Transcript for:
Research Methods in Psychology

foreign the second lecture here is going to continue talking about research methods in the last lecture I introduced really the three main categories of research methods that are typically used by psychologists so one of those was descriptive methods and then I mentioned correlational and experimental so I'm going to walk through each type with descriptive methods think about if you want to make observations of people or subjects in either a in a natural setting or in a laboratory setting there are a few excuse me kind of issues that might get in the way of really making accurate sort of observations so one of those is something called demand characteristics sometimes called the observer effect and this is the idea that if individuals know that they are actually being watched by someone not necessarily even if they know they're being watched by somebody that's taking measurements in any sense but if someone knows they're being watched their behavior might not be what it would be if they were not being watched so for example or one example prejudices might not be displayed in groups because individuals know that others are paying attention another issue is something called Observer Bias and this is the idea or really the issue that our expectations of what we might see in a situation or a research setting can actually influence the observations that we make can actually influence our perceptions of reality if we go into some circumstance thinking we are going to see a certain thing then we may be showing some bias couple ways to eliminate demand characteristics and Observer Bias one that will eliminate the observer effect is is by running a single blind study and what that means is the subjects are unaware of actually being observed okay so that will clearly help with the observer effect and then really the best way although it's not always easy to run a study like this but would be to do a double-blind study in this case not only do the subjects uh not only are the subjects not aware they're being observed but also the researchers that are actually taking the measurements or conducting the study do not know the actual purpose of the study so for example if graduates graduate students are out taking measurements they may not know the principal investigator of the lab's actual research question for example or what the what the pi intends to actually analyze or study so double blind would be great again not always easy to conduct a study that way a good example of uh what's called naturalistic Observation so making observations in in a natural setting is the work of Jane Goodall and you guys have probably heard of Jane Goodall one point or another but she devoted Decades of her life to studying uh chimp behavior in the wild so unobtrusively she would go make these observations and try and relate chimp behavior in some ways to human behavior um cooperation things like that child rearing a number of just a ton of work and a ton of studies that she put in to try and understand behavior that way some an issue a big issue with naturalistic observation is if you're talking about trying to understand some phenomenon really the naturalistic observation or any observational method uh can only describe so it can't explain anything causal um one advantage then would be that if you are doing naturalistic observation the behavior itself is occurring in a naturally in a natural situation so unlike if you're doing those observations in a lab for example and also if you're making your observations in a natural setting then single blind becomes a little bit easier and doing double blind studies also depending on whether or not the the person making the observations is aware of the purpose of those observations so another descriptive method is something called a case study and in a case study typically one or maybe a few individuals that are unique are studied in depth so in great detail in the hopes of being able to generalize to the general population so case studies oftentimes will be the result of some uh Oddity or something that occurred during development something surgical so a lot of neurological cases present good opportunities to do case studies because we can try and understand Behavior through kind of the Neuroscience or the biopsychological lens but a great example of case study one of the most famous is patient hm so here's hm here at the bottom we know his name now it was Henry malazin for a while he was just hm he passed away oh 10 12 years ago but hm in his 20s he had severe severe epilepsy and it interfered with his daily life um and so what neurosurgeons hoped they could do is figure out where his seizures started in his brain and they did that and then remove that part of the brain in the hopes of curing essentially the epilepsy so they figured out that in hm's brain seizures seizure activity began in the part of the brain called the hippocampus so if you look over here at a normal brain the hippocampus is in What's called the medial temporal lobe we'll go into more detail on that when we talk about biopsychology but you can see here's a normal brain compare that to hm's brain they just basically ripped out a giant chunk of his brain they ripped out the hippocampus on both sides so there's a hippocampus on the right in the right and the left hemisphere if you look at this section it shows you really the same thing just in a little different view what's called a coronal section if you look at a normal brain there's the hippocampus and hm hippocampus was was removed so what happened then no one really could have necessarily predicted there wasn't a lot known at the time about what the hippocampus was important for there were some ideas but in hm's case and there were other cases like hm but in hm's case he really didn't have much of a life he had severe epilepsy so the intention was to uh cure it and they did have uh a dramatic effect on reducing his seizure activity by removing his hippocampus didn't 100 cure it but uh it it did a good job at its intention of doing that but what happened to hm was pretty amazing so from the point of his surgery on HM could never form a new memory and he could never form a new specific type of memory so he couldn't form new memories for for facts or events everything in his past life he pretty much remembered fine but he couldn't go in the forward Direction and and make new memories in fact the doctor one of the famous uh psychologists that studied them for decades Brenda Milner would have to introduce herself to hm every time she met him even though she met with him a lot so there you go there's an example of a very famous case study there are lots of examples particularly in neurology of things like the idea of the case study is again like I said to try and study one sort of unique occurrence to understand a little bit better about the population and the problem with the case study then you might already can anticipate is that maybe the individual cases are misleading maybe hm had a weird memory from the beginning and so studying hm wouldn't be a good idea if you're trying to generalize uh it turns out that we now have tons of evidence showing us that the hippocampus is indeed essential for forming new memories but at the time like I said the doctors really didn't know it would have that dramatic of an impact on hm's memory formation so case studies they don't necessarily allow us to draw generalizations although sometimes they can put us on the right track it would be a little bit of an error to necessarily always generalize from one individual case last descriptive method that I'll tell you about is a survey so I'm sure all of you have taken surveys at one point or another uh it kind of seems like we do this all the time but a survey is nice in the in the sense that you can get a lot of data pretty quickly it's cheap all you need is paper and some willing subjects um you want to take a representative sample if you're doing a survey that's going to reflect uh basically the population that you're trying to Target so if I'm trying to Target uh learning strategies and undergraduates obviously this is a ridiculous example but I'm not going to give my survey to a bunch of folks at a nursing home for example so a representative sample of the population of interest is important and should be randomly selected you don't want to be targeting let's say drug generalizations about undergraduate learning and then study a unique sort of University maybe I don't know you might not want to avoid some of the ivy league or something or not just limit yourself to studying undergraduates in Alaska or whatever might bring in a compound something like that so it's important to randomly select subjects and have a representative sample surveys can give us lots of information very quickly and in a cheap and easy sort of way some examples of some survey questions that are just not so good so first one please indicate how much you agree or disagree with each of the following statements about the Child Care Program here is um there's something wrong with one of these options with this option sorry so I feel welcomed by staff and other youth at the center all right so why don't you guys maybe pause this and think about what might be wrong with that option so the problem here is that uh it's it's kind of like a a double hit so to speak so I feel welcomed by both staff and the youth and there's no option to choose one or the other so maybe for example the individual feels welcomed by the staff but the youth are just you know terrorizing the person so it doesn't really give the option to choose one or the other so a better way would be to provide both options I feel welcomed by the staff at the center and then also I feel welcomed by other youth at the center okay here is another example so in this case community community organizing is hard do leadership trainings help you feel prepared for Community organizing and again here's your your choices uh much more prepared somewhat more prepared slightly more prepared not more prepared so why don't you guys pause again and then see if you can find couple issues with this one okay so this is kind of like a leading question right so Community organizing is hard it's that statement itself kind of is already biased what if the individual reading this question does not feel at all like Community organizing is hard asking the question in this way or making the statement in this way sort of leads the individual kind of um to start thinking that way and then if you look at the options much somewhat slightly not more prepared we can probably all agree that much and not we can distinguish between those but maybe slightly and somewhat a little bit ambiguous a better way to ask that and this is what you commonly see with survey questions so the leadership trainings prepare me for Community organizing and then your typical strongly agree agree disagree strongly disagree those seem a little less ambiguous of choices another example so your survey question should be specific here's an example of maybe where that goes around so should government do something about health care you can imagine asking that to some person who might just be willing to write an entire book about that okay so not very specific uh second option should government pass a law on Health Care again not super specific it's kind of an open question somebody could really go on and on and on about that and then finally maybe a little bit better should government pass the law on health care that would guarantee health coverage for all and that's pretty good that's pretty specific you can kind of answer that with a yes or a no all right another example and you see this kind of thing sadly quite a bit so what is your age pretty straightforward question uh and here are your options 0 to 10 10 to 20 20 30 to 40 40 plus so hopefully you guys can see what's wrong with that what if you happen to be 10 or 20 what which one do you Circle think about the nightmare for the individual that has to score maybe it's hundreds or thousand surveys where do you kind of put uh the choice 20 for example okay so that's all I'm going to say about descriptive methods of course you're reading the chapter and I'm going to move on and talk a little bit about correlational methods so I had already kind of briefly went through this but if you remember correlation kind of tells us the relationship between variables uh maybe it's two variables maybe it's two or three or four or whatever but the relationship between two naturally occurring variables hopefully allowing us to predict one variable if we know some level of the other foreign so a correlation will tell us the extent to which two variables or two factors kind of differ or vary together and giving us some idea how well one variable can predict the other when two events are correlated they are related to each other and here's a good example uh nice weather correlates with the use of flip-flops so let's say you live uh I guess in Texas you can kind of wear flip-flops a lot of the time but let's say you live in up north you live in Minnesota for example uh you don't see a lot of folks wearing flip-flops in the winter time uh that might be dangerous so yes we can probably find a pretty strong relationship between temperature or the weather um and the individuals wearing flip-flops but importantly correlation doesn't allow us to draw any causal conclusions correlations allow us to make predictions but we can't say one variable causes the other so there's nothing necessarily about the temperature that kind of like a puppet pulls you into the closet and causes you to put flip-flops on there are obviously other factors that are going to influence your shoe choice so although these things are related we wouldn't say that you know Sunny 75 80 degree weather causes the use or it causes the wearing of flip-flops okay a little more on correlation so we can distinguish between what are positive correlations and negative correlations and if you've read the chapter you already have a good idea of what that might mean but a positive correlation indicates a direct relationship two events or two variables will increase together or decrease together so if one variable goes up the other does as well if one variable goes down the other does as well so an example here and this is published work the longer that children are breastfed the greater the their later academic achievement on some standardized tests or whatever it might be the longer breastfed the greater the achievement so one variable breastfeeding is going up so is the other variable that's being measured academic achievement an example of a negative correlation is when when one variable increases the other will decrease or vice versa the more children and youth use various media the less happy they are with their lives so you can see as media use social it might be goes up happiness goes down so maybe that's some survey or something like that about level of Happiness one variable going up the other going down that's a negative correlation I want to make the point that there can be whether it's positive or negative doesn't really have anything to do with the strength of the relationship you can have a strong negative correlation just as well as you could have a strong positive correlation it just tells us the direction of the relationship so correlations let us make predictions if I know how long a child has been breastfed I can draw some conclusion or some make some prediction about their later academic achievement because we know there is a relationship like I said correlations do not let us make cause effect explanations so why might breastfeeding be related to academic achievement there could be many possible explanations correlation just shows us the association whether or not there is one and whether it's a positive Association or a negative association so the strength of the the strength of a correlation is determined by a a mathematical test which I'm not going to have you guys walk through I will just tell you the output of that mathematical test is something called the correlation coefficient and this is a measure of the the strength of a correlation and if that the number you get when you run this mathematical test is close to zero you have a week or maybe almost no correlation farther from zero the stronger the correlation and the possibilities the possible results of your mathematical test are going to range from a minus one to a plus one again closer to one in either direction is going to indicate a stronger relationship and zero would indicate no correlation so here you're looking at something called a scatter plot the scatter plot just graphically depicts the relationship between two variables so I have given you guys I had given you guys an example experiment where I talked about is uh sleep or lack of sleep can impair memory performance so let's use that I think it's a decent example let's use that um as and pretend as though we were doing a correlational study we didn't have bring subjects into the lab necessarily and dictate the amount of sleep they got we just maybe give out a survey or something and ask them how many how many hours of sleep they got and then gave them a test a memory test so if you look at this scatter plot on the left you see memory performance on the y-axis okay and on the x-axis we have hours of sleep and you can eat pretty easily see that as hours of sleep go up so does memory performance here we have basically a perfect relationship it's a straight line so that might be a correlation of one which you never see in psychological studies but nevertheless for the sake of illustration you see this is a perfect relationship you can draw a straight line through these points as hours of sleep go up we can predict memory performance will as well again we don't know why correlation just lets us see the association moving to the scatter plot in the middle here is a correlation of minus one and let's let's let's say let's say for example uh I know this isn't true but let's say that memory performance is still here on our y-axis and hours of sleep is still here on our x-axis let's say it turns out this is ridiculous but just pretend let's say that as hours of sleep go up memory performance goes down and that relationship is perfect we know that's not true but just for the sake of it so here we have again a straight line and it's in the other direction so we have a straight line as sleep goes up memory performance goes down and we have a strong negative correlation let's call that minus 1.0 that's as strong as you can get of a negative correlation this is as strong as you could get of a positive correlation so our mathematical test is going to tell us the relationship between these variables and spit out a number that is between minus one and plus one if there's a minus in front of the number we have a negative correlation if there's a plus we have a positive correlation okay finally on the right here we have a scatter plot showing basically no relationship as hours of sleep go up you can't predict really at all what the memory performance would be okay somebody that gets a lot of sleep might have a poor memory performance somebody that gets little sleep might as well and some and vice versa someone that gets a little bit of sleep has a great memory performance same thing for individuals with a lot of sleep so there's really no relationship you can't really draw much of a line through these points okay so if you really want to get a sense of a relationship between variables of Scatter Plots very useful and you would you would calculate this correlation coefficient to determine a mathematical relationship between the variables again keeping in mind that the correlation coefficient is only going to range between minus one and one and the closer it is to 1 whether it's plus or minus one the stronger the relationship closer to zero and it would indicate not much of a correlation so here is this is a published study way back in 1975 showing the relationship between the number of toasters in one's home and the use of contraception and stop yourself pause it say is this a positive or A negative relationship and does it look like it might be a strong relationship and hopefully you came to the answer that this is indeed a positive relationship and it looks fairly fairly strong wait a second that's pretty goofy how could number of toasters in the home be related to use of contraception that seems ridiculous okay but it's true it's published this is work this somebody did the study so use of contraception is predicted by a number of appliances in the home in this case toasters so that makes you think is that there really some direct relationship there do you think that one thing's causing the other looks like a pretty strong corally so let's think about it for a second and this will lead us into something called the third variable problem which is very important a very important thing to understand when you're thinking about correlation okay so it could be sure and some weird Universe it could be the number of toasters in the home somehow predicts contraception or there could be some variable we're not accounting for that's related to both of these things to both toasters and contraception turns out we still don't know we can't draw causal conclusions but one variable that happens to be related to both things is that it turns out that uh socioeconomic status or amount of income affects number of appliances in the home okay that should be fairly intuitive but also uh uh propensity to use contraception as well so money happens to be related to both of these variables it's our third variable it's not accounted for in the scatter plot but it might be again we can't draw causal conclusions but maybe potentially that might be one explanation for this relationship okay so like I said that's called the third variable problem a causal relationship between two variables cannot be inferred from the naturally occurring correlation between them because of the ever-present possibility of the third variable and that is the reason that we cannot draw causal conclusions when correlational when we do correlational studies here's an example that I think maybe will sear in this idea of the third variable problem and so I hope you're not about to eat lunch because this is a little bit a little bit disgusting but again I hope it's something that is memorable and you'll think about it so back in the early 1900s uh palagra was a somewhat common condition it was a vitamin deficiency vitamin B uh it was an epidemic for a while in the southern states uh it was associated with um these skin lesions and weight loss and some cognitive issues and nobody really knew what was causing it prior to this fellow in the upper right corner here Dr goldberger it was believed to be due to infectious spread unsanitary conditions okay so Dr goldberger didn't think so Dr goldberger believed it was due to malnutrition okay and he proved it he did an experiment really uh with a small n a small number of subjects but he did an experiment and a little bit of a snippet of his experiment is down here on the bottom so I'll read this so uh finally he selected two patients one with scaling sores the other was diarrhea he scraped the scales from the sores mixed the scales with four cubic centimeters of urine from the same patient added an equal amount of liquid feces and rolled the mixture into little dough balls by the addition of four pinches of flour the pills were taken voluntarily by him his assistance and by his wife so he proved and he did none of none of whom contracted palagra so he proved that it was not likely due to unsanitary conditions or infectious spread and then after further study he went on to show it's due to vitamin deficiency okay so this is our third variable problem both malnutrition and unsanitary conditions happen to be happen to be correlated so if these two variables are correlated and we're also looking at pellagra as being our other variable it's hard obviously it's hard to draw causal conclusions as to which one of these two variables might be responsible for the palagra epidemic Dr goldberger kind of eliminated one of them by doing this experiment that I described below okay so because these two variables themselves are correlated that led to our third variable problem in this case our third variable that is not likely to be causing the palagra epidemic was unsanitary conditions even though it was strongly correlated with pellagra it also happened to be correlated with malnutrition which we know now to be the cause okay sorry about that guys and I know probably won't read that again try not to think about the details there but just the big picture in terms of the third variable issue that was happening in the understanding of this condition okay so how do psychologists come to conclusions so we talked about correlation when two events occur together or are related to each other um that's a correlation and we can use a mathematical process to determine the strength of that correlation we can use Scatter Plots to get an idea if the relationship is positive or negative again if one variable goes up does the other go up or down um an example lack of sleep correlates with poor memory I've walked through an experimental example of that too uh but importantly like I've tried to highlight correlation doesn't tell us if one thing causes another in order to figure out causal relationships we have to do and experiment so I'm going to walk into experimental methods keep an eye out guys uh this week I will post up just some practice exercises you don't have to turn this in but I'll post up some practice exercises related to correlation and also a few related to um to experimental methods as well so look keep an eye out for that okay so in order to determine to determine causation we need to do an experiment and that's the method I'll talk about so I've talked about descriptive methods again I'll highlight the the fact that descriptive methods don't allow us to draw causal conclusions they're basically just observations of a phenomenon sometimes those observations are made in very few individuals like case studies the observations may be made in a natural environment or a lab setting and we can use Serve things like surveys or self-reports which allow us to gather a lot of data in order to to do observational tips and methods as well or descriptive types of methods as well okay so correlation I just kind of walked through allowing us to use one variable to predict another finally I'm going to go through experimental methods and experimental methods are the most powerful because the researchers that do experiments are manipulating the variables in the experiment to determine cause and effect okay so clearly that's going to be more powerful than these other methods because we can figure out if one thing is causing another and that's kind of the ultimate goal all right so like I said the experimental methods allow researchers to isolate cause and effect and it's really because the researchers in an experimental study are going to be determining or manipulating manipulating is the important thing to keep in mind manipulating the variables in the study okay so take another example does breastfeeding have an effect on later academic achievement or maybe they correlate because of some third variable if we do an experiment we're kind of getting away from that issue that plagues correlational studies because we are manipulating variables okay this slide is showing you some important characteristics of experimental methods and I'll walk through these with examples and more details but experimental methods involve random assignment control and experimental groups and also I've already described to you guys independent and dependent variables remembering the importance typically if we're talking about an independent variable that's going to be the variable that researchers actually manipulate in their experimental study okay so Random assignment and we're doing a random assignment uh we assign participants to the experimental and control groups more or less by chance there's sometimes something called matching will happen and I'll describe that in a second here but typically with random assignment participants are assigned to a particular group in the experiment might be many experimental groups and many control groups but they're assigned randomly by chance so what that's hopefully doing is eliminating any pre-existing differences that makes may be present between the groups and thus allowing the researcher to then manipulate one or two variables and look at the effect on the outcome a control group you guys probably know this a control group is the group that just isn't going to receive the experimental treatment isn't going to receive the the manipulation so easiest way to think about that would be in some drug trial study for example you guys probably know that one group of subjects or patients is going to receive the drug that might be under study for potential approval by FDA or whatever so the the group that gets the drug is going to be the experimental group and the group that gets up something like a placebo or a sugar pill not the experimental drug would be our control group and we're hopefully then going to be able to make comparisons between those groups to determine if the experimental drug under study is having some beneficial effect okay I mentioned that we do random assignment but it's also important in many cases this doesn't always necessarily have to be true but in many cases if you want to have a a broad subject pool let's say you have uh like this this is showing us let's say you have for example young and old males and females that's what this is trying to show us so it's a pretty Broad sample but you want to ensure does not happen you do not want to have an experimental group for example that has all females and a control group that has all males unless you're studying sex differences then you might want to do something like that along those lines but if you are not interested in that you wouldn't want to divvy up groups based on set you want to have an equal number roughly of females in one group and and males in that group and vice versa in experimental versus control groups you don't want to have the opposite case where one group is overloaded uh with males or females same thing is true of age you don't want to have all elderly folks in in the experimental group and all the younger folks in the controller so we do something called matching which is still a random assignment we're randomly moving subjects into one group or the other but along the way ensuring that we have an equal number of young and old males and females in each group okay so that would be random assignment with matching which is typically done now if we have a homogeneous sample let's say we have all 20 to 25 year old females then that's not going to be as big of an issue all right because we're gonna we're gonna be divvying up our subjects randomly into groups and we're not so worried about overloading one group with males for example because they don't exist in our study to begin with so this is matching is something that will happen as needed again it still involves random assignment but it's just ensuring that we don't have groups that are overloaded with one particular type of subject all right so uh going back to uh the breastfeeding academic achievement example here is an example of running that study as an experiment so after doing random assignment we we're going to make sure that we have an equal number of high and low socioeconomic status individuals and groups and high and low maybe IQ score something like that so Random assignment with matching and then we create these two conditions we have our experimental group and our control group our experimental group is going to receive the experimental treatment which in this case we're going to say is breastfeeding the control group is going to be formula fed and then at age eight we are going to give those subjects those kids at this point an intelligence test or some standardized test and see if our manipulation had any any effect okay does breast milk versus Formula have some impact on later achievement later intelligence scores in this case I already told you guys it does that's published work but this is how we might go about running that as an experiment okay so I know I'm repeating myself here but this is important to really get straight and you'll see these types of questions on the exam so you really need to understand if you read a blurb or think about the design of a particular experimental study can you point out the independent variable and the dependent variable and there could be more than one independent variable there could be more than one dependent variable but you have to be able to determine what they are so again I'll remind you the independent variable is the factor that or the variable that is being directly manipulated by the researchers okay we have control of this so the researchers in the previous example the breastfeeding study determined what was going to be a group assignment either to a breastfed group or a formula fed group this assignment was independent of the subjects the subjects didn't their behavior had nothing to do with the assignment of that variable they were in one group of the other um and it did not depend on their behavior or the outcome in any way so manipulated variable is the independent variable or variables could be men the dependent variable then is this is what is being measured and in the breastfeeding example the outcome what's being measured is oops that's an error sorry about that but it's intelligence it was an intelligence test at age eight sorry about that so not six but same same idea so it's an intelligence test some years down the road from the experimental manipulation so this would be kind of a long-term study looking at subject performance uh several years after the actual manipulation but nevertheless hopefully you can see the independent variable the distinction between the independent variable and the dependent variable in that example I already mentioned this previously I've given you guys an example um in class or in the previous if you're in class for Tuesday but the previous audio lecture so I had given the example of sleep and memory the sleep of a negative impact on memory and in that case sleep was being manipulated by the researchers we brought subjects into the lab put them in one group that got a lot of sleep eight hours whatever versus another group um randomly assigned to receive I think I said two hours of sleep so there you go and then what was being measured in that example was the again the the memory performance okay so control group again I know I've already worked through it but control group is going to be the group that is not receiving the experimental treatment best way to think about that is a drug study because you can immediately say right drug or Placebo but the group that's not receiving the manipulation that the researchers think is going to have some some impact it's going to produce some measurable measurable outcome a distinct outcome when you're talking about the dependent variable so a control group we have an experimental group that's going to get the treatment again we could have several experimental groups if you think about a drug trial example again you could have a group that's getting 10 milligrams of a drug another group that's getting 20 30 and so on so you could have several different experimental groups okay so keep that in mind I'm giving you guys pretty straightforward examples where you just have two groups you can think of circumstances where you might have several okay be careful with your um your group assignment and setup remember make sure there's random assignment but also with matching as needed okay so importantly when you do an experiment you want to get rid of as many or or account for as many extraneous variables and kind of reduce their impact you only want to be looking at the effect of your independent variable so that's where things like random assignment and matching are going to be helpful because again I'll give the same example if you think about we want to study the differences in let's say memory Performance Based on level sleep we do not want to end up in a situation where our group getting eight hours of sleep is all people over the age of 40 and our group that's getting in the two hours of sleep is all people under the age of 40. so we don't want to have that age be a variable in our study we only want to be studying the effects of sleep okay here's a look at our table again I've walked through now all of these different types of methods and you should be obviously you need to be familiar with all of them for the exam the exam is going to have a lot of questions about these different types of methods so make sure you understand the differences the strengths and weaknesses of these different types of psychological research methods again I'll highlight overwhelmingly experimental methods are going to be the best in terms of determining causal factors but of course it's not always um ethical let's say for example or possible or feasible to do an experiment let's say it's too expensive you need equipment or something to run it so it's easier to do a correlation or it's not ethical because the manipulation would be psychologically harmful potentially something like that so experiments when possible are going to be the best method but they're not always possible these other methods cannot tell us causal relationships experimental studies can and there's the power something you need to be familiar with when you're thinking about experimental design or how just measurement in general when we're doing a study we need to be able to actually Define so that we can measure our outcome variables so let's say we're interested in this would be an example for me so in order for me to determine whether or not you guys have learned in this class I have to come up with what's called an operational definition of learning learning is a psychological construct it could mean many things a lot of things could demonstrate learning I could measure it in many different ways but I have to come up with a systematic objective consistent way to measure learning in this class I've just given one example so let's say I measure learning in this class or at least at least for the final unit of this class by giving you a final exam and let's say I Define quote unquote learning as being a performance of it's arbitrary right now but uh 80 or better in the final exam okay I Define that as good learning or sufficient learning but importantly I had to come up with a measurable definition okay this might be um uh easier to think about if you think about something that's more difficult like operationalizing sleep in Kim's experiment that might seem pretty straightforward actually but if you think about it how do we systematically determine the amount of sleep between different individuals uh in an objective way all right some people snore hey if they're snoring they're asleep uh some people are in REM sleep so their eyes are moving are they really asleep or are their eyes just moving well they're probably in REM sleep but how do we know um are they easily aroused from sleep or are they in deep sleep so in order to out the scene it's more complicated than it might seem at first so how do we operationally Define sleep maybe we use an EEG where you strap electrodes to their head whether while they're in the lab and you measure brain activity we have a pretty good handle on different stages of sleep and what they look like on an EEG readout so that's probably going to be the best way but you might not be able to afford an EEG machine so what are you going to do so these are things you have to think about how do I measure sleep another example let's say we were going to study aggressive behavior in children aggressive behavior in first graders I don't know or maybe daycare uh kids in a date specific daycare for example how are we going to Define aggression we have to measure it right we have to say all right this group is acting more aggressively or showing more aggressive behavior than the control group or or something like that or the experience vice versa the experimental group how do you define it well maybe we count the number of outbursts but how do you define Outburst so maybe we count the a number of physical attacks pushing pinching scratching biting things like that maybe we think that certain things like isolation is an aggressive behavior in some way so a kid that's just off in the corner and and not interacting with the group do we Define that as being aggressive uh maybe maybe not do we Define yelling screaming taking people's toys taking other kids toys so you can see there's a lot that goes into figuring out how we're going to systematically Define and measure our dependent variables our outcomes and that's an always an important thing to keep in mind the fact you shouldn't really jump into doing an experiment this should be obvious but you shouldn't jump into doing an experiment or any method really until you know that you have a systematic way to define and measure uh whatever is your your measure uh love how would you operationally define love and measure it give a survey how would you do that so things to think about when you're using any any of any method here's an example of operational definitions that's kind of personal so my wife and I we were gonna we ended up moving but when we were initially looking to move we were looking at school districts so we had to kind of figure out all right where do we want it was Maggie our daughter where do we want Maggie to start school and so I was looking at different elementary schools just in different districts and I came across a website of an elementary school that gave operational definitions of lots of things but I'm just giving you a couple of examples so some of these are goofy and straightforward but I hope you get the idea so the school defined tardy they had to figure out how do we measure and Define tardy well you know pretty pretty straightforward tardy is not in your seat when the bell rings great uh how are we going to measure and Define operationally fine reading comprehension So reading comprehension means that Maggie understands what she is reading as evidenced by blank out of whatever correct responses on questions about the text good so there is a systematic objective it's fairly straightforward definition of what we mean by reading comprehension and how we measure computation means that Maggie is able to add subtract multiply or divide using numbers to obtain a final correct answer straightforward but simplistic and and well-defined writing legibly means that Maggie writes letters within words this strips me up every semester and I use this but uh and I never remember why it worked within words that are recognizable in isolation that the words are recognizable in isolation oh okay that does make sense letters within words that are recognizable and words that are recognizable okay so anyway so sorry about that but writing legibly this is our definition of writing legibly good so here are examples of operational definitions of things that are relevant for young kids in elementary school there are different types of ways to design experiments lots of different types of ways to design experiments and this can be this is a whole class in and of itself okay so if you guys go on and take experimental psychology you'll learn a lot more about this type of thing but I just want to introduce you to to it here in not too much detail but just to give you some idea of different types of ways we could go about trying to run an experimental study so there are designs called within subjects designs or each subject in the experiment is going to be exposed to every level of the independent variable or treatment condition okay so this is powerful in some ways because it allows the subjects to be their own control group all right we're going to study them at time point x and then time point Y after some manipulation and we're comparing their behavior to their own behavior so thus in doing that we eliminate a lot of extraneous variables right there's not going to be any between subjects variables age difference isn't going to matter it's the same subject or sex or you get the idea so it's the same subject we're comparing them to themselves so that can be powerful as a design if it's possible to run it this way okay you could think of cases where it might be in Practical you might not be able to use the same subject more than once it might be a longitudinal type study where you're studying different populate or different samples within a population at different time points it might be a drug study where you can't keep using the same drug you don't know if it's safe so there are lots of reasons to use or not use within subjects designs but they can be powerful if possible to use them a between subject design is different so in this case each subject is only exposed to one condition and we are comparing subjects in different conditions to each other not to themselves with two different subjects okay so our sleep and sleep in memory uh example those were two different groups of subjects we compared their performance to each other okay we didn't use the same subject sleep deprived them one time and then give them adequate sleep and compare their behavior to their own behavior we chose different groups of subjects a mixed model design is going to use elements of both of these within subjects and between subjects excuse me I'm about to give you an example that example of a mixed model design and in choosing the mixed model I think it'll give you a better understanding of the between and within subjects designs as well so um Bill wants to test the effects of drug X on blood pressure in males and females with hypertension so they have pre-existing condition of high blood pressure already and we're looking at whether or not um some experimental new drug is going to help lower their blood pressure so the groups were matched for uh for pressure at the beginning of the study blood pressure at the beginning of the study for age body weight other risk factors maybe smoking or drug use things like that at the start of the experiment okay uh the Baseline blood pressure was taken over a three-month period and then both groups were exposed to a three-month treatment with drug X so what's going to happen I want you guys to think about pause this think about it in good detail in fact I'm not going to tell you the answer I want you to think about it what's the dependent variable in our study and what are the independent variables are there just one is there more than one and what are they okay now I'm going to show you the results which I just made up the results of this fictitious study are as follows so here is blood pressure on the y-axis and we have no drug and after the three three month exposure to drug X males are shown in blue females are shown in Red so think about how you would describe what happened in this study take a moment and do that okay hopefully you you want to pause this or think about independent variable or variables dependent variable and what happened what are the results so in this made up study it looks like we have we have equivalent systolic pressure to begin our study okay between males and females after three months on the drug we seem to be having a beneficial effect only in females okay hopefully everybody sees that so breaking this down in terms of how we designed this and the comparisons we can make between our groups in the study we have within subjects comparisons we're comparing the male subjects to themselves bid the drug after three-month exposure have any impact on their systolic pressure and the answer is maybe a tiny bit but doesn't seem to be very much that's a within subjects comparison same males getting a baseline period with no drug and then exposure to the drug and we take our dependent variable readings after three months but we do seem to have a beneficial effect in the females you see a pretty dramatic reduction in blood pressure after three month exposure to the drug we can still do the same within subjects comparison with the females and we see an effect but we can also compare make comparisons between males and females so we have a between subjects comparison here as well at the end of our study what is the systolic pressure of males versus females okay so this is a mixed model study it has within subjects and between subjects comparisons finally when talking about psychological research studies you guys will be encountering this to some degree although with the nature of things at present and covid and all that you're probably not going to be going into any Laboratories to do studies typically many semesters you can but in terms of ethical considerations and psychological research there is what's called an IRB institutional review board that is a panel of typically faculty administrators at every University or any every University that does human research that will determine if a study is Humane ethical or and will meet certain requirements so here are some considerations when you're thinking about ethics in a psychological research study and you guys should think about this as you are engaging in Sona and and doing and being a subject in certain studies so you guys should be able to give informed consent you must participate voluntarily and this irks me a little bit because you have to do Sona for a grade this is one issue I have with Sona but that's another story so uh you should be able to choose to participate or not and that is true you guys can join a study or not you have to join some study but it doesn't have to be study X or study why you get to pick um what you will be a participant in uh you cannot be you cannot be coerced into research participation again kind of boo because you have it for a grade but nevertheless you are not being coerced into participating in any specific Direction and this is true if if we're doing studies that are not for credit for a class individuals people students whoever may not be coerced into participating in research uh deception this is an issue historically deception used to be A-Okay before there are all these regulations and it led to some pretty bad outcomes which we'll talk about later in the semester but can't be deceived in most cases if the deception is determined to be something that might have the potential to cause psychological harm or physical or any sort of harm deception can be used but at the end of us out of the study deception can be used from the beginning as you're participating but at the end of the study you have to be told debrief you have to be told what the study was about and your role in it uh your personal information is confidential check that yep it will be um or the study will not be approved and when you're writing these protocols to run a study and and submitting them to the IRB you have to state in great detail how you Ensure all of these things are true I already kind of mentioned information about the study that you're in must be provided to you at some point so like I said deception can be used if it's not going to cause harm but at some point you will be told about the point of the study and your role within it