To get a complete picture of how and why something happened

In this case, we will want to look for multiple causes, each of which may play a different role. Some might be background conditions, others might spark the event, and others may be influences that sped up the event once it got started. In this case, we often speak of near causes that are close in time or space to the event itself, and remote causes, that are further away or further in the past. We can also describe a chain of causes, with one thing leading to the next, which leads to the next. It may even be the case that we have a feedback loop where a first event causes a second event and the second event triggers more of the first, creating an endless circle of causation. For example, as sea ice melts in the arctic, the dark water absorbs more heat, which warms it further, which melts more ice, which makes the water absorb more heat, etc. If the results are bad, this is called a vicious circle.

To decide who is responsible

Sometimes if an event has multiple causes, we may be most concerned with deciding who bears responsibility and how much. In a car accident, the driver might bear responsibility and the car manufacturer might bear some as well. We will have to argue that the responsible party caused the event but we will also have to show that there was a moral obligation not to do what the party did. That implies some degree of choice and knowledge of possible consequences. If the driver was following all good driving regulations and triggered an explosion by activating the turn signal, clearly the driver cannot be held responsible.

In order to determine that someone is responsible, there must be a clearly defined domain of responsibility for that person or entity. To convince readers that a certain party is responsible, readers have to agree on what the expectations for that party in their particular role are. For example, if a patient misreads the directions for taking a drug and accidentally overdoses, does the drug manufacturer bear any responsibility? What about the pharmacist? To decide that, we need to agree on how much responsibility the manufacturer has for making the directions foolproof and how much the pharmacist has for making sure the patient understands them. Sometimes a person can be held responsible for something they didn’t do if the action omitted fell under their domain of responsibility.

To figure out how to make something happen

In this case we need to zero in on a factor or factors that will push the event forward. Such a factor is sometimes called a precipitating cause. The success of this push will depend on circumstances being right for it, so we will likely also need to describe the conditions that have to be in place for the precipitating cause to actually precipitate the event. If there are likely factors that could block the event, we need to show that those can be eliminated. For example, if we propose a particular surgery to fix a heart problem, we will also need to show that the patient can get to a hospital that performs the surgery and get an appointment. We will certainly need to show that the patient is likely to tolerate the surgery.

To stop something from happening

In this case, we do not need to describe all possible causes. We want to find a factor that is so necessary to the bad result that if we get rid of that factor, the result cannot occur. Then if we eliminate that factor, we can block the bad result. If we cannot find a single such factor, we may at least be able to find one that will make the bad result less likely. For example, to reduce wildfire risk in California, we cannot get rid of all fire whatsoever, but we can repair power lines and aging gas and electric infrastructure to reduce the risk that defects in this system will spark a fire. Or we could try to reduce the damage fires cause by focusing on clearing underbrush.

To predict what might happen in future

As Jeanne Fahnestock and Marie Secor put it in A Rhetoric of Argument, “When you argue for a prediction, you try to convince your reader that all the causes needed to bring about an event are in place or will fall into place.” You also may need to show that nothing will intervene to block the event from happening. One common way to support a prediction is by comparing it to a past event that has already played out. For example, we might argue that humans have survived natural disasters in the past, so we will survive the effects of climate change as well. As Fahnestock and Secor point out, however, “the argument is only as good as the analogy, which sometimes must itself be supported.” How comparable are the disasters of the past to the likely effects of climate change? The argument would need to describe both past and possible future events and convince us that they are similar in severity.

Identify possible causes

If other writers have already identified possible causes, an argument simply needs to refer back to those and add in any that have been missed. If not, the writer can put themselves in the role of detective and imagine what might have caused the event.

Determine which factor is most correlated with the event

If we think that a factor may commonly cause an event, the first question to ask is whether they go together. If we are looking for a sole cause, we can ask if the factor is always there when the event happens and always absent when the event doesn’t happen. Do the factor and the event follow the same trends? The following methods of arguing for causality were developed by philosopher John Stuart Mill, and are often referred to as “Mill’s methods.”

• If the event is repeated and every time it happens, a common factor is present, that common factor may be the cause.
• If there is a single difference between cases where the event takes place and cases where it doesn’t.
• If an event and a possible cause are repeated over and over and they happen to varying degrees, we can check whether they always increase and decrease together. This is often best done with a graph so we can visually check whether the lines follow the same pattern.
• Finally, ruling out other possible causes can support a case that the one remaining possible cause did in fact operate.

Explain how that factor could have caused the event

In order to believe that one thing caused another, we usually need to have some idea of how the first thing could cause the second. If we cannot imagine how one would cause another, why should we find it plausible? Any argument about agency, or the way in which one thing caused another, depends on assumptions about what makes things happen. If we are talking about human behavior, then we are looking for motivation: love, hate, envy, greed, desire for power, etc. If we are talking about a physical event, then we need to look at physical forces. Scientists have dedicated much research to establishing how carbon dioxide in the atmosphere could effectively trap heat and warm the planet.

If there is enough other evidence to show that one thing caused another but the way it happened is still unknown, the argument can note that and perhaps point toward further studies that would establish the mechanism. The writer may want to qualify their argument with “may” or “might” or “seems to indicate,” if they cannot explain how the supposed cause led to the effect.

Eliminate alternate explanations

The catchphrase “correlation is not causation” can help us to remember the dangers of the methods above. It’s usually easy to show that two things happen at the same time or in the same pattern, but hard to show that one actually causes another. Correlation can be a good reason to investigate whether something is the cause, and it can provide some evidence of causality, but it is not proof. Sometimes two unrelated things may be correlated, like the number of women in Congress and the price of milk. We can imagine that both might follow an upward trend, one because of the increasing equality of women in society and the other because of inflation. Describing a plausible agency, or way in which one thing led to another, can help show that the correlation is not random. If we find a strong correlation, we can imagine various causal arguments that would explain it and argue that the one we support has the most plausible agency.

Sometimes things vary together because there is a common cause that affects both of them. An argument can explore possible third factors that may have led to both events. For example, students who go to elite colleges tend to make more money than students who go to less elite colleges. Did the elite colleges make the difference? Or are both the college choice and the later earnings due to a third cause, such as family connections? In his book Food Rules: An Eater’s Manual, journalist Michael Pollan assesses studies on the effects of supplements like multivitamins and concludes that people who take supplements are also those who have better diet and exercise habits, and that the supplements themselves have no effect on health. He advises, “Be the kind of person who takes supplements — then skip the supplements.”

If we have two phenomena that are correlated and happen at the same time, it’s worth considering whether the second phenomenon could actually have caused the first rather than the other way around. For example, if we find that gun violence and violence within video games are both on the rise, we shouldn’t leap to blame video games for the increase in shootings. It may be that people who play video games are being influenced by violence in the games and becoming more likely to go out and shoot people in real life. But could it also be that as gun violence increases in society for other reasons, such violence is a bigger part of people’s consciousness, leading video game makers and gamers to incorporate more violence in their games? It might be that causality operates in both directions, creating a feedback loop as we discussed above.

Proving causality is tricky, and often even rigorous academic studies can do little more than suggest that causality is probable or possible. There are a host of laboratory and statistical methods for testing causality. The gold standard for an experiment to determine a cause is a double-blind, randomized control trial in which there are two groups of people randomly assigned. One group gets the drug being studied and one group gets the placebo, but neither the participants nor the researchers know which is which. This kind of study eliminates the effect of unconscious suggestion, but it is often not possible for ethical and logistical reasons.

The ins and outs of causal arguments are worth studying in a statistics course or a philosophy course, but even without such a course we can do a better job of assessing causes if we develop the habit of looking for alternate explanations.