01 · The Question
Where in the causal chain is your research problem?
Suppose you want to study low academic performance. You discover that students with poor attendance tend to perform less well, so absenteeism becomes central to your study.
But what exactly are you investigating?
Absenteeism could be treated as a possible cause of lower performance. Yet absenteeism may itself be a consequence of other conditions, such as health difficulties, employment demands, transportation problems, disengagement, course scheduling, or experiences at school. Those conditions may have consequences of their own.
This is why asking whether something is “the cause” or “the consequence” can be misleading. Many research problems sit within a longer chain of relationships. Your task is not necessarily to trace that chain to its ultimate beginning. It is to identify which part of it your study is examining and avoid claiming more than your evidence and research design can establish.
03 · What You Need to Know
Causes and consequences depend on the relationship you are investigating
A phenomenon is not inherently a cause or a consequence
Consider employee burnout. In one study, burnout might be examined as a possible consequence of excessive workload. In another, it might be investigated as a predictor of intention to leave. In a longer explanatory sequence, workload may contribute to burnout, which may in turn contribute to turnover intentions or actual turnover.
Burnout has not changed. Its analytical position has.
This means you should avoid asking whether a variable “is a cause” in the abstract. Ask instead: a possible cause of what? A consequence of what?
| Phenomenon |
When treated as a possible consequence |
When treated as a possible cause or explanatory factor |
| Student absenteeism |
Examining factors associated with students missing classes |
Examining whether attendance patterns contribute to differences in achievement |
| Teacher burnout |
Examining workload, working conditions, or organizational support |
Examining retention, absenteeism, or instructional outcomes |
| Technology anxiety |
Examining prior experience, support, or characteristics of a technology |
Examining technology adoption or avoidance |
| Patient nonattendance |
Examining access, scheduling, communication, cost, or other barriers |
Examining continuity of care or subsequent health-service use |
The examples illustrate possible research relationships, not established causal claims. Whether any proposed relationship is causal depends on the evidence.
Start by separating the observed condition from its proposed explanation
Suppose institutional records show unusually high employee turnover. That observation establishes an outcome that may require explanation. Interviews then suggest that some employees perceive workload as excessive.
You now have at least two pieces of information: turnover is occurring, and some employees report workload concerns. You do not yet necessarily have evidence that excessive workload caused the turnover.
A common mistake is to collapse those steps into a problem statement such as “excessive workload is causing high employee turnover.” Unless the causal relationship is already well supported in the relevant context, the wording may decide the explanation before the study tests it.
This is one reason researchers need to distinguish an observable condition from the underlying problem that may be producing it.
Draw the proposed causal sequence before writing it into the problem
A simple causal map can expose assumptions that are difficult to notice in prose.
Imagine that a researcher proposes the following sequence:
Possible antecedent Students experience unstable internet connectivity.
Possible intermediate consequence Students participate less frequently in synchronous online sessions.
Possible later consequence Students miss instructional interactions and learning activities.
Possible academic outcome Academic performance may be affected.
This map does not establish that the sequence is correct. Its value is conceptual. It forces you to state which relationships you are proposing and shows that an intermediate condition can simultaneously be a consequence of something earlier and a possible cause of something later.
Temporal order matters, but it is not enough to establish causation
For one phenomenon to cause another, the proposed cause must precede its effect. If an outcome occurred before the supposed cause, that causal explanation cannot work in the direction proposed.
Yet temporal precedence alone is insufficient. Ice cream sales may increase before some outcome without causing it. Two phenomena can change together because of another factor, selection processes, contextual changes, or coincidence.
Research design therefore matters greatly when you move from describing an association to making a causal claim. Depending on the question, causal inference may require attention to temporality, comparison conditions, confounding, alternative explanations, measurement, selection, and the assumptions of the analytical approach.
Watch Out
Words such as causes, leads to, results in, produces, affects, and impacts make causal claims. Do not use them merely because two variables are associated or because one explanation seems plausible.
A consequence can become a legitimate research problem
Researchers sometimes assume that a study is superficial unless it reaches the deepest possible cause. That is not a useful rule.
Suppose an intervention has already been shown to increase teachers' administrative workload, but little is known about how that additional workload affects instructional planning. Investigating the consequence can be theoretically or practically important even if the study does not explain why the intervention increased workload in the first place.
Likewise, research might examine the consequences of misinformation, natural disasters, policy changes, disease, discrimination, technological disruption, or economic shocks without claiming to explain their origins.
The relevant question is whether the consequence represents a meaningful uncertainty that research can address.
You do not need to find an ultimate cause
Once researchers begin moving backward through a causal chain, there is rarely an obvious place where they must stop.
If low achievement is associated with absenteeism, what causes absenteeism? If transportation difficulties contribute to absenteeism, what causes the transportation difficulties? If household income affects transportation access, what explains household income?
A single study cannot indefinitely follow every causal branch. Eventually, you need a theoretically meaningful and methodologically feasible boundary.
When the explanatory chain expands beyond what one project can reasonably investigate, you may need to reduce the problem to a manageable part of the larger system.
Reverse causality may complicate apparently simple explanations
Sometimes the presumed direction of influence is uncertain.
Imagine an association between low academic confidence and poor academic performance. Lower confidence might contribute to poorer performance, but poor performance might also reduce confidence. Both processes could operate over time.
Cross-sectional evidence showing an association at one point in time generally cannot resolve such temporal ordering by itself. The research problem should reflect that uncertainty rather than presenting one direction as established without adequate evidence.
Another variable may produce both the supposed cause and consequence
A relationship between two variables can also arise because both are influenced by another factor.
Suppose students who participate less in online classes also receive lower grades. Reduced participation may contribute to lower achievement. But employment obligations could plausibly reduce participation while independently limiting study time. Health, prior preparation, course difficulty, or other factors might also matter.
This is why identifying a plausible mechanism is different from demonstrating a causal effect. Your research problem should preserve that distinction.
The wording of your problem can quietly determine the causal direction
Compare “the effects of teacher resistance on technology adoption” with “factors associated with variation in technology adoption.” The first formulation assumes a directional relationship and identifies resistance as an explanatory factor. The second is more open.
Neither form is automatically superior. If theory and prior evidence provide sufficient justification for examining a particular causal pathway, a focused causal formulation may be appropriate. If the relationship remains uncertain, however, framing the research problem too early around one explanation can constrain what the study is able to discover.
04 · A Practical Example
Is low engagement the cause, the consequence, or both?
Hypothetical Example
Low engagement in an online course
A researcher notices that students who interact less frequently with an online course tend to have lower completion rates. The researcher initially describes low engagement as the cause of noncompletion.
Observed relationship Course data show that lower recorded engagement and noncompletion occur together.
Initial causal interpretation The researcher assumes that low engagement causes students to stop completing the course.
Alternative direction Students who have already decided to withdraw may begin engaging less, meaning impending noncompletion could partly explain declining engagement.
Possible common causes Workload, employment, academic difficulty, technical access, health, or competing responsibilities might influence both engagement and completion.
Better problem formulation Unless stronger evidence supports a causal direction, the study can focus on the relationship between patterns of engagement and course completion while investigating plausible explanations for that relationship.
Research implication The researcher can choose a design appropriate to the intended inference rather than treating an observed association as a causal finding from the outset.
The example illustrates why causal language should follow from the research question, design, and evidence rather than from the intuitive order in which the researcher first notices the variables.
06 · What This Means for You
Choose the part of the causal chain your study can actually investigate
Begin with a neutral description of what is already known. Then write down the explanatory relationship you think may exist. Place the proposed antecedent before the outcome and identify any intermediate steps you believe connect them.
Next, challenge that sequence. Could the direction operate in reverse? Could another factor influence both variables? Is the proposed mechanism supported by theory or previous evidence? Are you studying an association, a prediction, a mechanism, or a causal effect?
Your wording should then match the level of inference your study is designed to support.
A simple decision framework
If you only know that two phenomena occur together
Describe the relationship without automatically labeling one as the cause of the other.
If theory and prior evidence suggest a direction but uncertainty remains
Present the causal pathway as something to investigate rather than as an established fact.
If your research question genuinely concerns causal effects
Use a design and analytical strategy capable of supporting the intended causal inference and state the required assumptions.
If your focal phenomenon is a consequence of something outside your scope
Study the consequence if it represents an important unresolved question, while making the boundary explicit.
If the causal chain extends far beyond your resources
Select a theoretically and practically defensible segment rather than claiming to explain the entire problem.
The goal is not to make every research project causal. It is to know where your study sits within the proposed chain and ensure that your claims remain proportionate to what the design can establish.
07 · A Quick Checklist
Before calling something a cause or a consequence
Check the proposed relationship:
State the outcome or condition you are trying to explain.
Identify the factor you believe may precede or influence that outcome.
Draw the proposed sequence so you can see where each phenomenon sits in the causal chain.
Check whether the supposed cause actually occurs before the proposed consequence.
Consider whether reverse causality could plausibly explain the relationship.
Identify plausible common causes or alternative explanations.
Distinguish evidence of association from evidence capable of supporting causal inference.
Confirm that causal wording in the problem statement matches what the proposed design can support.
Set a defensible boundary rather than attempting to trace every possible cause and consequence.
