And so, ultimately, what we just did here, in blue and in pink, is discuss the two types of data collection methods. A researcher can do an observational study where the subject chooses what to do. Alternatively, a researcher can conduct a controlled experiment where a subject is assigned what to do. Let's take a closer look now, alright? In the next example, I gave two case studies. I want us to read through them and ultimately decide: what's the data collection method? Why did I pick it? What relationship, association, or causation exists? So, let's do this together. A study took a random sample of adults and asked them about their bedtime habits. The data showed that people who drank a cup of tea before bedtime were more likely to go to sleep earlier than those who didn't drink tea. So, what's the treatment here? The treatment is whether or not a person drank tea before bedtime. That is the treatment variable. Whether or not a person drank tea before bed. And the outcome we're studying is their bedtime habits. That's the outcome variable. Does drinking tea make you go to sleep earlier or later? So, the researcher did their thing. My question is, how did the researcher collect this data? Did they do an observational study or did they do a controlled experiment? Right, this is an observational study. There's one key word here that's telling us this is an observational study. It's 'asked'. See, the key word here, 'asked', emphasizes the fact that these subjects got to choose. They got to choose, 'Am I going to drink tea or not?' The researchers didn't tell them, 'You're gonna drink tea.' It just asked them, 'Hey, what did you choose to do last night? To drink or not drink tea?' Observational studies come from when subjects choose what to do. So, what is then the relationship between this variable of drinking tea before bedtime and what time you go to sleep? Is this an association or causation? Look back on the previous page. Which did we say observational studies will end with? And in this case, it's association. In general, when we're working with observational studies, it will always lead to an association. I have you guys do this for the study in 'B' as well. Another study took a group of adults and randomly divided them into two groups. One group was told to drink tea every night for a week while the other group was told not to drink tea that week. Researchers then compared when each group fell asleep. I want you guys to see, it's the same variable. It's still the same variable of 'to drink tea or not to drink tea'. I want you to see, it's the same outcome variable of what time they're ultimately going to fall asleep. And so, what I'm trying to emphasize here is that this is a very, very similar case study. This is a similar case study in that we're looking at the same outcome variables, the same treatment variables. And yet, how this case study, how this data was collected was different. I'm going to give you guys a quick moment to do each of the three: what was the data collection method? What was the justification? What is the relationship? And we'll come and talk about it together in a second. What word in the prompt is telling us? Yeah, this is a controlled experiment. What word? Yeah, it's 'told'. See, that word 'told' is emphasizing the subjects were assigned what to do. And so, now that we know we're looking at a controlled experiment, the researchers are ultimately going to collect their data. And what type of relationship can they then establish between the treatment and the outcome variable? Drinking tea and bedtime. Association or causation? Causation, exactly. Controlled experiments will always be associated with causation. What we just did in examples A and B establishes two famous trios, just like Ron, Harry, and Hermione. Ultimately, we are going to see that observational studies will always come when the subjects get to choose what to do, and that the ending relationship will always be a milder association. Whereas the other famous trio is going to be when we're doing a controlled experiment, when subjects are assigned what to do, meaning whatever results we get can establish a causation. A question I'm always asked is, 'Oh, Shannon, association's weaker, causation's stronger. So why don't we always push forward for controlled experiments?' Well, many studies have shown that children of mothers who smoke while pregnant tend to be less well-developed physically and intellectually than children of mothers who do not smoke while pregnant. And so, given this particular study, what do you guys think it is? Observational or experimental? Yeah, like everyone is saying, observational. Totally, totally. And this is an important study, right? This is an important study because it's about protecting the health of children. So, someone might beg the question of, 'Why don't we conduct a controlled experiment? Why can't we do a controlled experiment here? What is the issue about making this control?' Yeah, it's unethical. It might hurt someone in the midst of it. You guys know I have two kids. If a scientist came up to me and said, 'For the sake of science, smoke while you're pregnant,' I would have punched that guy hard in the face because there's no way I would want to put myself or my child at harm's risk. And while this controlled experiment will lead to a strong relationship of causing, smoking will hurt children, not might hurt children, but will hurt children, while the controlled experiment is feasible, the issue with it is that it is not ethical. Ultimately, that is the reason why observational studies, while giving a weaker result, are sometimes more powerful than controlled experiments because of the ethics behind forcing people to do something. And that's why you need both. That is the reason why we need both observational studies and controlled experiments.