How to Identify What a Research Study Was Actually Trying to Answer?

What question did the study really address?

A paper may tell you that it investigated whether an intervention “works,” whether an exposure “affects” an outcome, or whether two variables are “related.” Those descriptions sound like research questions, but they are often too broad to tell you what the study actually answered.

To understand a study, you need to reconstruct the question at the level at which the research was actually conducted: who was studied, what was examined, what comparison was made, what outcome was measured, when it was measured, and what relationship or effect the analysis estimated. This matters because the conclusions you can reasonably draw are constrained by that operational question, not by the breadth of the topic or the ambition of the introduction.

Reconstruct the question from what the researchers actually did

How to identify the question the study actually addressed

Start with the stated objective, but treat it as a claim to verify

Look first for sentences such as “we aimed to,” “the objective was,” “we investigated whether,” or “we hypothesized that.” Reporting guidance such as STROBE asks authors of observational studies to state specific objectives, including prespecified hypotheses. CONSORT 2025 similarly calls for randomized-trial reports to state specific objectives related to benefits and harms.

A clearly stated objective is useful because it tells you what the investigators say they intended to examine. It does not, by itself, establish that the study design, measurements, and analysis actually addressed that objective.

Sometimes the question is stated plainly. In other papers, you may have to infer the research question from several parts of the paper. That reconstruction should be conservative. You are trying to identify what the evidence addresses, not to write a better research question on the authors' behalf.

Separate the broad research problem from the study question

A research problem can be much broader than an empirical study. For example, researchers may be concerned about whether remote work affects employee well-being. A particular study might survey employees from three companies at one point in time and examine the association between self-reported days working remotely and a psychological well-being score.

Those are not equivalent questions. “Does remote work affect employee well-being?” suggests a broad causal question. “Among employees in these companies, is self-reported remote-work frequency associated with concurrent well-being scores?” is closer to what that particular cross-sectional study may have addressed.

Identify who the question is about

The population is part of the question, not background decoration. A study conducted among first-year university students, adults attending one specialist clinic, or employees of a particular organization does not automatically answer the same question for all students, all patients with the condition, or all workers.

Begin by identifying the population that was actually studied. Later, you can consider how far the findings might generalize beyond it. For now, the goal is simply to state accurately who generated the evidence.

Identify what was examined

Next, determine the central exposure, intervention, predictor, condition, experience, or phenomenon. The appropriate term depends on the study. In a randomized trial, this may be an assigned intervention. In an observational study, it may be an exposure. In qualitative research, the central phenomenon might instead be an experience, process, practice, or meaning that participants were asked to describe.

Do not substitute the broad topic for what was actually examined. “Social media use,” for example, could mean time spent on a platform, frequency of checking, active posting, passive browsing, problematic-use scores, or exposure to particular content. Those operationalizations can produce meaningfully different questions.

Determine whether there was a comparison

Many research questions become clearer once you identify what is being compared. A clinical trial might compare a new treatment with usual care. An observational study might compare people with different exposure levels. A before-and-after study may compare the same participants across time. Some descriptive and qualitative studies may have no formal comparison group at all.

Do not invent one because a familiar framework seems to require it. Instead, identify the comparison the researchers actually made, if any.

Identify the outcome or object of explanation

The outcome is what the study ultimately measured, estimated, predicted, explained, described, or interpreted. Be precise about its operational definition.

“Academic achievement” could mean course grades, grade point average, standardized-test performance, pass rates, or another measure. A study using one of these has not automatically studied every reasonable meaning of academic achievement.

This is why identifying the outcome that was actually measured can materially change your understanding of the research question.

Include timing when timing changes the question

A treatment effect measured after two weeks is not necessarily the same research question as the effect measured after two years. Likewise, a contemporaneous association between an exposure and outcome differs from a longitudinal question in which exposure precedes the outcome.

When timing matters, include when the outcome was measured in your reconstruction of the question.

Let the study design constrain the verb you use

The verb in your reconstructed question carries methodological meaning. “Describe,” “estimate,” “compare,” “associate,” “predict,” “explore,” and “cause” do not make equivalent claims.

For example, a cross-sectional observational analysis can estimate an association between measured variables, but the mere presence of that association does not establish that changing one variable would cause the other to change. Conversely, a properly conducted randomized trial may be designed to estimate causal effects of assignment to an intervention under specified conditions.

Check the primary analysis because it reveals the statistical question

Two studies can collect similar variables but ask different questions through their analyses. One may estimate the difference in mean outcomes between groups. Another may estimate an adjusted association. A third may focus on whether an exposure predicts an outcome after accounting for selected covariates.

Find the primary analysis and ask what quantity it was intended to estimate. In quantitative research, this may be a mean difference, risk ratio, odds ratio, hazard ratio, correlation, regression coefficient, diagnostic-accuracy measure, or another estimand or parameter. The statistical method is not itself the research question, but it can reveal exactly how the question was operationalized.

Distinguish the primary question from questions discovered along the way

A paper may report subgroup analyses, additional outcomes, alternative models, sensitivity analyses, and exploratory findings. These can be informative, but they should not automatically be treated as the question that originally motivated the study.

STROBE explicitly distinguishes stated objectives and prespecified hypotheses from important analyses that arose during data analysis. CONSORT 2025 likewise asks trial reports to identify important changes after a trial began, including outcomes or analyses that were not prespecified.

When this distinction matters, determine what was prespecified and what appears to have been decided later. Otherwise, a striking secondary result can quietly replace the original question in your interpretation of the paper.

A useful reconstruction often resembles PICO, but PICO is not universal

For intervention and many clinical questions, the familiar PICO structure can be useful: Population, Intervention, Comparator, and Outcome. Evidence-based medicine resources commonly use these components to structure focused questions.

But forcing every study into PICO can distort the research. Observational studies may be better described in terms of population, exposure, comparison, and outcome. Diagnostic, prognostic, qualitative, descriptive, methodological, and exploratory studies may require different elements. Use the framework that exposes the actual structure of the question rather than the framework whose acronym you happen to remember from methods class.

Turning a broad claim into the question the study actually answered

Common mistakes when identifying what a study was trying to answer

Use the reconstructed question as the anchor for critical appraisal

Before deciding whether a result is convincing, useful, generalizable, or important, write down what you think the study actually asked. One sentence is usually enough.

Then test that sentence against the paper. Can you point to the participants who generated the data? Can you identify the exposure, intervention, or phenomenon? Is the comparison real rather than assumed? Does your wording name the measured outcome rather than a broader construct? Does the verb fit the design? Does the primary analysis correspond to the relationship you have described?

This exercise gives the rest of your appraisal a stable reference point. Instead of asking vaguely whether “the study is good,” you can ask whether its design and execution provide a credible answer to the particular question it actually addressed. Academic papers become considerably easier to interrogate once the question stops moving around.

Before interpreting the findings, reconstruct the study question

Questions about identifying a study's real research question

The study question is defined by more than the sentence stating its aim

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.