. The last experimental. So this is going to involve the event relationship to between more variables. And do much changing on manipulating one of the variables theries as you already talked about both designs. And then we record or collect data, what obser the change in the dependent variable that result from our manipulation of the. That's what we're looking at. We're moving and sh and manipulating one and seeing if it causes an effects or change in the other. That's what we're looking for. So have experimental research, we are looking for causation not just correlation. We're not just looking to see due to variables moved together. No, we're actually looking to see if we make a change in one variable, do we see a subsequent change in the other word? If we make another change in that variable, we shift it more, we change it fast. We take it away. Do we need a consequential change in the independent variableag yet, okay, you're able to shift and manipulate the independent variable and consistently see a change of the dependent variables, then you know you have causation, a change in one causes a change in the other. They've already kind of gone over this multiple things, so I will just briefly say this again, but you've got the independent variable dependent variable, the independent ones you what we're going to manipulate and change, whatever. that looks like. um at a very simple level of experimental research, you can have one level of your independent variable, and then nothing, right? You can have your experimental group and your controller. The group that gets the treatment, that group that does not. So that is your very basic experimental research where you just have two groups and one of them is to control groups. But even still, you should see a change in the depependent variable to the group that is receiving treatment and you should see no change for the group that is not receiving treatment, right? That would be causation. Now, of course, you can have multiple levels of the independent variable, we're not gonna get too much into that. In this course, um, but two is kind of the minimal, right? treatment, no truth, and then you can move beyond that. The dependent variable is the one that is being measured. It is hopefully changing. If you see no change in the dependent variable when you're making changes to the independent variable, you've got a big problem, right? That means that your independent variable that you manipulating actually has nothing to do with the behavior that you're trying to observe. It doesn't impact it at all, and you're going to have no results, no adjacent. It's very disappointing. It does happen. and it's disappointing, but not happen. Um, so it is the uh the outcome orependent measure. Now, something I briefly mentioned that I have gone too much into depth, yet are the confounding variables, so the confounding or they're also called you probably heard them called extraneous variables. These are other variables, other than your independent variables. So anything that is not your independent variable can be a confounding variable, and it can cause and change in the dependent variable if you have not accounted for and controls something that has to be at and avoid it at all costs. Let's say let's say we're doing a study and we are trying to decrease the amount of smoking individuals engage. Hi, so we're trying to help them. We're trying to decrease their smoking paper. And our treatment is going to be some sort of meditation and relaxation techniques that they can learn because of that is based on the research that people smoke war when they're experiencing higher levels of stress. So how can we decrease their stress? Let's teach them various coping mechanisms, deb breathing techniques, meditation techniques, other things that they can do to decrease their stress and hopefully have a decrease in theopy behavior. Okay, great. So we implement our treatment. But what if we forgot to ask participants? if any of them had gotten pug onto to the doctor recently and had some maybe vac about their health, if they received some not so great news about their health, could that be a variable that is intacting how much they decide to smoke after that document? Absute, right? The doctors that said, hey, you' lungs are not looking for good, or you've got something precursors to cancer, we're gonna have to run some tests. That type of news could certainly impact someone who's smoking and could result in a change in their smoking behavior, they might leave that doctor's office and go, okay, wow, I really need to stop smoking. But if we didn't ask them that, we don't know. We don't have that information. So, we've moved forward, we implement our procedure and our treatment, and theyreased their smoking and we go, wow, our treatment works really great. Look at all these people that stop smoking. But in fact, all those people went to the doctor got not so great, there was a decided to not smoke, regardless of whether or not you taught them had a meditate break, right? That is a confounding variable that will throw your data because you did not account for it. Whenever we're doing a study like that, on any type of addictive behaviors for illness, if you're doing a medication study, you have to ask all of those questions. You have to get all of that information up front, because those are come down in variables that can change the behavior that you did not account for and you are not manipulating or control. So now we can't make the claim that if we um, you know, give individuals, um different mechanisms to decrease their stress, it will decrease their snow people. We can't make that claim anymore because that's not what caused the meaning. Or at least we don't know for sure that that's what we're doing. So confoundingles are a big bump. we run into these a lot, and I will tell you that when we are designing a research study um when you're working in a lab and you're working with researchers, it is intimidating to bring a research project to the lab. I mean, I did it a lot inad school. We were required to do this. You have to do this when you're doing research, but you bring your research question and your proposal for how you're gonna run your study to the lab. you put it up there and literally everyone in the room writs it apart. Everyone sits there for an hour or two and says, what about this confounding birdle? What about this? Well, this one's gonna throw your data. Well, this one's not gonna work. Well, you have an accountant for this, they rip it apart. It doesn't feel great in the moment. However, that is how you identify all of the compounding variables and you find a way to account. so that you have good data in the end. It's very important piece of research and experimental research specifically. We do want to avoid them at all costs. Okay, so here's another example. Let's say a researcher investigate whether giving students more time to study, reduces their tests anxiety. Okay. What is going to be the dependent variable here? What are we measuring? What are we looking at? We wouldn't want to take it again. Test anxiety. levels of anxiety when you're taking a test, right? That's what we're measure. We're trying to change that, okay? So that's gonna be the behavior that we're looking at. What is the independent variable here? Time to set, the amount of time that you're set, whatever that may be, okay? So the DV is test anxiety or levels of anxiety will take the test whatever you will word that, that's what we're measuring. Am amount of study time is what we're looking at for the independent version. Now, when you're taking a test, there are multiple things that happen that have nothing to do, maybe, with the amount of time you study. Can we reduce test anxiety by making sure that you study at least a minimum amount of time? Yes, we can reduce your test anxiety a little bit. But there are also other factors that if we if I was running this study as an experimental research, not just as like the naturalistic observation in a classroom, like let's just see if we can help. If I was actually running an experimental res research that many things that I have to account for. I need to account for type of tests. What if I get half of my participants, the tests is the morning and half of my participants the test in the afternoon? That's the I founding variable. Maybe the students in the morning are more stressed out because they didn't have time to relax in the morning and get ready for this test that I'm about to do them yet. Right? They're getting ready, they're in traffic, they're driving here, trying to park and so and so forth. Yes, we might run into that in the afternoon, but you still have got more time in the day. to get ready for it. So that's a confounding marriage, time of test. Another confounding variable would be temperature in the room. If it's too cold or too hot, you've got one room that's hotter, one room that's colder. That can impact someone's test anxiety. When you're feeling anxious, if I'm sure everyone has felt that feeling at one point in their life, it doesn't feel great to then also be hot and sweat. It usually makes that anxiety a little bit worse. You start to feel kind ofustrophobic and you're like, I don't know what's going on. I'm getting really hot. I don't feel good, I'm getting kind of dizzy, like, and your anxiety skyrock. right? So, I wanna make sure that the temperature in my room every time participants are taking the test, it has to be exactly the same, or usually within a couple degrees of the temperature. Okay, so these are just a few examples I can go on and on about all of the things that would impact you while you're having an exam that would impact your test anxiety. I need to help for all of those things, and every participant in all of my different groups would all have to have the same things so that I can truly say it was the amount of set. and it wasn't possibly due to during the room, time of the test, the room that they're in, how close they're sitting to each other and so on. and and that and that's part of the roofing unit ofart process, right? If I came to my lab and just said, oh, I'm gonna do this. They're like, well, what else, what else are you controlling for? I'm like, nothing, you know,'ll be fine. They're gonna rivet apart, right? All of those confounding variables that we need to account for. Um, a study involved investigating how manipulating the accuracy with which feedback is delivered, affects a number of work tasks that can be completed by college. So this is essentially, say, a student is doing a work task, and if I give you no feedback on that, as to whether you're doing it track, if I give you feedback that is correct and it matches, I say, yes, that's correct. or if I give you wrong feedbacks. So that's what we're talking about when we were saying a different type of feedback on your ability to complete a task. So, what is the independent variable here? What are we manipulating? Yeah, that's hypo feedback, right? We're gonna change that. It's gonna be different. What's the dependent variable that we're measured? It's the behavior we're looking at here? Yeah. Uh, number of work task. Correct, yes. How many workops did they actually complete? Do they get more done when they're getting positive feedback? Do they get more done when they're getting no feedback at all? You probably don't get more done when they're getting negative. You back would probably be my hypothesis, but we're gonna look at them, right? We're gonna count how many tasks they get done based on the type of feedback that they are given. And then we see how this impact that dependent varies. How does that impact the behavior that they're engaging? So, there will be questions like this on these things. This is like a perfect example. It will be this exact one, I'll change the words I'll change a thing. And I'll ask you these questions. What is the independent? What is the dependent variable? And sometimes I'll put a confounding variable in there and I'll say like, identify the confounding variable. and you'll hopulate pick one of the choices. So very similar to what I've test questions would look like for something like that. But the good to be able to look at examples and pull these things up about. If you're in any type of research class, statistics class, you need to be able to have this very uh, you guys could go over these ones.. I'm not gonna keep going, but you get you get this. All right. So experiments typically involve two groups at a minimum, which I talked about already, but you're gonna have atom minimum, your control group, and your experimental. The control group is a group of participants that does not receive the treatment. No treatment. Okay. Um, you don't change it. You essentially just measure their behavior, but you don't expose them to anything. Um, so work tasks with the feedback, that would be the control group is no feedback, okay? So we just allow the students to complete tasks as they normally would, we do not interject, we do not give the feedback one way or another. We just sort of let them carry on with their day as they normally would and we count how many works how they could. Versus, the experimental group are the ones that are going to receive some type of treatment. Now, as I've said before, minimally, you've got one experiment group and your control but you can have multiple experimental groups and a controll. And so you can have two or three different types of feedback. Those would be your experimental groups, and you can still have a group that received no feedback. if we're looking at study fine with students, you can look at, you know, two hours, four hours, six hours a week. Those are your experimental groups. the other students, you would sort of just allow them to either or you would prevent them from studying at all, or you just would not manipulate the study time for them, you would allow them to study however long they normally do and have them report on. So you would just basically specify that this group did not have a controlled set amount of set, and then they would report on how many hours they affected. versus the other three experimental groups would have a set amount by the time that you're controlled. Okay. Here's a question, a clinical psychologist conducts a study that involves ten people. He thinks he can cure depression by giving his science a particular type of drug. So he prescribes the drugs and finds his 60 days later, all clients show fewer signs of depression, as the psychologist includes he has cured depression. So what's the problem? There's a lot of problems here, but like, what's the main simple problem with what we understand in this particular research research? What has not been done? Yes.... doesn't describe. There is no controller, right? Every single person got the drug. There's no control group. So how do you know that that drug improved their depression? If you do not have a control group, you have no comparison to make, the whole point of having a control group, the whole reason we do it is so that if the drug does work, let's say that the psychologist is correct, this drug works, it cures depression. If you have a controlled group, we have a group of participants who didn't get the dress, what should happen for them? is someone over here? I take a guy? What should happen for people who are naked? Is it control with this? What do you expect? Yes. Yes, they should save the same, right? They're not getting the drugs. So they shouldn't get better. And then the people in the experimental group who are getting the drug if the drug works, they should get better. And you have that comparison. You now you can definitively say, okay, look at all these people that did not get the drug in my control group, they didn't get any better. The symptoms of depression persisted. But look at all of my participants, my experimental your, we saw a significant improvement in depression symptoms. Okay, now maybe you have a plane. But if you give the drugs to every single person, you have nothing to compare. How do you know what your drug is not something else that you're their depression? Maybe a bunch of people were unemployed and during that time that they were given the drug, they got a job. Back didn't improve someone's levels of depression, especially if it's a situational depression. course, there's a depression that is biologically, you know, that's a different type of depression, but there's also situational depression. And if you have an accountant for every single situation that person is in, those are confounding variables that can impact in this case, levels of depression. Do you know how control group, you have nothing to compare. You cannot make this. big problem. can't rule out any other expavation. So that's why minimally we always have to have at least a control group and an experimental group. And as I said, you can have more than that, but the bare minimum requirement, no treatment, treatment, control group, experiment. So when we use, um control groups and experimental groups, individuals are randomly assigned to each group, and I've kind of talked about this a little bit that the need for this in order to make sure that the participants in each group represents the larger population. That's the, right? You're never going to be able to access the entire population. You're not going to be able to access every single person who's ever experienced depression or has symptoms of depression in a drug site. You're going to have to randomly assign participants to certain groups and hope that they represent the larger population of people that experience symptoms of depression, right? So that is the point of random assignment to different conditions. Usually, the experimenter, I mean, ideally, the experimenter doesn't even know who's in which group, in a drug study that is ideal. We call it a double blind assignment where the participant doesn't know if they're getting a drug or not and the experiments or also does't know if they're getting the drug or not. Why? Because bias can be introduced? If they're participant thinks they're getting the drug, um, they can have sort of placebo effects, right? If you've ever heard of that, the placebo effects, or they think they're getting better because they're underlyression, they're getting the drug. researchers can also treat participants differently based on if they know who is getting the drug and who's not, and that will impact the data that they're collecting on that person's behavior. So, ideally, like the perfect scenario, nobody knows what's going on. There's a lab that assigns the drug and puts it in an envelope and assigns names, randomly and they give the envelope to their researcher and there's a red pill and there's a blue pill, but the researcher doesn't know which one is which. One could be trained it, one could be placebo, we don't know, and that's very important, but kind of nobody knows what's going on. until the end. And that's how you get the best data. out of something like this. Now, group should be comparable to each other. um, they should be assigned to the group based on Chancel, essentially. Um, usually we use some sort of computer programming to randomly assign numbers, two people and then randomly assign those numbers into the groups that were trying to produce. This can be very difficult to do. The smaller your participant pool, in fact, the more impossible this gets. So that's why a lot of research studies try to get so many participants and absorbid an amount of participants, um or why you need to run several studies to build your participant pool, before you can make any sort of claim about your data. because the smaller you participant pool gets, the less representative of the population they will be. because you do need to think about things like, um intelligence or um education level personality type socioeconomic status, ethnicity, um, their income, there's so many things that you have to think about and a smaller you participant will gets, the less representative of all of these things it will be. And then we run into the problem that your participants didn't actually represent the larger population, and your data really only applies to that very small group, and it cannot be applied to the larger group, which is always the goal. The goal of research is to collect data with a smaller amount of people, but you hope that you can go and apply those kinds and those results to the larger population. If you are looking for a drug that cares depression, you want those results to be good and to be representative of the larger population so that you can then produce a drug that can be distributed to people who have symptoms of depression and it cures them, right? You don't only want that drug to work for 60 people that you ran the side with, and that's it. and it doesn't work for anyone else. So, random assignment does help with this, but also large participant groups are going to make sure or ensure that you have a representative family of the larger published. Okay. yeah. So some other important things we have to considerable we're running research. and just terms that you should be aware of, so a confederate is someone who is employed by the researcher or is a researcher themselves that is going to participate in the study and pretend to be a participant. So they're gonna essentially take part in the study. um, the participants will not know that that person is a confederate, obviously it's a secret, so this involves some level of deception, usually in the study that we have to present to our review board and make sure that all of that is okay. But when you use a confederate, it's usually because you are conducting a study that people know they're being observed, they're going to change their behavior. So that's why we have confederates. When I was in grad school, a a grad friend of mine, she was in a different lab, and I was helping with her study. She ran this really interesting study on graphic Ed, so people would eat very, very fast. And I'm not just talking like, you know, kind of fast. We're talking like a burrit that big, is gone in one minute or less, like gone. And so, like barely chewing their food, like, wrap it, rapid, you. And as you can expect, there is a lot of health concerns that come first. We had some children who were rabbit eaters in that study. um there was significant choking hazards that had already occurred with some of those personents because they're eating much too fast, too large in bites. Um, but before we could run our study with children, we had to make sure that it was safe and it was not going to impact them too greatly, so we ran it with college students here on campus. Um and when we first started running it, you realized very quickly that they knew they were being observed, and so they were slowing down their eating. They were still eating fast, but it wasn't quite as fast as they had reported in their interviews when we were trying to pull participants. So what did we do? We got conf better. So we had a sticker researcher in there. and we left, so we who were identified as the researchers, we were like, hey, um, we're gonna be back a little later, and we're just we're gonna ask for your report on how fast you ate, but we gotta go. We'll be back later. Maybe pizza in the middle of the room, help yourselves. And then we actually had another researcher in there who was a participant, but she was a confederate. And she had to eat with them, which was difficult because she had to eat very, very quickly, so that they didn't know that she was a confederate. But that is an example of what we would do. Now, she had a time where she had different time on her too, that she was like collecting data for certain people in different sessions, so we could get a truer representation of how fast those people ate and their behavior was a different because they didn't know that they were being observed. So that is a perfect example when we could use the compatory. Um, replication and I've already sort of talked about this before, but we always wanna ask if we can rep replicate the results that have been found. This is extremely important. Scientific understanding is based on the accumulation of knowledge. The more knowledge we have, the more data we have on a on a body of research, the greater our scientific understanding is of that res research, of that behavior, of that phenomenon or theory, or whatever it is that we're investigated, the more research we have, the better we understand it. Replication is foundational to science moving forward. If we adjust did research for the heck of it, just to entertain ourselves to stimulate our reins or whatever research we just want to do, it doesn't help science, it doesn't move us forward at all. We have to publish it and then other scientists, other researchers have to replicate it and move the science forward. It's an extremely important part of research and without it, it really would kind of be pointless to do research at all. The point is to accumulate the knowledge and move the science forward. I've gonna talk about briefly about significant outcomes. If you take a statistical course, they get into this in great detail. But whenever we're looking at data, we're looking for what is called significant outcomes, statistically significant differences. We're not just looking for minimal differences between our groups, between our control group and our experimentsal groups, or even between our different experimental groups, we're looking for significant changes. big changes, changes that make a difference in people's lives. and a difference in their behavior changes, not just very small minuscule differences that maybe we can kind of say, well, there's a slight change. No, there must be a statistically significant dip. Now, of course, that is determined by the statistical analysis that are run. um, or if you're doing a study that's sort of based on kind of like a real world problem, um, things like when we work with children with autism and things like that, um, or any individual with a developmental disability, we're looking for um learning outcomes, so do they make significant jumps in their learning outcomes or their development? E cognitive or physical development, right? So they need to be meaningful differences as, you know, we're not just looking for tinyunicule changes, we're looking for meaningful, statistically significant differences between our groups. Experimental bias is something we always have to be aware of, these are going to be factors that could impact your dependent variable, a bias from the researcher, a bias from the in from the first incipant. Those can impact the data that you get in the way that they be hidden um any expectations that you are the persistent have can surely impact how they are behaving. We always need to account for that and make sure that we're, you know, making sure that doesn't do. Well is a false treatment. I've already kind of mentioned this before, but we typically see this with any sort of drug study um, but it's just the no treatment. They're given a pill that doesn't have any chemical properties to it, so it shouldn't impact their um system.? So if it impacts them in any way? That's what we need when we say alpha seat. And then finally, I've also talked about this already, but double blind means both the experimenter and the person do not know who's receiving treatment and who's not. That is the ideal standard to lose a another one in experiment, nobody knows. And it prevents
Flashcard (10)
Studied by 0 people
Note