Should a Research Question Specify a Comparison Even When the Comparison Seems Obvious?

If Everyone Knows the Comparison, Do You Still Need to Say It?

Suppose your study evaluates a new teaching strategy. You write:

“Does the new teaching strategy improve students' examination performance?”

The intended comparison may seem obvious. Presumably, you mean improve performance compared with students who do not receive the strategy, students receiving usual instruction, or perhaps the same students before the intervention.

But those are not necessarily the same comparison.

Whether a comparison should appear explicitly in the research question depends on whether it is part of what gives the question its meaning. In many comparative studies, leaving it unstated creates genuine ambiguity. In other research questions, no comparator is required at all.

Specify the Comparison When It Changes What the Question Means

A Comparison Is Part of the Scientific Question, Not Merely a Design Detail

Many quantitative research questions ask whether one condition differs from another. The comparison may involve an intervention versus usual care, an exposure versus no exposure, one diagnostic method versus a reference standard, one group versus another, or one condition versus a meaningful baseline.

In such questions, the comparator helps define what the result means.

This is reflected in PICO, a widely used framework in clinical and health research: Population, Intervention, Comparison, and Outcome. The comparison identifies the main alternative against which the intervention or exposure is evaluated. Variants such as PECO use the same basic logic for exposure-based questions, while PICOT adds time when it is relevant.

These frameworks are particularly useful for comparative questions, but they are not universal templates for all research. As with using frameworks such as PICO, PICOT, or SPIDER, the structure should fit the question rather than dictate it.

“Better” Has No Meaning Without a Reference Point

Consider:

“Does AI-assisted feedback improve students' academic writing?”

Improve compared with what?

Possible comparisons include:

• no feedback;
• instructor feedback;
• peer feedback;
• automated non-generative feedback;
• students' own performance before receiving AI-assisted feedback;
• another form of AI-assisted feedback.

A positive result against no feedback would not necessarily establish superiority to instructor feedback. Likewise, improvement from baseline does not by itself establish that the intervention caused the improvement if other explanations remain plausible.

The comparator therefore determines the claim the study is positioned to evaluate.

An “Obvious” Comparator May Not Be Obvious to Someone Else

Researchers often work so closely with a project that unstated assumptions begin to feel self-evident.

Suppose a researcher asks:

“Is a flipped classroom more effective for teaching programming?”

The researcher may have conventional lectures in mind. A reader might instead assume blended learning, self-paced online instruction, or whatever teaching approach is currently standard in that institution.

If those alternatives would produce different studies, the comparison is not sufficiently obvious to leave unspecified.

The Comparison Should Match the Decision You Are Trying to Inform

A comparator can be methodologically convenient yet scientifically unhelpful.

Imagine that a university is deciding whether to replace instructor feedback with an AI-supported feedback system. Comparing AI feedback with no feedback may demonstrate that receiving feedback is better than receiving none. But that is not the decision the university faces.

The relevant question may instead require comparison with the current instructor-feedback practice.

Choosing a comparator is therefore not merely about finding a control group. Ask what alternative would actually matter if the study were used to inform a scientific or practical decision.

Different Comparators Can Produce Different Conclusions About the Same Intervention

An intervention does not possess an abstract quantity called “effectiveness” independent of comparison and context.

A new approach might outperform no intervention, perform similarly to usual practice, and perform worse than an established alternative. All three findings could be simultaneously true.

For this reason, comparative questions should make clear what claim is being tested. “Does X work?” often conceals a more useful question: “Compared with what, for whom, and for which outcome?”

This becomes especially consequential when a question uses causal terms such as “effect” without a design capable of supporting the intended inference. An explicit comparison improves clarity, but it does not by itself establish causality.

Comparison Does Not Always Mean a Separate Control Group

When researchers hear “comparison,” they sometimes imagine only an intervention group and a separate control group.

Comparative evidence can take several forms depending on the design. Researchers may compare two independent groups, different interventions, exposed and unexposed populations, different conditions experienced by the same participants, measurements across relevant periods, or a diagnostic test against an appropriate reference standard.

The research question should describe the scientifically relevant contrast. The precise implementation of that contrast belongs to the study design.

Some Research Questions Do Not Need a Comparator

Not all worthwhile research asks whether one thing differs from another.

A descriptive question might ask:

“What proportion of first-year students report using generative AI for assessed coursework?”

A qualitative question might ask:

“How do doctoral students experience institutional policies governing generative AI use?”

Neither question inherently requires a comparison group. Adding one merely because a framework contains a “C” would alter the research question rather than improve it.

This is consistent with the broader principle that a research question can be descriptive without asking about relationships or effects.

The Comparator Can Sometimes Be Implied Without Creating Ambiguity

There is a difference between a comparison being absent from the scientific question and a comparison being linguistically implicit.

For example, “Is intervention A non-inferior to the current standard treatment for outcome Y?” already communicates a comparison, although the exact standard treatment should be defined elsewhere and may need to be named in the question when ambiguity is possible.

Similarly, a tightly established context may make a comparator readily understood. But brevity should not come at the cost of reproducibility or interpretability. If naming the comparator removes a consequential ambiguity with only a few additional words, doing so is usually worthwhile.

Do Not Add a Comparator Merely to Make the Question Look More Rigorous

Researchers sometimes turn a naturally descriptive or exploratory question into a comparative one because comparison appears more sophisticated.

That can create arbitrary groups.

If your actual purpose is to describe how early-career researchers experience peer review, you do not automatically need to compare them with senior researchers. The additional group requires justification, participants, analysis, and interpretation. It should exist because the contrast answers a meaningful question, not because two groups look more empirical than one.

The same principle applies when deciding whether adding more detail actually improves a research question. Every added element should perform useful conceptual work.

How the Comparator Changes the Question You Are Answering

Common Mistakes When Choosing or Stating a Comparison

Ask What Your Result Will Be Relative To

If your question contains words such as “better,” “worse,” “higher,” “lower,” “difference,” “effect,” “improvement,” or “more effective,” identify the reference point implied by that language.

Then ask whether stating it would materially clarify the question.

A useful comparator sharpens the scientific contrast. An unnecessary one merely adds another group to recruit, measure, analyze, explain, and eventually defend to Reviewer 2.

Before Finalizing the Comparison in Your Research Question

Questions About Comparators in Research Questions

Make the Reference Point Clear When the Question Depends on It

How to Cite This Guide

This guide is intended to be read, shared, and used in research, teaching, and academic work. If you draw on its ideas, explanations, or other content, please acknowledge the source by citing the guide. Doing so gives appropriate credit and helps your readers locate the original resource.