Association vs causation

Observational studies are an important tool in science and medicine, allowing researchers to investigate associations between different factors in large groups of people. However, a key limitation is that these studies can only show links, not whether one thing causes another.

What is an observational study?

In an observational study, researchers watch and record information about participants without changing what happens to them. Common examples include studies looking at diet, exercise, or lifestyle habits and comparing health outcomes between people with different exposures.

Association does not mean causation

Observational studies can detect correlations—what tends to happen together—but just because two things are linked does not mean that one causes the other. As an example, people who eat more fruits and vegetables may tend to be healthier, but the study cannot confirm that fruit and vegetable consumption is directly responsible for better health.

The problem of confounding

A confounder is a third factor that influences both the supposed cause and the outcome. For example, people who exercise often may also eat healthier diets, sleep better, and have other healthy habits. These other factors could be responsible for health differences, not just exercise alone.

Reverse causation

Sometimes, it is possible that what looks like an effect is actually a cause. For example, people with early signs of illness might change their behaviour in response, such as eating less or being less active, which can make it look like their behaviour caused the illness when it's actually the other way around.

How do researchers try to address these problems?

• Statistical adjustment: Researchers use statistical methods to account for known confounders, like age, gender, or smoking status. However, they cannot adjust for everything, especially unknown or unmeasured factors.
• Prospective vs. retrospective design: Some studies follow people over time and collect data as events happen (prospective), which can reduce some biases.
• Triangulation with other evidence: Researchers may look for similar associations in different studies or use experiments (like randomised controlled trials) to test whether changing the exposure changes the outcome.

Headlines and overinterpretation

Media coverage of observational studies often treats associations as if they prove cause and effect, which can be misleading. It's important to ask whether a study is observational or experimental and to be cautious about overstating what the findings mean.

Key takeaways

• Observational studies can identify patterns and generate hypotheses but do not prove causation.
• Confounding and reverse causation can distort interpretations.
• Further research, especially randomised controlled trials, is often needed to test causality.

Frequently Asked Questions

• Do observational studies have any value if they can't prove causation?
  Yes, they are valuable for identifying potential links, generating research questions, and informing further studies. They are often the first step before more rigorous testing.
• How can you tell if a study is observational?
  Look for language describing how researchers watched or surveyed people without changing what they did, and whether there was no assignment of interventions by the researchers.
• Are randomised controlled trials always possible to do?
  No, sometimes it is not ethical or practical to assign exposures, such as smoking or specific diets, so observational evidence is all that is available.
• What should I watch for in headlines about scientific studies?
  Be wary if headlines claim that one thing causes another based only on an observational study. Check for mentions of confounding, limitations, or a need for further research.