 I think we'll start this out with the demo. We've got the Hilbert chain, as you're going to have eight hands. You grab by the pumpkin fuller. We might be wrong. We're funny, but not always a joke. Back because research methods is awesome, and you wanted to learn more about single-subject analysis, and you already probably watched the other video, and if you haven't watched the first video on single-subject research, you need to go back and watch that one before you watch this one, because this one won't make a lick. It won't make sense, but you'll be missing a whole bunch of context, and you'll be really informed about something you know nothing about. So you need to go watch the other video first, so you know something about what I'm ready to tell you about, which will make it all make more sense. All right, so single-subject research, the studying of one individual at a time. You can study them multiple, but analyzing them one at a time. We're going to limit the aggregation of data. We're not going to look at averages. We're not going to look at medians, for the most part. There's some times when that stuff is useful, do not get me wrong, but when we're talking about the prediction and control and the study of behavior in a nutshell, we're going to use a single-subject approach. So single-subject analysis, again, it's not about the size of the sample. It's about what you do with the data that you get from your studies. And I'm going to continue forward with some talking about internal and external validity. Internal validity, the ability of your experiment to detect an effect if one is there. Say that again. Internal validity is the ability of your experiment to detect an effect of the independent variable if an effect exists. A study that is designed in a piss-poor fashion won't detect the effect of a variable. You'll implement whatever intervention it is, but your design will be so crappy that you won't be able to detect it. That is low internal validity. We're going to study internal validity with some other videos and I'm going to talk pretty heavily about single-subject logic, reversal designs, multiple baseline designs, alternating treatment designs, the Abbott design, which that's a new one, but we'll come back to that one. So all of those types of things we're going to talk about in some other videos, but this we're going to focus on how single-subject research works with both internal and external validity. So internal validity, again, we're looking at behavior change and how that syncs up with condition changes, high levels of internal validity using the designs that we're going to talk about in the other videos. That's it. That's all I'm going to talk about here. But external validity is really where people start to get torquey with single-subject research. Why? Because they don't understand it. They miss the point. And because science is a very complex thing, sometimes you have to pay attention to the nuances in order to understand the big picture. Here's one of those nuances. In fact, the rest of this video is that nuance. So external validity is the ability of your findings to generalize beyond your sample, or to be valid in the real world. There's all sorts of different types of external validity. There's varying levels of it. I'm not going to get into all the different types because I'll just talk myself into a circle for the next hour, and you don't have time for that, neither do I. So group research studies large numbers of people. We often use a fancy rule called n equals 20k. Your sample size is determined by the number of conditions. k's multiply that by 20. You got yourself your sample size. You got a couple of conditions, let's say two. We're going to need about 40 people to run a genuine group-sized experiment on that, to have enough power, to have the internal validity to detect the effect that's there. If it is there, single-subject research says, now we can do that with one person, maybe two. Heck, you know what? Just grab three or four, maybe five, because that's convenient. Okay? So we're going to work with those five, we're going to do the same thing. Like how? How? Well, that's in the other videos, but the important point that I'm trying to make is that when we use these large groups of people, we often talk about representative samples. So in general and in large group research, we talk about make sure you get a representative sample, and there's amazing amounts of procedures to pull that off. And that's awesome, and that's really good about how you do group design, and there's this wonderful logic and all this stuff, and all these different things that fit together, and then statistics feeds into all of that stuff, and someday we're going to do some series on statistics, so hold your horses for about a year, and then we'll get to that. So anyway, so all the statistics, all the stuff allows us to analyze this and apply it to the population as a whole. In other words, external validity. With group research, with large sample research, you're talking about using the way in which you select your participants to address issues about external validity. There's other ones too that have to do with the experiment, but that's a big part of it. So then how in the world can a single subject research, how in the world can a single subject study even have internal external validity? How can you do it with a subject, with one subject or two or three? I'll tell ya, this magical little thing called replication. That's the key, replication, right? Any finding in science needs to be replicated in order for it to be valid. Some of you may remember a study that came out not too long ago that talked about how awful psychological research is as a whole. How unreplicable a vast majority of the findings are. I'll try to find the research and put it down in the comments. That's largely what I'm talking about. If you have drastically high levels of internal validity, if we're way up here with internal validity, single subject research designs, we'll get there, then when we get a finding and we try to replicate it and we are successful, we have some pretty powerful science. So there's different types of replication. There is systematic replication. There is direct replication. There's inter-subject replication and inter-subject replication. When we're doing a study, we get replication across, maybe we do five people at once or five animals or whatever it is. We're using single subject analysis. We see the same patterns for one, two, three, four, five. Not identical, but similar patterns. There you go. There's one type of replication for you. Then we try it with some slightly different subjects. Again, maybe next week, do the same study over again. Two, three, four, five. I know I'm drawing extension curves, but they're pretty easy. So he did this again. So now we have more research. Now let's try it on a different species. Five more. One, two, three, four, five. In a different species and so on and so forth and under different conditions and with different behaviors and with different contexts and in different environments and all of this other stuff and all of a sudden you find out that the study of one individual actually generalizes to others. So if you establish using replication, the fact that you have a sound finding, you will have higher levels of external validity than you do using group designs. This is the sticking point that I personally have with large group research. It doesn't do any good in predicting the behavior of an individual. So as some other behavior analysts have once said if you really want external validity only study one person at a time because then you're going to get valid information about that one person and you might be able to predict the behavior of another person rather than weird ephemeral means that don't really physically exist in the real world anywhere. Bay of analysis folks is a hard science as hard as physics, as hard as biology, as hard as chemistry. It's not a soft science at all which is why I personally and many other people don't think it belongs inside of psychology departments at all. But alas it does, it's there. So we have to deal with it. So the logic is drastically different. Remember external validity comes from getting a finding with one person with a very solid finding using strong high levels of internal validity testing that with multiple individuals and then doing more types of replication with other individuals, other similar experiments and so on and so forth. We'll come back to another video talking about the different layers of replication with systematic direct and all that fun stuff. So I hope that's enough to get you going on single subject research. We're going to come back and do more on experimental design later. If you thought that was pretty cool maybe you should go check out the rest of the playlist on the RBT issues.