from home. Do you agree? Here is a thin paragraph:

Working from home has many benefits. Employees save time and money. Also, they are less stressed. According to research, 68% of workers are more productive at home. Therefore, companies should allow it.

Four claims, zero development, one fabricated number. Now the same idea, developed:

The strongest argument for home working is the time it returns to employees. In large cities, a commute of an hour each way is normal, which means office workers surrender ten hours a week before any work is done. When that travel disappears, the hours reappear as sleep, exercise or family time, and a rested employee concentrates better than an exhausted one. A programmer who starts the day after breakfast rather than after a crowded metro ride is simply in a better condition to work.

One idea, but the reader is walked through the mechanism and shown a concrete, believable instance. Nothing needed inventing, because the example is drawn from ordinary observable life — which is precisely the register of evidence the task asks for.

Hypothetical examples work too, if you frame them honestly: If a small business had to pay for office space in central Tashkent, the saving from even two remote days a week would be substantial. The conditional framing tells the truth — this is an illustration, not a report — and illustration is all the task requires.

Which mistakes waste the most words?

• The example dump: for example, phones, laptops, tablets and smartwatches — a list is not an example. An example is an instance examined, not items named.
• The irrelevant personal story: My cousin failed his exam because of computer games supports almost nothing on its own. Generalise it: when gaming displaces sleep, results follow — a pattern most families with teenagers will recognise turns anecdote into argument.
• Fake research, fake famous people: invented quotes from Einstein score no better than invented percentages, and for the same reasons.
• The example that changes the subject: a paragraph about the cost of university that illustrates itself with a story about online courses has quietly switched topics, and coherence pays the bill.
• Supporting the easy claim: candidates develop the obvious point at length and leave the essay's most contestable claim bare. Spend your best example on the claim a sceptical reader would resist most.

Every mistake on this list is cheaper to fix at the planning stage than during writing: choose each main point and its illustration together, and the mid-paragraph scramble that produces invented studies never happens. There are more essay walkthroughs and criterion-by-criterion guides on the WiderAI blog.

The deeper shift is this: stop thinking of examples as decoration required by a rubric, and start thinking of them as the argument itself. A claim the reader can picture is a claim the reader accepts. That is what "relevant, extended and supported" means in practice — and no percentage sign is involved.