These two terms are always interchanged especially in the fields of health and scientific studies. Every time we see a link between an event or action with another, what comes to mind is that the event or action has caused the other.
What immediately caught my eye was those figures in the 90s — clearly, studying abroad makes you irresistible to grad schools and employers. I was surprised at just how large the academic and career benefits were that came as a result of studying in another country.
My second thought was: However, something about the information stuck with me. A few days later while listening to a data skeptic podcast it hit me: I had assumed that studying abroad caused students to have better grades and career prospects, when all the statistics showed was that the two were correlated.
Sometimes, especially with health, these tend towards the unbelievable like a Guardian headline claiming a diet of fish leads to less violence. The real explanation is usually much less exciting.
For example, students who take music lessons may perform better in school, but they are also more likely to have grown up in an environment with a large emphasis on education and the resources needed to succeed academically.
These students would therefore have higher school achievement with or without the music lessons.
Taking music classes and school performance happen to rise in tandem because they are both products of a similar background, but one does not necessarily cause the other. Likewise, people who stay in school longer typically have more resources which also means they can afford better health care.
Most of the time these mistakes are not made out of a deliberate effort to deceive although that does occur but out of an honest misunderstanding of the idea of causation.
What the statistics, especially those in the study abroad email, show is a selection bias. In each study, the individuals Causation and correlation do not come from a representative slice of society, but instead are all drawn from similar groups, leading to a skewed result.
While it might be possible that studying abroad did somehow motivate lagging students to graduate on time, the more likely explanation is that students who choose to go abroad were those in a better position academically in the first place.
They would graduate on time with high GPAs regardless of whether they went to another country. It takes a lot of work and preparation to go to another country to study for a year, and the students who feel confident enough to do so are the ones who are on top of their studies.
In this real-world case, the selection bias is towards better students. The sample of students who study abroad is not indicative of students as a whole, rather, it includes only the best-prepared students and therefore it is no surprise that this group has significantly better academic and career outcomes.
The study abroad experience may look great in hindsight, but if we selected only the best students and had them do anything, it would be misleading to say the phenomenon led to better grades. Say for example we own a bottled water company and we want to gather some positive stats to help with sales.
We hire a few students to stand outside the honors class and only give our water to the top students. We then conduct a study that shows conclusively that students who drink our brand get better grades. Because we selected a specific group of subjects to include in our study, we can make it look as though our water caused an increase in grades.
The study abroad statistics come from what is known as an observational study. Rather than constructing an experiment, an observational study observes some process in the real world with no cannot control over the independent variable, in this case the students who chose to study abroad.
Observational studies cannot prove cause and effectonly associations between different factors such as achievement and studying in another country.
In order to prove one process caused another requires a randomized controlled trial with subjects represent the entire population. In this case, carrying out a randomized controlled trial would require selecting a random subset of students across the range of academic performance, sending some to study abroad, and keeping a control group home.
We could then analyze the results to determine if there were significant differences between the two groups. If there was, then we would probably carry out more studies controlling for more variables, until eventually we were satisfied there was no hidden effects and we could establish a causal relationship.
Civility and Honest Discussion. By posting about this, I was not trying to call out the office. Although the email did state: However, when a single topic is presented surrounded by a sea of facts, our natural inclination is to draw a causal link, a tendency marketers and companies take advantage of with regularity.
I believe all the statistics in this case are valid, but we still need to avoid assigning a cause and effect relationship. Without randomized controlled trials, we cannot say one activity caused another, and all we can claim is that two trends are correlated.
This is a small example, but it illustrates an extremely critical point: However, the world usually does not have defined causes and effects, and we must settle for correlations.
This view of the world may make headlines less exciting it turns out chocolate is not a miraculous foodbut it means you will not be fooled into buying products or taking actions that are not in your best interest due to questionable evidence.
As always, I welcome constructive criticism and feedback.UNDERSTANDING RESEARCH: What do we actually mean by research and how does it help inform our understanding of things?
Today we look at the dangers of making a link between unrelated results. Here. Describes and gives examples of fallacies of causation.
Conflating correlation with causation is one of the most common errors in health and science reporting. This work is licensed under a Creative Commons Attribution-NonCommercial License. This means you're free to copy and share these .
For two variables, a statistical correlation is measured by the use of a Correlation Coefficient, represented by the symbol (r), which is a single number that describes the degree of . — Dagmar Herzog, New York Times, "Surviving Europe’s Horrendous 20th Century," 6 Apr. While this is a correlation rather than a causation, another study found that taking a substance called haloperidol, which disrupts dopamine’s effects within the brain, led to a measurable decrease in sensation seeking behaviour.