The one factor universally required to establish causality is temporality: the cause must come before the effect. While several other factors strengthen a causal argument, temporality is the only one considered absolutely necessary. Without it, you cannot logically claim that one thing caused another.
This question comes up frequently in research methods, epidemiology, and statistics courses, often in multiple-choice format. The answer choices typically include factors like correlation, sample size, biological plausibility, and temporal order. Understanding what each factor contributes to a causal argument helps you see why temporality stands alone as the non-negotiable requirement.
Why Temporality Is the Only Absolute Requirement
Temporality means the proposed cause must happen before the observed outcome. This sounds obvious, but in practice it can be surprisingly hard to confirm. In a cross-sectional study (one that measures everything at a single point in time), you might find that heavy alcohol use and tuberculosis often appear together. But which came first? It’s possible that the social consequences of tuberculosis led to increased drinking, not the other way around. This problem is called reverse causality, and it’s one of the main reasons researchers insist on establishing time order before making any causal claims.
Every major framework for causal reasoning treats temporality as foundational. Directed acyclic graphs (a visual tool used in modern causal inference) are built entirely on the temporal ordering of variables. The Bradford Hill criteria, the most widely taught checklist for evaluating causation, list temporality as the only criterion that is strictly necessary rather than merely supportive. You can have a valid causal relationship that lacks some of the other criteria, but you can never have one where the effect comes before the cause.
The Bradford Hill Criteria
In 1965, epidemiologist Austin Bradford Hill proposed nine factors for evaluating whether an observed association between two things is likely to be causal. These are not a rigid checklist where every box must be ticked. Instead, they’re guidelines: the more criteria a relationship satisfies, the stronger the case for causation. Here’s what each one means in plain terms.
- Temporality: The exposure comes before the outcome. This is the only criterion considered mandatory.
- Strength of association: A larger effect size makes causation more plausible. For example, if people exposed to a substance are four times more likely to develop a disease than unexposed people, that’s harder to explain away as coincidence or bias than a 10% increase would be.
- Consistency: The same relationship shows up across different studies, populations, and settings. If only one study finds an effect and dozens of others don’t, the case for causation weakens considerably.
- Specificity: The exposure leads to one particular outcome rather than a broad range of unrelated outcomes. This criterion is considered the weakest of the nine, since many real causes (like smoking) produce multiple effects.
- Biological gradient: More exposure leads to more of the outcome, a dose-response pattern. If heavier smokers get lung cancer at higher rates than light smokers, that gradient supports a causal link.
- Plausibility: A known biological mechanism can explain how the exposure could produce the outcome. This one comes with a caveat: plausibility depends on current scientific knowledge, which changes over time. Something that seems implausible today might become well understood tomorrow.
- Coherence: The causal interpretation doesn’t conflict with what’s already known about the disease or exposure from other branches of science.
- Experiment: Experimental evidence (like a randomized controlled trial) supports the relationship.
- Analogy: A similar exposure is already known to cause a similar outcome, making the new claim more believable by comparison.
Why Correlation Alone Doesn’t Work
A strong statistical association between two variables is one of the Bradford Hill criteria, but by itself it tells you nothing about causation. Two things can be correlated for reasons that have nothing to do with one causing the other. The most common explanation is confounding: a third variable influences both the exposure and the outcome, creating the illusion of a direct link.
For a variable to be a true confounder, it must meet three conditions. First, it has to be a risk factor for the outcome on its own. Second, it must be unevenly distributed between the groups being compared. Third, it cannot be a consequence of the exposure, meaning it sits outside the causal pathway you’re investigating. Ice cream sales and drowning rates both rise in summer, but the confounder is hot weather, not some dangerous property of ice cream. Without accounting for confounders, even a very strong correlation can be meaningless.
How Experiments Strengthen Causal Claims
Randomized controlled trials are considered the gold standard for establishing cause and effect. The reason is straightforward: when you randomly assign participants to either receive an intervention or not, you balance out all the characteristics between the two groups, both the ones you can measure and the ones you can’t. Any difference in outcomes can then be attributed to the intervention itself rather than to some lurking confounder.
No single study definitively proves causation. But randomization reduces bias in a way that no other study design can match. This is why drug approvals, public health recommendations, and clinical guidelines lean so heavily on trial data. When randomized experiments aren’t ethical or practical (you can’t randomly assign people to smoke for 20 years), researchers rely on the full toolkit of observational evidence, the Bradford Hill criteria, and modern causal inference methods to build the strongest case possible.
The Counterfactual Framework
Modern causal inference often uses a “what if” approach called the counterfactual framework. The core idea is simple: to know whether a treatment caused an outcome for a specific person, you’d need to compare what actually happened with what would have happened if that same person, at that same moment, had not received the treatment. Since you can never observe both scenarios for the same individual, researchers use study designs and statistical methods to estimate that comparison across groups.
This framework makes the requirements for causality very explicit. You need a clearly defined exposure, a clearly defined outcome, a way to account for confounders, and, once again, a confirmed time order. Every modern method for causal inference, from randomized trials to advanced statistical modeling, is essentially an attempt to approximate that impossible counterfactual comparison as closely as real-world data allows.
Putting It All Together
If you’re answering a multiple-choice question about which single factor is required to establish causality, the answer is temporality. The cause must precede the effect. Everything else, including strength of association, consistency across studies, biological plausibility, dose-response relationships, and experimental support, strengthens a causal argument but is not individually required. A causal relationship can exist without a known biological mechanism, without a perfect dose-response curve, and without a large effect size. It cannot exist if the proposed cause happened after the outcome.
In practice, establishing causality is rarely about meeting one criterion. It’s about accumulating evidence across multiple dimensions until the weight of that evidence makes a causal interpretation far more convincing than any alternative explanation. Temporality is simply the foundation that everything else is built on.