You are not currently logged in. Please create an account or log in to view the full course.
Experimental Designs
Generating Lecture Summary...
Generating Lecture Summary...
Generating Vocabulary List...
Generating Questions...
Generating Questions...
- About
- Transcript
- Cite
About the lecture
In this lecture, we introduce experiments, focusing in particular on: (i) recognising experiments as the ‘gold standard’ in research design, due to their ability to establish the presence of causal relationships between variables; (ii) two key terms in experimental design – independent and dependent variables, talking through what each refers to; (iii) outlining how to identify an experiment as a method of study; (iv) the different types of experiments; (v) the benefits and drawbacks of laboratory, field, and natural experiments in the context of differing types of validity and unwanted effects on the results; (vi) within and between subjects designs, including matched design as a sub-type of between-subjects design.
Laboratory Experiment – Scientific study conducted in a laboratory or other such workplace, where the investigator has some degree of control over the environment and can manipulate the independent variable(s).
Field Experiment – A study conducted outside of the lab in a 'real-world' setting.
Naturalistic Experiment – Data collection in a field setting, without laboratory controls or manipulation of variables. These procedures are usually carried out by a trained observer, who watches and records the behaviour of participants in their natural setting.
Within-Subjects Design – Also known as a repeated measures design, within which the effects of treatments are seen through the comparison of scores from the same participant observed under all of the experimental conditions.
Between-Subjects Design – Also known as an independent groups design, within which participants are assigned to only one of the experimental conditions and each person provides only one score for data analysis.
Matched-Pairs Design – A derivative of the between- subjects design involving two participant groups in which each member of one group is paired with a similar participant from the other group(s). Participants may be 'similar' if they share one or more characteristics that are not the main focus of the study but could still influence the outcome.
Order Effects – In a within-subjects design, order effects are the influence of the order in which treatments are administered or tasks are completed. This can result in performance differences between participants which are not due to variables of interest to the study. This can be combatted through counterbalancing.
About the lecturer
Dr Eoin O’Sullivan is an associate lecturer in the School of Psychology and Neuroscience at the University of St Andrews. Dr O’Sullivan teaches the first-year undergraduate research methods course and is interested in uncovering novel teaching techniques in the field of research methods and statistics, within psychology. Some of Dr O’Sullivan’s recent publications include ‘Automatic imitation effects are influenced by experience of synchronous action in children’ (2018) and ‘Understanding imitation in Papio papio: the role of experience and the presence of a conspecific demonstrator’ (2022).
So if we have an aim in a research project and we've developed some hypotheses,
00:00:07we've chosen a population and a way of sampling from that population,
00:00:12and the next thing we need to decide on is
00:00:16what specific method will we use to test our hypothesis.
00:00:18To answer our questions,
00:00:22we could use surveys, questionnaires.
00:00:24We could interview people to learn more in depth detail about experiences.
00:00:26We could observe real behaviour
00:00:31or we could conduct an experiment. And that's what we're going to talk about today.
00:00:33Experimental designs.
00:00:37And so experiments are often considered that the gold standard in research design
00:00:39because with experiments we can get a really good idea of causation.
00:00:45Does one variable
00:00:50cause a change in another variable?
00:00:52Now you can examine relationships between variables,
00:00:54things that vary in a lot of different methods.
00:00:58You can observe behaviours and and see whether certain
00:01:00behaviours occur in the context of other behaviours.
00:01:04So whether there is some relationships going
00:01:06on between different variables that you're measuring,
00:01:08but with experiments, you as a researcher are in control of one of those variables.
00:01:11You do something at the start of the
00:01:16study to change the experience of your participant.
00:01:18Um, by making this change and then measuring the outcome of it,
00:01:22you can be more confident that that change or that change in
00:01:26your in your first variable has caused a change in your outcome.
00:01:31Variable.
00:01:34These variables have names.
00:01:35The thing that you're changing is called the independent variable,
00:01:37and the thing that you're measuring as an outcome is called the dependent variable.
00:01:40If you ever reading about a study, are learning about a study and you want to,
00:01:44you want to check what method is the researcher using?
00:01:47The best way of identifying an experiment is to ask yourself,
00:01:51Did the experimenter or researcher change something in the design of that study?
00:01:55Did they change the experience of a particular participant that the, uh,
00:02:01manipulate or change one of those variables to see
00:02:05how that would impact a different variable and outcome?
00:02:07If that is the case, then the researcher has conducted an experiment,
00:02:10and as I said,
00:02:16it's one of the best methods for identifying causation because as an experimenter,
00:02:17you can try to keep everything else the exact same in terms of participants
00:02:21experience while just changing that one small thing that you are interested in.
00:02:25So there are a number of different types of experiments,
00:02:30and the most common one, the one that best known,
00:02:35is the lab based experiments of the laboratory experiment.
00:02:38In this type of experiment,
00:02:42your participants will will come into a laboratory
00:02:44setting a setting that is very controlled,
00:02:46where you as an experimenter can keep everything as consistent as
00:02:48possible and just change the thing that you're interested in examining.
00:02:52What's good about these lab based studies is
00:02:56that you can very clearly control things,
00:02:59and so it might be easier for other researchers to
00:03:01do the same thing as you by controlling the environment.
00:03:04It also has something called good internal validity because
00:03:07you're controlling everything and only changing your independent variable,
00:03:11then you can be confident that if you find
00:03:15a difference in your outcome in your dependent variable,
00:03:18you can be confident that that change is because of a change
00:03:21in your independent variable and not another variable that you aren't measuring.
00:03:24These other variables are called extraneous or confounding variables.
00:03:28In a lab based setting, you try to control for those ones,
00:03:33manipulate your independent variable and examine
00:03:37the outcome on your dependent variable.
00:03:40However, there are some issues with lab based studies, Uh,
00:03:43in a laboratory environment doesn't really resemble the real world.
00:03:47And so you may get some behaviours that are that
00:03:52are fake or not really representative of real behaviour.
00:03:54This is called poor external validity or ecological validity.
00:03:57Another thing that might happen in a laboratory experiment
00:04:02is demand characteristics when participants come into a laboratory setting
00:04:05and they expect to behave in certain ways.
00:04:09And they might, uh,
00:04:11they might start guessing what the experiment might be
00:04:12about which may lead to them changing their behaviour.
00:04:15And it's an example of a lab based experiment.
00:04:19Would be, um,
00:04:23say we're interested in examining the impact of caffeine on attention.
00:04:24Um, so here are aims to understand whether consuming caffeine affects attention.
00:04:29So we're interested in caffeine that's our independent variable.
00:04:34Whether that has an effect on attention are dependent variable.
00:04:38We invite participants into our lab,
00:04:41and we, uh, manipulate one thing whether our participants receive caffeine or not.
00:04:44So the group that received caffeine,
00:04:50they are experimental group and the group that don't there are Control group,
00:04:52and what we might do is actually give the control group,
00:04:56decaffeinated coffee or decaf coffee so that they experience a
00:04:58similar taste in a similar experience to the other participants.
00:05:02They're just not receiving the caffeine.
00:05:05So we changed that, whether participants has caffeine or not.
00:05:08And then we measured attention to see
00:05:11whether consuming caffeine might change someone's attention.
00:05:14Uh, for that we'd have to develop an attention task.
00:05:19Let's say we had a video recording of people walking down a busy street,
00:05:21and we asked our participants to count every
00:05:25single person that walks by in the video.
00:05:28In that way, we've got very controlled, uh, test or task to measure attention.
00:05:31That will be the same for every single one of our participants.
00:05:37If we see that there is a difference in the accuracy on that test week,
00:05:39if there's a difference depending on whether our participants and caffeine or not,
00:05:45we can be more confident that that difference
00:05:49is caused by our manipulation in this case,
00:05:51caffeine.
00:05:54Now an issue with this experiment, as I as I said, is that it's it's unrealistic.
00:05:56This isn't how people normally drink coffee,
00:06:00and so maybe something about that situation
00:06:02will lead to abnormal or different behaviour.
00:06:04and so a different type of experiment that we could conduct is a field experiment.
00:06:07In this case,
00:06:11an experimenter would still manipulate an independent variable
00:06:12to examine its impact on a dependent variable.
00:06:15However, it would take place in a more naturalistic setting out in the wild.
00:06:18Now the advantage here is that things are more realistic.
00:06:23It has greater ecological validity.
00:06:26But you have less control over other variables,
00:06:29these other confounding variables.
00:06:33And they may have an impact on your results,
00:06:35which may make it less obvious whether your independent variable
00:06:37is actually having an effect on your dependent variable.
00:06:40Let's take our coffee example.
00:06:44We could try to conduct that study as a as a field experiment.
00:06:45We might go to a coffee shop as participants walk into the coffee shop.
00:06:49We can say that we will give them a free cup of coffee,
00:06:53but they may receive one with caffeine or one with decaf coffee,
00:06:56and they won't know what they're receiving.
00:07:00Um, so you randomly allocate participants to your conditions,
00:07:02caffeine or non caffeine,
00:07:06and then you ask them to do the attention task.
00:07:07Now we could show them something on a computer
00:07:11and they would again count the number of people.
00:07:14Or instead we could just ask them to count the number of
00:07:16people that are passing by the door of the coffee shop again,
00:07:19making the task a little bit more realistic.
00:07:22So here, as I said,
00:07:24we've got a more realistic measure of the impact of coffee on
00:07:26people's behaviour because it's taking place in a more realistic setting.
00:07:29We have a more realistic task. As it relates to attention,
00:07:32however, we have less control over the situation.
00:07:36And so there could be other factors that come into play that
00:07:38make it more difficult for us to draw conclusions about our data.
00:07:41The final experiment type of experiment is a natural experiment.
00:07:46In this case,
00:07:50the experimenter actually doesn't do the manipulation and themselves.
00:07:51Something changes in the real world.
00:07:54And then researchers swoop in to see whether
00:07:57that change has an impact on another variable
00:07:59again. The advantage here is that it's more realistic.
00:08:03You're less likely to have demand characteristics because
00:08:06it's taking place outside of that laboratory setting.
00:08:09Um,
00:08:12and one thing that's very advantageous about this
00:08:13method is that sometimes you can use it
00:08:15when you wouldn't be able to manipulate a variable in an experiment or a lab,
00:08:17a lab or a field experiment because it might be unethical, for example,
00:08:22to manipulate that variable.
00:08:27For example,
00:08:29a number of field or a number of
00:08:30natural experiments have taken place in recent years.
00:08:31Examined the impact of the introduction of smoking bans in restaurants and bars.
00:08:34Um, so this can happen in countries where some cities will have introduced bands,
00:08:41but other cities don't have the bands,
00:08:47and so experiences come in and there's this thing that's manipulated,
00:08:50whether there's smoking going on in restaurants and clubs and other public places,
00:08:52they can examine the impact of this change on various health outcomes.
00:08:57The experimenter isn't doing the manipulation,
00:09:02but they're able to examine how these changes
00:09:04in the same places are impacting other variables.
00:09:06For example, the small, uh, in this case, the impact of smoking bans on help.
00:09:10Now,
00:09:17an important thing to mention here is that we would
00:09:18not be able to do this in a lab setting.
00:09:21We cannot blow smoke in our participants faces to
00:09:22examine the impact of third hand smoke or second hand
00:09:26smoke because that would be unethical That's why it's important
00:09:28that we can do it as a natural experiment.
00:09:33But again, the major issue here is that there's less control over what's going on.
00:09:35And so, um, it is.
00:09:40It is a little bit more difficult to draw solid conclusions about about your data,
00:09:41so there are different types of experiment.
00:09:49But another thing that you need to think about with
00:09:50experimental design is how you're going to test your participants.
00:09:53We need to participant be tested once or more than once
00:09:56if they're tested more than once you've got a within subjects design.
00:09:59Another worth another term for that is repeated measures design.
00:10:03Um, if you're only testing every participant once you've got a between subjects,
00:10:08design or an independent samples design
00:10:12and a final design is a special type of between subjects.
00:10:15Design a matched design where you match certain
00:10:19participants in different groups based upon certain characteristics.
00:10:22So in the context of a between subjects design,
00:10:26each participant is randomly assigned to one condition,
00:10:30so participants arrive in the lab and you can randomly
00:10:34placed them into your experimental condition or your control condition.
00:10:36In the example,
00:10:40we discussed about caffeine and the impact of caffeine
00:10:41and attention that would be a between subjects design.
00:10:43Some participants get caffeine, some get decaf coffee without caffeine,
00:10:47and you examine the impact on attention.
00:10:52Now the advantages here is that you've
00:10:55got a random spread of participant variability,
00:10:57and there is no risk of something called an order effect,
00:10:59which is an issue with within subjects design, which I'll discuss in a minute.
00:11:02One of the issues with a between subjects design
00:11:07is that when you randomly allocate participants to different groups,
00:11:10different conditions,
00:11:13it could be that a difference exists there already.
00:11:15So, for example,
00:11:19let's take our our study looking at the difference in caffeine and attention.
00:11:20If we were to take participants and randomly
00:11:27allocate them to one of those two conditions
00:11:29and maybe by complete chance,
00:11:31we randomly allocate a lot of older participants to one group
00:11:33and younger participants to another group Now, in that case,
00:11:37we may find that there's a difference between our groups
00:11:41when it comes to their performance on the attention task.
00:11:43But because there is that already different,
00:11:46there is already a difference in terms of age of our Group A and Group B.
00:11:48It's unclear whether that difference is caused by our independent variable Kathy,
00:11:53or whether it might be due to that, um, that unmeasured
00:11:58and pre existing difference between our groups.
00:12:02Another issue with between subjects design is that people vary.
00:12:06People are very different in various ways.
00:12:10And so when you've got two groups of people randomly sampled, um,
00:12:14they will be different already in in their various measures, for example,
00:12:19people have very different performance abilities when it comes to attention.
00:12:25Some people are very good at paying very close attention to
00:12:29things other people less so and so in your two groups,
00:12:31you're going to have this massive amount of variability.
00:12:35And so the difference actually within your group could
00:12:37be quite large when it comes to attention.
00:12:40And so if you're looking at the effect of coffee on attention,
00:12:43it could be relatively small in comparison to this large individual variation.
00:12:46What that means is, when you're testing your, uh,
00:12:52your hypothesis testing whether your individual or your independent
00:12:56variable has an impact on your dependent variable,
00:13:00you might find that there is no difference because that tiny
00:13:02effect of caffeine is hidden or masked by the much larger,
00:13:06very individual variability that already exists there.
00:13:10So another way of designing your experiment is
00:13:15with it with a within subjects design.
00:13:18And in this case, your participants are measured on multiple occasions.
00:13:20Um, in the example of, uh, of caffeine and its impact on attention.
00:13:25On day one, for example, we might, uh,
00:13:30have our participants come to the lab and test them on,
00:13:33um, give them a cup of coffee and examine their performance,
00:13:37their their task performance and attention.
00:13:41After having caffeine on the second day,
00:13:43we test them again without caffeine and see whether there's been a change in,
00:13:45um in their attention accuracy.
00:13:49Um, so here the advantages is that you eliminate that individual differences.
00:13:52Those individuals vary abilities,
00:13:58um,
00:14:00and so it's easier to identify an impact of that individual
00:14:00of that independent variable because you're testing the same person twice.
00:14:04For example, I may have a high level of attention already,
00:14:08and and so I have this high rate of attention.
00:14:12Without coffee, that little bit of coffee, it improves my attention.
00:14:15It's easier to identify it.
00:14:19Um, that's called greater statistical power.
00:14:21It's easier to identify an effect if that effect exists,
00:14:24and so you may need fewer participants to actually test your hypothesis.
00:14:27However,
00:14:32one of the main disadvantages with this approach relates to order effects.
00:14:32If you're testing people on multiple occasions,
00:14:36it could be that by testing them twice you're already changing
00:14:38and
00:14:42how they're going to perform on your dependent variable.
00:14:43If you test people twice on a measure of attention,
00:14:46it could be that on the second attempt at the second day,
00:14:48they're better because they've had some practise.
00:14:52Or they may be worse because they're getting bored with the task.
00:14:54And so they attend to the task a lot less.
00:14:57And so then, when you're measuring people on multiple occasions,
00:15:02it's less clear again whether your independent variable is having an effect on your
00:15:05dependent variable or whether it is this repeated testing that's having an impact.
00:15:10And so when you have this repeated measures design within subjects design,
00:15:15it's important to to counterbalance the presentation of your conditions.
00:15:19So some of your participants should experience one
00:15:25of the conditions first and the other second,
00:15:27and the other half of your participants should have the opposite and order.
00:15:29So some participants,
00:15:34in the example of coffee and attention should have caffeine first and
00:15:35decaf second and the other participants will have decaf first and caffeinated.
00:15:39Second,
00:15:44This allows you to find out what the impact of
00:15:45caffeine is by getting around the impact of order.
00:15:49The final type of design I'm going to talk about,
00:15:54uh takes advantage of the positive aspects
00:15:57of both between and within subjects designs,
00:15:59so there are different in a matched designs.
00:16:02There are different participants, so it's a between subjects design.
00:16:05However, in your different groups,
00:16:09you match your participants based on certain characteristics.
00:16:12So let's take the example of age.
00:16:16You want to make sure that you don't have very
00:16:19different age can distributions in your two different groups.
00:16:21So what you might do in your two different groups is ensure that there
00:16:26are a certain number of participants that are over the age of 60.
00:16:29A certain proportion of your participants are between the ages of 40 and 60
00:16:33and a certain proportion of your participants are between 20 and 40 and you
00:16:38make sure that there's a balance of those participants in both of your conditions.
00:16:41What's positive about this is that
00:16:48you're reducing group variability you're reducing the
00:16:49chance that there's going to be a difference between your groups to start
00:16:53off with and ensuring that any difference observed in your groups is due
00:16:56to your independent variable and not due to another pre existing difference.
00:17:01One drawback with this method is that it does require
00:17:05a lot of effort and a lot of resources,
00:17:08and so it can be quite time consuming and
00:17:10is used a little bit less in scientific research.
00:17:12So in this lecture, we've covered experimental design.
00:17:17We've discussed why it's so important in understanding causation
00:17:20discussed a number of different types of experimental design,
00:17:24from the lab based experiments to the natural experiment.
00:17:27We've also discussed different ways of using our participants in experiments,
00:17:31whether we're using the same participant just once or on multiple occasions,
00:17:35you know, within subjects design.
00:17:40I hope you learned a little bit about the importance
00:17:42of experiments and why they're so useful in psychological research.
00:17:44
Cite this Lecture
APA style
O'Sullivan, E. (2021, November 17). WJEC Eduqas A Level Psychology (2024) - Experimental Designs [Video]. MASSOLIT. https://massolit.io/options/wjec-eduqas-a-level-psychology?auth=0&lesson=4180&option=6009&type=lesson
MLA style
O'Sullivan, E. "WJEC Eduqas A Level Psychology (2024) – Experimental Designs." MASSOLIT, uploaded by MASSOLIT, 17 Nov 2021, https://massolit.io/options/wjec-eduqas-a-level-psychology?auth=0&lesson=4180&option=6009&type=lesson