Statistical Analysis (02:49)
From Title: Psychology Research in Context
Using memory tests, psychologists must choose an appropriate statistical test to determine if their data results have statistical significance.
Item Number: 40117
Date Added: 01/07/2010
Online Classroom Ltd.
Filed Under:Introduction to Psychology
Related Resources ⇧
Segments Transcript Related
Interactive Transcript Tutorial
– “Science” literally means “knowledge.” But it’s come to be associated with subjects like physics, chemistry, and biology. So what do we mean by science? How does science work? And is it true? Science is the systematic and logical pursuit of knowledge through specific methods. One of these is the hypothetical deductive method.
– The hypothetical deductive model is the way that we study science. So we start with observations about the world and develop theories about the way that the world might work. We test those theories, and we use the findings from our data to draw conclusions about our theories, modify those theories, and then start again.
– We’re going to illustrate how this approach works in psychology by looking at one of Piaget’s experiments. As a young man working with children, Piaget noticed that very young ones consistently gave wrong answers to certain questions.
– Piaget was looking at the way that children developed, and he believe that children, as they got older, changed the way that they thought. So their cognitive processes changed with age.
– One of the ways Piaget assessed children’s thinking was to present them with what he called conservation tasks. That is the ability to recognize that quantity doesn’t change even when display is altered. Children were shown counters arranged in the same way. The experimenter then spread one of the rows out. Most young children thought there were now more bricks in the longer row. For Piaget, this showed that young children didn’t yet have the ability to conserve number. And this was consistent with his theory that children’s cognitive abilities developed over time. So are Piaget’s findings true? Well, yes and no. Yes, because he showed objectively that children’s thought patterns develop in distinct stages. But no, because all scientific findings are temporary. They’re always open to question, evaluation, and criticism. And this is just what happened in Piaget’s conservation experiment. It was challenged by McGarrigle and Donaldson.
– McGarrigle and Donaldson felt that children actually learned these cognitive skills much earlier than Piaget thought from his experiments. They’d watched real children and felt that they could solve problems at a much younger age. McGarrigle and Donaldson took Piaget’s experiments, but instead of carrying them out in a very abstract way, they tried to set a context for the children that would help them to understand the experiment and to understand why they were having to estimate the numbers of counters. McGarrigle and Donaldson introduced a character called Naughty Teddy. Naughty Teddy came and knocked the counters across the table, and the experimenters then rearranged them and asked the children whether or not there were still the same number of counters. In this experiment with the context provided, children as young as three or four could conserve number. This tells us that Piaget probably underestimated children’s cognitive development when they’re given a real-life context. Piaget wasn’t technically wrong. There was just still more to learn about children’s cognitive development.
– So Naughty Teddy showed us something about child development. But since then, research has moved on, and McGarrigle and Donaldson’s research is now being questioned and criticized. And the process will go on.
– Naughty Teddy taught us something about child development, but since then, our understanding has moved on even further. Each time we find something out, it generates new questions.
– The children we’ve looked at today won’t always be like this.
– And what do you want to be when you’re a big girl?
– A hairdresser.
– A hairdresser?
– They’ll develop and change the way they think about themselves and the world.
– I can’t decide.
– You can’t decide? There’s such a lot of jobs you can choose, isn’t there?
– When I do the dentist, I’m gonna– I’m gonna brush the people’s teeth. If–or–or if their teeth are– all of their teeth are falling out, I might put brace teeth in.
– And it’s the same with science. It’s the constant process of change and evolution.
– We find out things. We test our theories. And we get new data that tells us new information about the way that the world works.
– And so the hypothetical deductive model is based on observing events and developing or revising theories based on those observations, then devising hypotheses to test predictions from the theory, and finally carrying out research, relating observations to hypothesis. And so the cycle of a hypothetical deductive model will just keep on spinning. And the more we know, the more questions we’ll be asking. Statistics is the science of handling quantitative information. It’s concerned with how data should be collected, how they should be analyzed, and how we can draw conclusions from them. So what kind of statistics are useful in psychology, and what do they mean?
[mellow guitar music] Dan’s a partner in a small record company. He’s looking to sign a new band. One of the band’s he’s thinking about is Damn Fine Future.
– [singing] There goes my hero
– Have you seen these guys, like, around Blakeney and stuff?
– Oh, yeah, we went to see them at the–
– Yeah, they are quite popular out there.
– Cool, cool. – Really popular band. So it’s class to go watch.
– But he’s also been looking at Stained Glass Alice. Trouble is, he can only sign one of them.
– Well, we’ve seen them a couple of times. We all think they’re fabulous. Just something a bit new and different, aren’t they?
– Have you got a demo or two, anything like that?
– We’ve got two, actually. – Have you?
– So which one is likely to sell more records?
– Well, obviously, you can go by what you hear, but this can be very subjective. People like lots of different music. And it can be difficult to work out exactly what will sell.
– They’re the only band around here that have got, like, about eight–
– It makes it– it looks really class having them all singing and doing different things. And it’s the fact they all look really into it.
– It would be good to have some more objective evidence, for example, how many people go to see them and how regular are gigs?
– So can descriptive statistics help here, going back and looking at the band’s concert figures? One way to do this is look at measures of central tendency– the mean, the statistical average; the mode, the most frequent; and the median, the middle value– of all the people who have attended the band’s last five concerts. So let’s look at their concert attendances. Damn Fine Future have been seen by an average of 41.8 people. However, the mean average for Stained Glass Alice is much the same. But this doesn’t help Dan. So is there another way of using descriptive statistics that might help? We need to know whether the bands consistently draw audiences of the same size or whether there might be some sort of chance event that’s contributed to their average. A useful way of thinking about this is to look at measures of dispersion, or variations in the data. The usual way of doing this is to look at the standard deviation. A big standard deviation means there’s lots of variation in the data, whereas a small standard deviation means the data are quite consistent. This could be plotted in a bar chart to show the spread of scores around the mean. From the objective measures of the band’s concert figures, which one would you advice Dan to choose?
[upbeat rock music] But how do these techniques work in psychology? Educational psychologist Rhian Humphreys is using IQ tests to see if she can identify children who might need extra support with learning at school.
– Some children with dyslexia, for example, have problems with reading and writing, amongst other things, which can affect their ability to learn. But they are also still very intelligent. If reading and writing isn’t consistent with intelligence, then specialized support is needed to help these children to learn effectively.
– IQ tests were developed specifically to help psychologists and teachers understand why some children have difficulties learning and what sort of difficulties these are.
– Well, IQ tests have been designed so that an average person at a specific age will always score an IQ of 100. So, for example, the average IQ of a seven-year-old is 100, and the average IQ of a 20-year-old is also 100.
– So the mean IQ for the entire population is 100. The standard deviation is 15. This means that about 68% of people have an IQ that’s within 15 points of the mean, either above or below. So 2/3 of the population have an IQ between 85 and 115. Ninety-five percent of people have an IQ within two standard deviations of the mean, or within 30 IQ points. So nearly everyone’s IQ falls within the range of 70 to 130. We can categorize people’s intelligence using this knowledge. This kind of information can help Rhian in her work as an educational psychologist. For example, if a child has problems with their schoolwork, she can use the information about their IQ to work out whether the problem has to do with their intelligence or with a more specific learning difficulty like dyslexia.
– For example, a child with an IQ of, say, 95, this is within one standard deviation of the mean. This means that they have a pretty average intelligence and that their problems are probably due to a specific learning difficulty. However, a child with an IQ of, say, 70, this is two standard deviations below the mean. So this means that the help that they need will be different. This can be helpful in a way that a signpost is helpful when you’re looking for somewhere. It can help point you in the direction that you need to take.
– So we’ve seen here how using measures of central tendency such as dispersion and standard deviations can be useful in psychology and its applications. But they can also be useful in everyday life. Finding the right partner’s not easy. You’ve got to click, got to be right for each other. It’s a bit the same with psychological data. Once you’ve collected your data, you need to find the right graphical partner to display your results clearly. So in this program, we’ll be looking at ways of presenting research data in the right graphical form. It seems our patterns of dating, our ways of finding partners, may be changing. It’s been estimated that 50% of single adults in Britain have used Lonely Hearts ads or internet dating. Caitlin McLeod is researching this.
– Well, I’m looking at who’s putting these ads in, their ages, when they’re doing it, what they’re looking for, and what they’re offering.
– As Caitlin is collecting different types of data, she needs to select the best way of presenting it graphically.
– The first thing I looked at is who wants a date with whom. Well, for this type of data, I would use a pie chart, because it’s a really good way of displaying how a population can be divided into sections and then what proportion of the whole each section represents.
– In Caitlin’s sample, the biggest proportion was women looking for men at 47%, next men looking for women, 42%, then men looking for men, 7%, and women looking for women, 4%. Data needs to be in the form of frequencies, in other words, nominal. The bigger the proportion, the bigger the slice of the pie. A second question is when people put ads in. Is it pretty much consistent throughout the year, or are the peak times in seasonal variations?
– Well, I’m finding there are some seasonal variations. Numbers increase through Autumn and reach a peak around Christmastime and into January, you know, perhaps people writing New Year’s resolutions, “Must find partner.” And then the numbers decline and pick up again in the spring and early summer. For this data, I would use a line graph.
– Line graphs are used to show a trend over time or how a participant’s experiences change. The X-axis on the line graph must always use continuous units of measurement, for example, changes over time.
– A line graph is the best way to display this type of data about seasonal changes in the number of ads placed because the data’s continuous.
– Caitlin is also looking at the age of people who put in Lonely Heart ads.
– There’s a stereotype view that it’s only middle-aged and older people who put in these type of Lonely Hearts ads. But I’m finding that there’s a whole range of ages. My data actually shows that ages range from 19 to 87. what is interesting is looking at the ages of the person and the age of the partner they’re seeking.
– Here as Caitlin is looking at the relationship between variables, the age of the people placing the ads and the age of the partner they’re looking for, this can be presented with a scattergraph. The values are plotted on the graph and a line of best fit calculated so any relationship can be clearly seen. So you can see there’s a strong correlation with advertisers looking for potential partners around their own age.
– I’m also looking at the ads themselves to see what people want and what they’re offering, effectively what they’re buying and what they’re selling. And there are some interesting gender differences.
– I look for someone who’s faithful and who’s nice and who has a nice personality.
– She’s got to have long blonde hair.
– Good looks, brown hair, you know, nice personality.
– Somebody that can make me laugh and smile. I don’t really, like, look for features.
– Men tend to prioritize attractiveness and social skills and are generally looking for someone younger than themselves. Women place a lot more value on commitment and resources and are generally looking for someone older.
– And while males tended to stress economic resources as their main selling point, females stressed their attractiveness. This kind of information which shows frequencies and discrete data, such as males and females, can be presented in a bar chart. A bar chart is a diagram consisting of columns, the heights of which indicate frequencies. On the X-axis are discrete data. So what’s Caitlin been finding out from her research?
– Well, all of this shows us that although finding a mate through ads is relatively new, basic evolutionary processes are at work, the male offering protection, the female, desirability.
– Here we’ve been looking at presenting statistical data. Caitlin’s been collecting different types of data, which she’s presented in different ways. And in data presentation, like real life, the trick is getting the right match. When psychologists get data from lots of participants, it can be difficult to get a clear picture of exactly what’s happening. We need to be sure of what type of data we have and whether our results are statistically significant. And that’s what we’ll be looking at here.
– I remember some of the problems that I had with statistical analysis when I was a student, so I’m gonna try and make it a bit clearer here. I’m gonna ask these students to help me with some short tests on memory. I’m gonna use these to do two things. First, we’ll look at the types of statistical data that psychologists collect. And second, we’ll use an example to find out if we have statistical significance. Okay, who thinks they’ve got a good memory? Okay, if you’ll just go stand by that sign, please. And what about average memories? Who thinks their memories are just average? Okay, if you’ll just go stand over there, please. Thank you. And bad? You think your memories are just bad? Okay, if you’ll just go stand over there. Thank you.
– Putting people into categories like this is what psychologists call nominal data.
– Now what I’m gonna do is see what your memories are really like.
– Students are shown a tray with a number of items on it, and they’re given one minute to memorize as many as they can.
– Okay, minute’s up. Just gonna cover the tray.
– They’re given another minute to write down as many as they can remember.
– Right, here’s what was on the tray. If you’d like to work out your scores, we’ve got an umbrella, a stapler, a Kellogg’s Special K bar, a banana, a cassette… What we’re gonna do here is put the students into rank order. We’re not looking at individual scores here. We’re just looking at the order.
– In rank order, the person who got the highest score is ranked number one, next highest score number two, and so on until we get to the lowest score having a rank value of ten. Putting people or scores into order, lowest to highest or highest to lowest, is called ordinal data. But we don’t know how much higher or lower. For example, there might not be the same amount of difference between one and two as there is between four and five.
– Right, now what I’m going to do is get a different type of data by asking students for their actual results. Okay, let’s see your scores, please. How many did you get?
– This is what psychologists call interval data. We have real scores. And the size of each step on the scale is identical. The difference between one and two is the same as the difference between four and five, one remembered object. If possible, psychologists always try to collect this type of data because it’s the most meaningful. Now we know the difference between nominal, ordinal, and interval data, we can look at an example of how they’re used by psychologists to choose an appropriate statistical test.
– Okay, what we want to do now is see what your memory’s really like.
– We’re going to illustrate this looking at how we can use statistics to see whether boys or girls score differently on these memory tests. So this is an example involving interval data. But the same idea would be applied if we were using other types of data.
– Okay, time’s up. I’m just gonna cover the tray.
– The experimental hypothesis is that sex affects memory. So we expect to see that boys and girls have different scores. The null hypothesis is that sex does not affect memory, in which case, we’d see no difference between scores for boys and girls. So how do we decide which test to use?
– Well, let’s just recap and see what we have first. The experiment is a test of difference, a difference between boys and girls. The experiments were designed as independent measures because we’re comparing boys’ scores with girls’ scores. And the data are interval data, as each participant has a score representing the number of items that they correctly recalled. So for this experiment, we use a Mann-Whitney U Test.
– The result shows that U equals 22 and that P is less than 0.05, so we can reject the null hypothesis. We’re 95% sure that sex really does affect memory. Because the mean score for girls is higher, we know that girls tend to have better memories. It won’t be true for all boys or girls because the data vary within each group. But on average, we can expect this to be true. And our Mann-Whitney U Test has told us we can be 95% sure the difference between boys’ and girls’ memories is because of their sex, not because of chance.
– So here we’ve looked at different types of data and had to choose the right kind of test to see whether we have statistically significant results. Examiners will expect you to know about which test to choose and the meaning of P values, so good luck.
– Collecting and analyzing data is only part of the research process. Data also have to be interpreted and evaluated. And that’s what we’ll be looking at here with research student Tanya Wells. Psychologists are interested in studying aggression. But what is aggression?
– I think it’s more like anger, and there’s a lot of physical and verbal. Like, you can start raising your voice and get aggressive when you’re infuriated or frustrated by something.
– Like, if someone gets really angry, often you can over–get– like, lash out on someone. That’d be aggression. Or you can sort of bottle it up and just, like, walk away.
– People have different buttons and different, like, pressure points in where they can get angry and where they can’t get angry.
– But are there differences in the way that males and females express aggression?
– Mainly physical with boys. Girls just tend to, I think, be very two-faced and hiding their aggression.
– Boys tend to use physical. Girls tend to be more calm when it comes to being aggressive.
– And who’s more aggressive, boys or girls?
– Normally, people think boys, but I think girls get aggressive but in different ways. They show it in different ways.
– Research student Tanya is trying to find out.
– Well, I’m researching aggression. One of the things I’m looking at are gender differences in self-reported measures of aggression. And for this part of the research, I’m using questionnaire methods. Well, I have a number of statements to operationalize the concept of aggression. So respondents are asked whether they agree or disagree to a statement such as: “If anyone insults me or my family, they’re asking for a fight.”
– “If I’m arguing with someone, I tend to raise my voice.” Agree.
– “If someone hits me first, I usually hit them back.” Agree.
– “When people shout at me, I shout back.” Agree.
– So what’s Tanya been finding?
– Well, my findings show that on average, males did score higher on aggression in the questionnaire than females, as you can see from here.
– The standard deviation in Tanya’s data shows approximately the same degree of variation in each set of scores. However, for both sets of scores, the amount of dispersion was fairly large. What this means is, there was considerable individual variation in both male and female scores. So Tanya’s findings have shown higher levels of reported aggression in boys. But how can she be sure these differences are really due to sex and not just chance?
– Tanya needs to be 95% confident that the difference that she’s seeing is due to sex and not due to chance. She’s going to test this using a Mann-Whitney U Test. This is a test of difference between two groups, an independent measures test for data that’s at least ordinal, like our questionnaire scores.
– After calculation, U equals 5, and P is less than 0.05, so we can be 95% sure that males did scores higher on aggression than females. But there’s another question we have to ask in data interpretation: how confident can we be of the data themselves? And here psychologists have certain key data to help them. One of the most important is validity.
– Validity means that we want to check that something measures what we really think it measures. So in the context of Tanya’s questionnaire, does it really measure aggression, or does it measure something else that might be related to aggression but not quite the same thing?
– So can the validity of Tanya’s methods be questioned? Her supervisor’s interested in her measures of aggression.
– In some cultures, of course, bad language won’t be used very much at all. It might only be used in extreme circumstances. In other cultures, you might find it as a part of everyday language.
– People use, like, bad language in everyday situations.
– ‘Cause a lot of people just swear, and they don’t even realize half the time, whereas when you’re angry, you do it, and you know you’re doing it.
– But swearing shouldn’t be used as a sign of aggression.
– I’d swear, like, in front of my friends and stuff like– in general conversation, but then as soon as I go home, I wouldn’t swear in front of my parents.
– Are we really measuring aggression? If a students says that they don’t– they only use bad language when they’re very angry and another student says, “I use bad language all the time,” we’re not really necessarily measuring aggression. We’re measuring cultural norms and whether it’s okay in that culture to use that kind of language.
– Yeah, yeah.
– Reliability is another important criterion.
– Reliability means that the test should give the same results if it’s repeated with the same people or with the same population of people in different places or at different times.
– I think you’re right. Like, I think it would be good to take the questionnaire to, you know, obviously other schools in completely different areas so that, you know, you’re even working with kids who are, you know, of different backgrounds to see how aggression works with them.
– Absolutely, because you might find that children from different cultures display aggression in different ways. In Tanya’s study, we would expect her to find the same results, the pattern of boys being more aggressive than girls, whether she looked at private schools or state schools in different parts of the country or even in different parts of the world.
– And if the results are very different, then maybe the data aren’t reliable. Maybe they need reinterpreting. Or maybe the methods need refining.
– So I think to have maybe five points on your scale rather than just the two will give you a more sensitive measure of the level of aggression that each person…
– But generally, I think you can make a comparison between science and sport. You may be really good at sport, but someone may take your championship, beat your score. And it’s the same with science: someone’s going to find a better methodology or a better explanation, and that’s the way it progresses.
– We saw this in the first program with Piaget’s conservation experiment. It told us a lot, but McGarrigle and Donaldson raised questions about its validity with Naughty Teddy, and they came up with different results. And that’s the way science works: by building on what’s there and trying to push it that bit further.
132 West 31st Street
New York, NY 10001
Tech Support: 1-800-322-8755 – Press (6)
Report a Problem
Films Media Group, Films for the Humanities & Sciences, Cambridge Educational, Meridian Education,
Shopware and their respective logos are trademarks of Films Media Group.
© Films Media Group. All rights reserved.
FILMS ON DEMAND