How to Write a Science or Technology Argument Essay

A few years ago a student handed me an essay called “The Dangers of Social Media.” Twelve hundred words, nine sources, clean paragraphs, and not one sentence I could disagree with. That was the problem. It explained that teenagers use phones a lot, that some researchers worry about this, and that more research is needed. It was a tidy report wearing an argument’s clothes. When I asked her what she was claiming, she paused and said, “That it’s bad?” And I asked: bad compared to what, for whom, by how much, according to which evidence? She did not know yet. The second draft, two weeks later, was the best thing she wrote all term, and this guide is essentially what I told her in between.

Science and technology topics attract this failure more than any other kind of essay. The material feels authoritative, so students report it. The research feels unsettled, so they hedge everything into mush. The skill I want to teach you sits between those two reflexes: take a real position, and hold it exactly as firmly as the evidence allows.

The Essay That Explains Everything and Argues Nothing

Here is the quickest diagnostic I know. Read your thesis aloud and ask who would object to it. If the answer is nobody, you have written an explanation, not an argument. “CRISPR allows scientists to edit genes with unprecedented precision” is true, interesting, and completely inert. No biologist will fight you on it. “Germline editing of human embryos should remain banned until inherited off-target effects can be tracked across at least two generations” is a claim a reasonable scientist could dispute, and that friction is what makes it an essay.

Explanatory writing has its place. In a science argument essay, it belongs in the service of a claim: you explain how a technology works only to the extent that the reader needs it to judge your position.

So the first job is not research. It is finding the disagreement.

Turning a Topic into a Question You Could Lose

I ask students to write their topic at the top of a page and then generate questions underneath it until one of them scares them a little. “Artificial intelligence” produces nothing. “Should image generators be trained on copyrighted art without permission?” produces a fight. “Will large language models reduce the number of entry-level programming jobs within ten years?” produces a fight with evidence on both sides, which is even better.

The questions that work share three features:

• Informed people currently disagree about the answer.
• Evidence exists, or could exist, that bears on it.
• You can imagine a finding that would make you change your mind.

That last one matters more than students expect. If no possible evidence could move you, you are writing a sermon. Science argument essays borrow their ethics from the scientific method itself: a claim you refuse to expose to disconfirmation is not a claim, it is a preference.

Once you have the question, your thesis is your answer to it, stated as a position. If you want a deeper walk through sharpening that sentence, my guide on how to write a thesis statement covers the mechanics. For our purposes, the thesis needs to do two things: commit to a side, and signal how confident that commitment is.

Take a Position, Then Size It to the Evidence

This is where the genre gets interesting, because scientific questions rarely have a clean yes. Students respond either by pretending the uncertainty away (“Social media causes depression”) or by surrendering to it (“The effects of social media are complex and more research is needed”). Both are failures of calibration. The first overclaims. The second refuses to claim at all.

The honest move is to argue about likelihood, conditions, and trade-offs. Compare these three sentences:

1. “Smartphones have caused the teen mental health crisis.”
2. “It is impossible to know whether smartphones affect teen mental health.”
3. “The evidence that heavy social media use harms adolescent mental health is suggestive but not yet strong enough to justify national phone bans in schools; targeted limits on use during school hours are a better fit for what we actually know.”

The third sentence is longer, and every extra word works. It names a degree of confidence (suggestive, not strong). It names a policy threshold (national bans). It proposes an alternative sized to the evidence (school-hours limits). A reader can disagree with each part, which means each part is a real claim. And notice that it still takes a side. Calibration is not the same as neutrality.

A useful test: could your thesis survive the best study on the other side being true? If one contrary finding would demolish it, you have claimed too much. If no finding could touch it, you have claimed nothing.

The Worked Example: Phones and Teenage Minds

Let me build one out, because the method is easier to copy than to describe. I am choosing the debate over smartphones, social media, and adolescent mental health because it is live, because serious researchers disagree in public, and because the temptation to overclaim on both sides is enormous.

The public case for harm was made most forcefully by the social psychologist Jonathan Haidt in his 2024 book The Anxious Generation, which argues that smartphones and social media, together with overprotective parenting, reshaped childhood and drove rising rates of anxiety and depression among young people.

The pushback came quickly from within developmental psychology. In a review published in Nature the same year, Candice Odgers, who studies adolescent development, argued that most empirical work on social media and mental health had not found a large or consistently negative effect, that the correlations that do exist might run in the other direction (struggling teenagers may reach for their phones more), and that the book’s central claim about technology rewiring children’s brains was, in her words, “not supported by science.” She also warned that fixating on phones could pull attention away from other causes of youth distress.

A separate line of research using very large survey datasets has reported that the statistical association between technology use and adolescent well-being, while often present, is small in magnitude. I am deliberately not quoting a figure here because I could not verify the exact number while writing this, and that restraint is itself part of the lesson: if you cannot check it, do not print it.

Now watch what a student can do with this. A weak thesis picks a team: “Haidt is right” or “Haidt is wrong.” A strong thesis finds the position the evidence actually supports, which is somewhere neither camp fully occupies:

“Correlational studies cannot establish that social media causes adolescent depression, and the measured associations are modest, so sweeping claims of an epidemic driven by phones overreach. But the same uncertainty cuts both ways: the absence of strong evidence for harm is not evidence of safety, and low-cost interventions such as phone-free school hours are justified as precaution even while the causal question stays open.”

That thesis has a spine. It sides with Odgers on the evidence and partly with Haidt on policy, and it explains why. It can be attacked from both directions, which means it is doing real argumentative work.

A Table for Keeping Claims Honest

When I grade these essays, the most common problem is not a missing argument but a mismatch between what a sentence claims and what its source found. I now make students build a small table before they draft. Three columns: the claim they want to make, the evidence that supports it, and the limit of that evidence. For the phones example it might look like this.

Claim	Evidence	Limit
Teen anxiety and depression rates rose in many wealthy countries after about 2012	National health surveys and clinical data cited by Haidt and others	Rising diagnosis can reflect better detection and reduced stigma, not only rising incidence
Heavy social media use is associated with lower well-being	Large cross-sectional and some longitudinal surveys	Associations are generally small; direction of causation is unclear
Phones do not cause the crisis	Odgers and others note weak, inconsistent effects	Absence of strong evidence is not proof of no effect; measurement of “use” is crude
Phone-free school hours improve focus	Early school-level reports and teacher surveys	Few controlled studies; effects on mental health specifically are unmeasured

The third column is where the essay lives. Every limit you write down is a sentence your reader would otherwise have written for you, in red, in the margin. It also tells you how firm your thesis can be: if every row has a serious limit, your conclusion should be proportionally careful.

Citing Evidence Without Hiding Behind It

Students often treat citations as armor: the more references, the safer the claim. In a science argument, citations are closer to witnesses. Each one needs to say something specific, be named so the reader can check it, and be described accurately enough that an expert would nod.

Three habits separate strong evidence handling from weak:

Name the source in the sentence, not just in the footnote. “A 2024 review in Nature by Candice Odgers argued that…” is checkable. “Studies show that…” is not, and experienced readers hear it as a confession that you have not read the studies.

Say what kind of evidence it is. A randomized trial, a longitudinal cohort, a cross-sectional survey, a case report, and an expert opinion carry very different weight. A single sentence of method (“a survey of self-reported screen time”) tells the reader how much to trust the finding and shows that you know the difference.

Quote sparingly and paraphrase precisely. A short phrase in the author’s own words can anchor a paragraph. Paraphrase in your own sentences, keep the author’s meaning intact, and resist the urge to sharpen their finding into something more dramatic than they wrote.

The Counterargument Is Not a Speed Bump

Most students include a counterargument because they were told to, and you can tell: one paragraph, a weak version of the opposing view, dispatched in two sentences so the essay can get back to agreeing with itself. This is worse than useless, because it tells the reader you have not seriously considered being wrong.

The counterargument you want is the one that worries you. In the phones essay, if your thesis leans toward “the evidence for harm is weak,” the objection that should worry you is this: large-scale harms can be real and still produce small average effects in crude surveys, especially if the harm concentrates in a vulnerable minority. That objection does not destroy the thesis, but it forces a concession: the modest average association is compatible with serious harm to some teenagers, and policy should account for that.

Notice what happened. Taking the objection seriously did not weaken the essay. It made the thesis more precise and more defensible, and it let the writer move from “we do not know” to “here is what a precautionary approach looks like given what we do not know.” That move is the difference between a hedge and an argument.

When you draft the counterargument section, write it as if you were trying to win for the other side. Then answer it. If you cannot answer it, change your thesis; that is what the scientific method would ask of a hypothesis too.

Overclaiming, and How to Catch It Before Your Reader Does

Overclaiming is the characteristic failure of this genre. It creeps in through verbs. A study “finds an association,” and by the second draft it “shows,” and by the third it “proves.” A book “argues,” and two paragraphs later its argument has become a fact. A mechanism that is “plausible” becomes “established.” Each is a small inflation, and together they leave the essay claiming things no source supports.

Here is the reading pass I do with students on the final draft. Go through and mark every sentence that asserts something about the world. For each, ask two questions: does my named source actually say this, or something smaller? And does my verb match the kind of evidence behind it? Then apply a few swaps:

• “causes” becomes “is associated with” unless there is experimental evidence
• “proves” becomes “supports” or “suggests”
• “the research shows” becomes the name of the specific study and its finding
• “experts agree” becomes the named experts and what each actually said
• “it is clear that” is deleted; if it were clear, you would not need to say so

A second check is structural. Read your conclusion next to your table of claims and limits. If the conclusion is more confident than the third column of the table allows, you have drifted. Pull the conclusion back to where the evidence left you. A careful conclusion that admits what remains unknown, and still takes a position, is the signature of someone who understands how science works. A confident conclusion that ignores the limits is the signature of someone who read the press release.

That student with the social media essay ended her second draft with a sentence I still remember the shape of: that the honest answer to whether phones were hurting her generation was that nobody yet knew, that this was a reason for caution rather than complacency, and that anyone claiming certainty in either direction was selling something. She took a side, she sized it to the evidence, and she left her reader with something to argue with. That is the whole assignment.