I really like this example because ultimately it's emphasizing what treatment represents. Again, treatment variable is emphasizing what the subject gets. We literally think of drugs as a treatment, and so the treatment variable is whether or not this Alzheimer patient gets the drug. Now, in any scientific study, we want to see if the drug will work. We are talking about Alzheimer patients who have issues with memory, so the outcome would be: how did their memory do? How did they do on the memory test? That is the outcome variable. Now, remember, the treatment variable can be broken into two parts: those who said yes, they got the drug, and those who said no, I did not get the drug. And it's that yes/no that determines the treatment group versus the control group. Again, when you think of the word 'treatment,' think about literally what are they being given. The thing that's of interest, like this new drug, are those who get to go in the treatment group, and those who don't get the drug are those who go in the control group. Can you tell me what was the key word here that told me 'controlled experiment'? Yeah, "given" - patients were given either the drug or the placebo. In this class, most of what we're going to do here is going to be looking at controlled experiments. And the reason for that is because there does need to be a certain degree of control. And so, because of that, I do want to discuss then a little bit more about controlled experiments. Like we just saw in this example, what made a controlled experiment was one thing and one thing only, and it was that the researchers assigned and told the patients what to do. See, the only thing that makes an experiment is when the researchers assign the subject what to do. That is the only thing that makes an experiment an experiment. And so, the big question is then: what makes a good experiment? Alright, what makes a good experiment? What do I mean by this? Well, it's like having a dog. We know what dogs are: furry animals, four legs, tail, snout, cute little happy demeanor. And yet, the question is, what makes a good dog? A good dog is one that doesn't bark. A good dog is one that doesn't pee in the house. A good dog is one that doesn't jump all over you. You see, whether or not a dog pees in the house, it's still a dog. Whether or not a dog barks, it's still a dog. So, in the same way, experiments are simply when a researcher assigns a subject what to do. But I want to emphasize what makes a good experiment. And we have this thing called the gold standard. The first gold standard for an experiment is that you have a large sample size. Let's go back to our Alzheimer example and see, we looked at 550 plus 450 people. I want you to see here that this sample size was ultimately a thousand people. A thousand is incredibly large. Why is this? A thousand people being sampled so important? Well, see, large sample sizes account for variability. By looking at a large sample, it means that we're looking at all types of people: men and women, all types of ages, all types of ethnicities, all types of socioeconomic statuses, all types of health conditions. See, when you have a large sample like a thousand people, it increases your chances of getting a group that represents all people. What else was another gold standard? Another gold standard is when the subjects were randomly selected. And randomness, I want to emphasize, is what makes a good experiment but doesn't necessarily make it an experiment. But why? Why is it so important that the patients were randomly divided into the two groups? Well, it's because randomness ultimately minimizes bias. Randomness makes sure we have a good spread of all people in both groups, the treatment and control group. Both have men and women, both have people of all ages, have people of all ethnicities, have people of all health backgrounds. See, the idea of randomness not only makes sure that we have all of the people but that all of the people are in both groups, minimizing bias. What's another gold standard? Another gold standard is having a placebo. So, as we saw in this example, some Alzheimer patients were given that new drug, that new pill. Others were just given a sugar pill. The question is, why? Why do we even want a placebo? Well, so we can make a blind study where participants won't know if they were given that treatment and therefore cannot expect the treatment to work. See, that is then the fourth of the gold standards: you want to have a blind or double-blind study where a double-blind study has it that not only do the subjects not know if they got the drug or not, the drug, are in the treatment group or in control group, but more so the researchers don't know who went into each group. And so, the question is: how do you even do that? Well, a lot of times researchers will hire a third party to conduct the experiment so that third party will give the treatment or placebo to the control and treatment group, and those people won't know what they got. That third party will then collect all that data, erase all of the names so you don't know who you're looking at, and then give that data to the researcher. Now, at the end of the day, it's not always possible to have a placebo, it's not always possible to have a blind or double-blind study, but the two things you can always have to make a good experiment is a large sample and a random sample. You can always, always, always have an experiment where you just talk to more people, gather more data, and you make sure you talk to people randomly. And so, ultimately, large sample and random sample is something we will utilize heavily in phase two of statistics, and so we're introducing it here. We're going to come back to it a little bit later.