 This is one of those topics that I get kind of fired up about. So if you think I've been jacked up on some of the other videos, this is one that tends to get me a little even more fired up. Why? Because it's about science. See? I'm already going. All right. Gotta calm down. All right. Take a deep breath. If you haven't read Murray Sidman, you need to. All right? So Murray Sidman, the tactics of scientific research, wickedly dry book, but absolutely awesome. Okay? Especially if you actually understand it. And it's a challenging book to understand, but I think the key that I picked up when reading Sidman's book was the fact that you have a hard science in our field. And that hard science is because of what this video is going to tell you about, right? So it's important to understand these things for our science because these things in here in this little video that we're going to talk about now are what separates the field of behavior analysis, the science of behavior analysis, and the philosophy of applied behavior analysis, or radical behaviorism, or functional contextualism, any of that stuff. This is the stuff that separates us from the other social sciences. This is the stuff that makes us a hard science. There was an argument a long time ago that behavior analysis should never be tied in with psychology because it's not psychology. It's not the study of the psyche. It's the study of behavior, which means it should probably be tied in with biology. Because really, if you think about where behavior analysis is on the chain of sciences, it's an extension of biology. It's applied biology, right? It is not psychology. All right? We're studying the psyche. We're studying your behavior. So it's an accident to paraphrase and possibly even quote Mechakiesa in her wonderful book on radical behaviorism. The fact that behavior analysis is considered an area of psychology is basically an accident of history. And this video will tell you a little bit about why. Maybe the next video will depends on how long winded I get and how much we need to chop this up. So I'm already getting wickedly ahead of myself because I'm excited about this topic even though I may not seem like it. So the first thing we need to understand is that behavior analysis is a hard science. We're a hard science because of our focus on observable events. And because we're not going to make inferences, well, how the hell do we do that? And how do we pull this off in a setting that allows us to draw these conclusions and allows us to still, in other words, allows us to have the experimental control that we have to do the things that we wanted to do and to draw the conclusions that we want to draw. How can we actually do that? Because it's really, really hard, especially if you're borrowing from the social sciences and we don't. Largely speaking, we don't. Sometimes there is some overlap between the other social science methodologies and ours, but not always. And I would argue that when it happens, it may be a mistake. It may not. I'll get into that at another time when we talk about particular types of designs. So I'm going to back up the truck a long ways and go into what single-subject research is about. Okay? So single-subject research, it's another one of those missed numbers, right? Single-subject, small-end design, all sorts of things. Behavior analysis, the fuel of behavior analysis, in our entire philosophy, is founded on the concept, on the package of experimentation that is single-subject research. Now before we go further, we need to make something abundantly clear. Single-subject research is not about one person or two or three. It's not about four. It's not about five. I mean five. It's about all of them at the same time but analyzed independently, okay? Single-subject research is about how you analyze the data, not how many people participate in your studies. It has nothing to do with that. I've done single-subject research. I wish it would have been published by now but it's not largely because I'm just lazy to have it written up but on 40 or 50 people concurrently and I did that for several years. Okay? I've done single-subject analysis on multiple classrooms at the same time. Okay? 30 people being observed, right? Or 30 people in the classroom all being observed analyzing independent data, right? So single-subject analysis is about the independent analysis of the data. Why is that important? It's important because we're trying to predict and control individual behaviors. So we need a level of analysis that focuses on an individual behavior, all right? So if the goal of our field is to develop interventions and blah, blah, blah, and do all these things to affect the individual in a positive way based on the research and blah, blah, all these other things that you can go look up in the videos that I've talked about before that I'm not going to re-hash right now. If it's about that, then we have to have a science that matches it. And the only type of science we have that matches it is a single-subject research. So let me give you a very quick rundown of how this might be. Take for example, the rabbit that you see here, all right? I want you to pretend you're going to take a shot, take a shot, push, it hits up here. You missed the rabbit. All right? Six inches to the eye. Take another shot. Bang! You hit down six inches below the rabbit. Well, if you're a typical psychologist, a typical sociologist, and in all these other fields that you use group design, congratulations, you shot yourself a rabbit. Why? Because the average between six inches high and six inches low is... Okay? And you're going to go home with food? Really? Are you sure? No. You're not. All right? You missed the rabbit twice. So the single-subject analysis says, we're going to look at where bolt number A went. We're going to look at where bolt number B went, okay? So in our research, we're going to study each individual piece. We are going to look at a level of analysis that does not involve aggregating your data. When you aggregate your data, you lose detail. Aggregating your data means compiling it, putting it into a package, and then coming up with a number about that package. Oh, congratulations. There's an average of 2.2 children per family in the United States. What the hell is a .2 child? I don't know. I'm sure we could come up with something completely vulgar and inappropriate and all sorts of fun stuff to represent what a .2 child is, but I'm not going to do that here. I'm going to let you do that on your own in your head right now. And while you do that, go ahead and pause. And you can have those thoughts that are just wrong to have that I'm having right now. Anyway, but you can't tell because single-subject analysis, you can't observe this, but I can. Ha, ha, ha. Private events. Anyway, getting way ahead of myself or behind myself, depending on which order you're watching the video is in. Anyway, so let's get back to the levels of analysis in aggregating data. So we have a data aggregated and we take a mean, the average, the arithmetic average, the mean of that data set there that I'm looking at here, that average doesn't necessarily or doesn't have to represent an individual. Watch this. If the mean is in between my two hands here, we close off this data and we notice that there's not a soul in between. The mean is the line right in between my hands here, right? This is a group of people and this is a group of people. Everybody fell at the mean. Subsequently, the mean doesn't represent any one person. That's a problem. When you're trying to study the behavior of a person and when you're trying to predict the behavior of a person, you can't do it using means of large groups of people. There is value in doing large group research like that. It is talking about the average behavior of populations using things like central limit theorem and all that stuff to predict what might happen on a grand scale. And yes, even physics uses that to some degree, but physics also uses a single subject analysis under other conditions, just like we're going to use. So again, to summarize, we're not going to be worrying about these groups of people. We're going to be worrying about the individuals within there. So we're not going to aggregate our data. We're going to get a very granular level of analysis and molar level of analysis of all these people. On average, they're 23 years old or on average, they engaged in this much drinking. Now, I don't care how much the average drinking is. I care about how much you drink this weekend because that we can change. If you want to increase your drinking water, decrease your drinking alcohol, you get the idea, right? We need to know how much you did, not how much the average did. So our single subject analysis is going to be about you, about your behavior. We're going to track it. We're going to measure it. Now, it's really funny because I said we don't aggregate, but sometimes we do. We're going to get rates and we're going to do things like that. But we're literally going to be looking at your behavior one at a time. I might plot five on a graph at once. Why don't I plot 10? Why not 20? Why not 30? Why not 40? Only because the graphs become unreadable, not because you can't do it. But as long as I'm plotting one line per individual or per organism, OK? So I'm drawing an extinction curve here. It doesn't look the same for every organism. It's different based on their learning history, right? But what it does do over time is develop this really clear pattern that you can draw conclusions from. So I'm going to come back in another video because I'm getting totally way ahead of myself and I'm going to talk to you next time about external validity and internal validity using single subject research. So come back to that one. It's really cool, really exciting, totally surprising. Jokes on you. You made it to the end of the video. Like, subscribe, share. Bye.