How to Write an Article Review: A Step-by-Step Guide for College Students

Every term I assign an article review, and every term about a third of the class hands me a summary with a compliment attached to the end. Those papers are not bad. They are accurate, they are organized, and they are not reviews. A review is a verdict on an article, supported by evidence from the article itself, and the summary exists only so the reader can follow the verdict. If your draft could be printed as the abstract of the piece you read, you have not started yet.

This guide is the talk I give before the assignment goes out, written down so you can reread it at midnight the night before it is due.

What an article review is, and what it is not

An article review is a short academic paper, typically one to three pages, in which you evaluate one published article. The article is usually a scholarly article from a peer reviewed journal, sometimes a chapter or a long piece from a serious magazine, and the review’s job is to decide how well it does what it set out to do. You describe the argument so your reader knows what is being judged, and then you judge it: the logic, the evidence, the method, the fit between conclusion and data.

Students confuse the assignment with two neighbors. The summary stops after the description. The reaction paper skips the description and goes straight to feelings. A review sits between them: fair to the author the way a summary is fair, and committed to a position the way a reaction is, except that the position has to be earned with page references.

Feature	Summary	Article review
Main question	What does the author say?	How well does the author say it, and is it true?
Share of the paper	Nearly all of it	About a quarter, sometimes less
Author’s evidence	Reported	Tested against the claim it supports
Method	Named, at most	Examined for fit, sample, measures, limits
Your position	Absent	Stated in the first paragraph and defended
Page references	Optional	Required for every criticism
Typical grade for the other one	Fine, if a summary was assigned	A summary handed in as a review caps around a C

That last row is the grading rubric in most departments, and I would rather you hear it now than read it in the margin.

Read the article as a critic before you write a word

The single biggest predictor of a good review in my grade book is whether the student read the article more than once. The first reading is for comprehension. You are finding out what the author claims, how the paper is organized, and what the conclusion says. Do not annotate much on this pass. Almost everything looks important on a first read, and a page covered in highlighter is a page you will have to reread anyway.

The second reading is the one that produces a review, and it is a different activity. Now you read with a question at the top of every page: does this support the claim? Mark the thesis, then trace each piece of evidence back to the part of the thesis it is supposed to carry. Note the sample size, the time period, the population studied, the definitions of key terms. When the author cites a study, check whether they explain what that study measured or just borrow its conclusion. When they concede a limitation, ask whether the conclusion later ignores it. Write your reactions in the margin with the page number, in full sentences if you can bear it, because a margin note reading “weak” is useless three days later while “sample is 40 managers at one firm, conclusion talks about all remote workers, p. 6” is already half a paragraph of your review.

Some articles need a third pass, usually for the methods section in an empirical paper. A thirty page study written for specialists does not surrender its weaknesses in one sitting, and the reviews that impress me are the ones where the student clearly went back.

The structure of a review

There is no single mandated outline, but nearly every good review I have graded follows the same four moves. The proportions are what students get wrong, so I have put rough word counts against each move for a 1,000 word review.

1. Introduction, about 100 words. Give the full citation in the required style, name the author’s central claim in one sentence, and state your verdict. The verdict goes here, not in the conclusion. A reader who reaches paragraph five without knowing whether you found the article convincing has been reading a mystery novel, and reviews are not mystery novels.
2. Summary, about 200 words. Explain the argument in your own words in the order the author made it. Include the research question, the method in a sentence or two, and the main findings. Stay neutral. This is the one section where your opinion is unwelcome, and the reader uses it to check that you understood the piece before you started disagreeing with it.
3. Critique, about 550 words. This is the review. Organize it by issue rather than by page: one paragraph on the argument’s logic, one or two on the evidence, one on the method, one on what the author did well. Every paragraph makes one judgment, supports it with a page reference, and explains why the point matters for the conclusion.
4. Conclusion, about 150 words. Restate the verdict in light of what you showed, say what the article contributes despite its flaws or why the flaws outweigh the contribution, and name one question the article opens for further work. Do not introduce new criticisms here.

Notice that the summary is a fifth of the paper and the critique is more than half. If your draft inverts those numbers, the fix is not to trim the summary. It is to write the critique you skipped.

How to evaluate an argument, its evidence and its method

Students tell me they do not feel qualified to criticize a published scholar. Fair enough, and also beside the point. You are not being asked to out-research the author. You are being asked to check whether the article does what it claims, and that check has parts a careful undergraduate can perform.

Start with the argument as a chain. Write the thesis at the top of a page, then list the claims the author needs to be true for the thesis to follow. Now ask which of those claims the article actually supports and which it assumes. Most published articles rest on at least one unsupported step, often a definition that quietly does the work. If an author argues that social media harms adolescent wellbeing and measures wellbeing with a single self report question, the whole argument leans on that question meaning what the author says it means. Pointing that out is a legitimate criticism and a strong one.

Then look at the evidence in relation to the claim it carries. How large is the sample and who is in it? Does the population studied match the population in the conclusion? Are the numbers correlations presented as causes? Is a single study doing the work of a literature? Does the evidence come from the author’s own data or from citations, and if from citations, are they characterized accurately? You can answer most of these from the article itself without opening another source.

Method is where empirical articles live or die, and where students most often give up. Do not give up. You may lack the statistics to check a model, but the scientific method has parts that need no statistics: a question, a way of observing, a comparison, and an inference. Ask whether the comparison exists. A study that shows productivity rose after a policy change, with no group that did not get the change, cannot attribute the rise to the policy. Ask whether the measures capture the concept. Ask whether the author’s limitations section admits the weaknesses you found, because an honest limitations section is a strength worth crediting and an absent one is a finding in itself.

Praise is part of evaluation too. A review that finds nothing good in a peer reviewed article is usually a review that did not read carefully. Name the choice that made the article credible, whether a large representative sample, a clever comparison group, or a conclusion that stays inside its data, and say why that choice mattered.

Specific beats harsh, every time.

A short worked example

Suppose the assigned article argues that four day workweeks improve employee wellbeing without reducing output, based on a trial across several dozen firms. I will treat the article as hypothetical, since I am illustrating the moves rather than reviewing a real paper.

The introduction cites the article, states the claim in a sentence, and delivers the verdict: convincing on wellbeing, overstated on productivity. The summary explains the trial in neutral terms, including how wellbeing and output were measured and over what period.

The critique then opens on the productivity claim, because that is where the verdict rests. It might read like this:

The productivity finding depends entirely on self reported output from participating managers (p. 9), collected six months into a trial the firms volunteered for. Volunteer firms are firms that expected the schedule to work, and managers who lobbied for the change have reason to report that it did. The article acknowledges the self report problem in a single sentence (p. 14) and then presents the productivity result without qualification in the conclusion (p. 17). The wellbeing result rests on a validated survey instrument administered before and after, which is a far stronger design, and the contrast between the two measures is the article’s central weakness.

That paragraph names the claim, cites the exact evidence, explains why the evidence is weaker than the claim, and ties the criticism to the verdict. The following paragraphs would credit the wellbeing measurement, question whether six months is long enough for novelty effects to fade, and note what the article never measures, such as customer outcomes. The conclusion restates the split verdict and suggests that a trial with randomly assigned firms would settle the productivity question.

None of that requires expertise the author lacks. It requires reading the methods section twice and refusing to let the conclusion claim more than the data showed.

Common mistakes I mark down

The mistake that costs the most points is the one this guide keeps returning to: summarizing instead of evaluating. It appears in disguise, too. A student writes “the author then discusses the limitations of the study,” which describes the article’s structure without saying whether the limitations are the right ones or whether the conclusion respects them. Any sentence that could have been written by someone who agreed with the author completely and someone who disagreed completely is a summary sentence, and the critique should contain almost none of them.

A related failure is the verdict with no address. “The evidence is unconvincing” tells me you had a reaction. “The evidence is a single survey of 40 managers at one company, p. 6, offered in support of a conclusion about the national workforce” tells me you read the article. The second version is longer, and it is the one that earns credit.

Students also bury the verdict at the end, organize the critique page by page like a reading log, and criticize the article for not being a different article. An article about hospital nurses in Ontario is not flawed because it says nothing about teachers in Texas. It is flawed only if its conclusion talks about workers in general while its data do not. Scope is a criticism when the author overreaches, not when the author stayed home.

One smaller habit to break: quoting the article at length in the summary, which proves you can copy rather than that you understood.

Before you submit, run one test. Read the critique and delete, in your head, every sentence that makes a judgment. If what remains still reads as a coherent paper, you wrote a summary and it is time to go back to the margins of your second reading, where the review has been waiting the whole time.