Psychologists try to develop theories about behaviour. A psychological theory is simply an explanation, based on objective evidence, for some aspect of behaviour. But how are theories to be developed?
The psychologist may observe some particular behaviour that leads him or her to ask questions about its cause. This leads to the formulation of a hypothesis (a testable statement) about the behaviour. This hypothesis is then tested, using an appropriate psychological method. The results of testing are a record of what happens during the test.
The psychologist will then interpret the results, reaching a conclusion about whether or not the hypothesis has been proved. Depending on the results of the study, he or she may need to refine the original hypothesis and repeat the procedure again until satisfied with the conclusions.
A psychologist, Albert Bandura, noticed that children often imitate adult behaviour. He
wanted to find out whether children would imitate an adult’s aggressive
behaviour.
He formulated an hypothesis that children who see an adult commit an act
of physical aggression are more likely to commit acts of physical aggression
than those who see an adult behaving non-aggressively. Nursery school children
were divided into two groups. Just one group watched a film of a man smacking
and hitting a large inflatable toy (a Bobo doll). All the children were then
placed, one at a time, in a playroom and, after a short while were made
frustrated by having the toys they were playing with taken away. Bandura
watched their behaviour towards the Bobo doll, which was left with them. There
was much more aggression in the group which had watched the aggressive adult
than in the group which had not. He found that after watching the adult, both
boys and girls behaved aggressively towards the doll, performing many of the
detailed acts of aggression that the adult had displayed.
!SAQ 1
a) What hypothesis was Bandura testing?
b) What behaviour is being observed?
c) What conclusion did Bandura reach about his hypothesis?
d) Explain why the psychologist’s approach to his study is scientific.
Bandura’s study is an example of how an experiment can be used to test a hypothesis. There are also a number of types of research study in use in psychology.
Developmental psychologists are mainly interested in how children’s behaviour changes as they develop or grow older. One way to investigate this is by carrying out a series of observations of the same individuals over a period of time. The disadvantage of longitudinal studies is that we have to wait for the individuals to develop, so our research may take a long time, and we may lose track of them or they might move away.
Alternatively, we can study how behaviour changes as the individual gets older by taking a representative group or sample of children of the same age, observe their behaviour, and compare it with the behaviour of another representative sample of older children. This is known as a cross-sectional study because we take a representative cross-section of children from each age group and compare their behaviour. Cross-sectional studies are much cheaper and quicker to carry out than longitudinal studies.
The main problem with cross-sectional studies is that the groups contain different people who may not be comparable. Suppose we compare the intelligence test scores of young adults with a group of over 65s. We might find that the older people do less well. But we can’t conclude from this that intelligence declines with age. The younger adults may have had better (or at least longer) schooling than the older group. It also won’t be so long since they left school!
In the
& Reading: Types of Research Methods
!SAQ 2
List (in a table, if you prefer) one advantage and one disadvantage of the longitudinal and cross sectional types of research discussed above.
In the last section we said that psychology is concerned with the study of behaviour, and we made the point that psychologists should use objective scientific methods to study observable behaviour. This section looks at some of the ways psychologists try to achieve this.
Psychologists study their participants by observing them. Scientific observation involves trying to record behaviour in as unbiased a way as possible. There are several different types of observation, each with its own advantages and disadvantages:
Naturalistic observation. This involves observing people (or animals) in their natural environment. Their behaviour should be true-to-life, although it is always possible that the presence of the observer could influence their behaviour.
Controlled observation: Naturalistic observation may be more true-to-life, but there are many influences on the behaviour of the people being observed which are not controlled. If this is seen as a particularly important potential problem, psychologists will take the observation into a controlled environment, perhaps even a laboratory. Controlled observation may also take place during an experiment. For example, in the Bandura study discussed above, children were observed to see how they would behave toward the Bobo-doll. Controlled observation can be more reliable, but the behaviour under study may not be altogether natural. It may be distorted by the fact that it takes place in a different setting than usual.
Participant observation: This is a form of observation in which the investigator becomes a part of the group or setting that is being studied. Those being studied may not even be aware of the fact that they are being studied at all. Festinger conducted a well-known example of participant observation. Along with his co-workers, he infiltrated a cult group who predicted that their city would be destroyed by flood on December 21st. They were interested in what the group’s reaction would be when the prophecy turned out to be incorrect. It would be fun to report that the researchers drowned while taking notes, but, of course, they did not! With participant observation, you can get very realistic behaviour. The problem, however, is that by taking part, you may influence the behaviour of the people you are observing.
&
!SAQ 3
a) Which method of observation would you choose to study children’s play? Which method could be used to study the behaviour of adolescents in a gang?
b) What is observer bias and why is it important?
c) Give one advantage and one disadvantage of participant observation.
Surveys are widely used by social scientists. Opinion polls about voting and contemporary issues are a good example. Psychologists also study behaviour by using interviews and questionnaires. For obvious reasons, written questionnaires aren’t very useful with young children, but they come into their own with adults, particularly when you need to get information from a large number of people. The problem with questionnaires is that you can’t be sure that you are getting the truth, because people might be afraid or even ashamed to say what they really think. Interviews can be more in-depth than questionnaires. Both interviews and questionnaires have the problem of leading (or biased) questions. Here is an example:
Q. If you think carefully about the issue, it is
not really sensible to vote Liberal Democrat in the next election since that
party is very unlikely to be able to form a government.
A. Agree/disagree/don’t know.
üActivity 1
Rewrite the above question to make it less biased. Now, try to rewrite it so the bias is in the other direction (disagree). Look at examples of surveys to see whether you can find more examples of biased questions.
Find out what the advantages and disadvantages of yes/no (i.e. closed) questions are compared with open-ended questions in surveys.
&
These are in-depth studies of one person or a small group of people, e.g., studies of members of gangs. They are often used, together with clinical interviews, to study individuals with psychological problems. Freud’s theories (see unit D1) were based largely on case studies.
&
People often claim that two things go together. For example, John Bowlby claimed that maternal deprivation and delinquency were related. This is known as correlation. Correlation studies show us the degree to which two things go together, e.g., food intake and weight gain or loss; smoking and a bad cough.
If two things go together in the same direction, such as amount of work done and your GCSE grade, they are said to be positively correlated — as one increases, so does the other. However, if two things move in opposite directions, e.g., amount of alcohol consumed and snooker score, they are negatively correlated. For many pairs of things there is no (or zero) correlation.
The actual degree of correlation can be worked out precisely. But it is not necessary to do this at GCSE level. However, you do need to know that the strongest correlation is given the number +1 (–1, if negative), the weakest (i.e., no correlation at all) is given 0. Note that it is a very common mistake for students to say that a correlation of –1 is no correlation. In fact, it is just as perfect a correlation as +1.
&
!SAQ 4
Are the following positively or negatively correlated? Give reasons for your answer:
a) Temperature and rainfall in this country.
b) Children’s age and shoe size.
Remember that correlation cannot show that one thing causes another. It may just be a coincidence.
üActivity 2
A researcher once pointed out that between the two World Wars there was a correlation between the divorce rate and the number of apples imported into the country. Coincidence or cause? What do you think?
Psychologists use a variety of methods to study behaviour: they have to decide whether to use cross-sectional, longitudinal or cross-cultural approaches, and then whether to use observation, experiment, survey, correlational or case study methods. It is only then that they can think about how to select their participants and how to design their investigation — topics we will look at in the following sections.
All psychological research needs people or animals to study. The researcher has to select a sample of participants from the target population. Researchers usually want to be able to generalise from their sample to the wider population. They can only do this, if their sample is representative. Sampling is concerned with the selection of representative participants.
In sampling, researchers need to identify:
1.
The target
population under study. For example, all GCSE students in schools in
2. The number of participants to study. Too few participants and our sample will be unrepresentative (biased). Too many participants will take lots more time and money to study than is needed. As a general rule, however, the more participants, the more representative our study will be.
3. How to select the sample group.
1. Random Sampling is where every member of the target population has an equal chance of being chosen for the sample. We can only use random sampling, if we can get a complete up-to-date list of every member of the target population. We could then put all the names into a hat (or box) and draw out the number of participants we have decided on. This is known as the lottery method, for obvious reasons. Or, we could use a computer to generate a list of random numbers. Suppose the first number is 77; we then select the 77th name on our list of the target population and so on.
In practice, it’s often impossible to get complete lists (or sampling frames) for our target population, so other methods are often used instead. Some examples of sampling frames are the electoral register for all adults living in a particular area, or all the registers for pupils in a school.
&
2. Multi-stage sampling is where we overcome the problem of sampling frames by leaving them until the last stage. Suppose we wished to study all GCSE Psychology students in British colleges. It is obviously impossible to get a list of them all — it would take much too long as there are about 500 Further Education colleges. We can use multi-stage sampling.
First Stage
(a) Obtain
a list of all Local Education Authorities (there are about 90) from the
library.
(b) Use
the lottery method to randomly select a small number of Local Education
Authorities (say six).
Second Stage
(a) Our
sample now contains the names of six Local Education Authorities. Go back to
the library and get the name of every Further Education college in each Local
Education Authority.
(b) For
each Local Education Authority, use the lottery method to pick out one college
from each.
Third Stage
(a) We now
have the names of, say, six Further Education colleges that have been chosen
randomly. We can now write to each of the six asking for a sampling frame of
GCSE Psychology students.
(b) Again,
use the lottery method to select the required number of participants (say
fifteen) from each college.
(c) We
eventually end up with the names of 90 students from six colleges. We can be
reasonably confident that such a sample will be representative and it’s much
easier to obtain our sample using the above procedure than to write to 500 or
so colleges.
3. Quota sampling is the preferred method where no sampling frames (i.e., complete lists of names) are available. It is widely used by market researchers and social scientists and has been shown to be reliable. Here, the target population is sub-divided into groups by factors that might affect the research. For example, a survey of GCSE Psychology students might divide them into sub-groups by age, sex and whether they are full or part-time students. The researcher then finds a quota of people in each such group. They might choose, for example, 10 female part-time students over 30 and 10 male full-time under 20. This will be done in such a way that the proportions of each sub-group are the same in the sample as they are in the population. Since all the different sub-groups of the target population are included, we can be reasonably sure that our sample is representative.
&
4. Self-selected sample is where people are asked to volunteer for some research, for example, through advertisements in the newspapers. Since the type of person who volunteers to take part in research is not typical of the population as a whole, we can’t really generalise from such self-selected samples.
5.
& Reading: ‘Discovering Psychology’ page 261 ‘
!SAQ 5
a) What is the difference between a population and a sample?
b) Explain the difference between a random and an opportunity sample.
In an experiment the psychologist deliberately changes some part of the situation and then observes and records any changes in the behaviour of the subjects (the participants who are being experimented on). For example, in the study referred to at the beginning of this unit, Bandura had children watching an adult behave violently towards a Bobo doll, and then tested them to see how aggressive they were. Anything that can change in an experiment is called a variable. There are two types of variables that it is important to understand. The variable that is manipulated by the experimenter is called the independent variable (IV). The variable that is measured is called the dependent variable (DV). In this example these are:
IV = whether children watch the adult behaving aggressively
DV = amount of aggressive behaviour observed later in the children
The purpose of the experiment is to test whether the changes in the DV are the result of changes in the IV. Of course, such an experiment would have to be carefully controlled, since there are lots of factors (or confounding variables) that could cause differences in aggression. Such factors could be the children’s backgrounds and ages, how long they watched the film, and so on. There are two classes of such variables to be controlled in an experiment:
¨ Subject variables: relating to individual differences in participants e.g., differences in age, temperament, intelligence etc.
¨ Situational variables: such as the temperature in the room where the experiment is carried out, distractions, the instructions to subjects.
The researcher needs to be sure that all variables other than the IV are held constant, so that they can be sure that any change observed in the DV is due solely to the change in the IV. This is more difficult than it might at first appear!
!SAQ 6
Explain what is meant by the term ‘confounding variable’ in an experiment. Give an example of one, and say whether it is a subject or a situational variable.
While an aim describes, in general terms, the research question in which we are interested, a hypothesis is a precise prediction about what we expect to happen in an experiment. It is a testable statement that predicts the effect of a given change in the IV on the DV. For example, (i) that drinking one glass of wine (IV) will increase a person’s reaction-time (DV) or (ii) that the average number of words in the sentences of seven-year old children will be greater than that of four-year old children.
!SAQ 7
Identify the IV and DV in the second example of a hypothesis given above.
Hypotheses must always be carefully worded in such a way that it is clear exactly what the IV and DV are, and what effect it is predicted the IV will have on the DV. The process of carefully specifying just how a variable is to be measured is known as operationalising variables. To do this, it is usually necessary to define the DV in such a way that it is measurable on some sort of scale (quantifiable). With reaction time this could be in terms of milliseconds. Some variables are not so easy to quantify; for example, Bandura had to devise a measure of aggression in his experiment on observing aggression.
üActivity 3
Imagine you had to observe children’s aggressive behaviour in a playground, comparing the amount shown by girls and boys. How would you define aggression in such a way that you could measure it?
In the example of the experiment by
Bandura we would probably find that some children showed aggression after
watching the adult, while others did not. Does this mean that the hypothesis is
proved or not? Bandura needed to predict how likely it was that any changes in
aggression (the DV) occurred by chance, rather than as a result of changes in
the IV. If chance was the likely cause of the changes in the DV, then he would
have accepted what is termed the null
hypothesis. In fact, he showed that the majority of children
who watched an adult being violent, were aggressive, but that few of the
children who did not watch the adult being violent, were also aggressive
towards the doll. He used statistics to
show that chance was very unlikely to have produced a difference as great as
the one he had observed.
!SAQ 8
Explain the difference between an hypothesis and a null hypothesis.
In any psychology study the fact that the participants are aware that they are being studied can affect their behaviour, so we need to make sure that any observed changes in the DV really are the result of changes in the IV, and not simply due to experimental effects. One way to do this is to divide participants into two groups, an experimental group where the DV is changed, and a control group where there is no change in the DV. In all other respects the groups must experience the same conditions, however, including being observed. This is what Bandura did. The group that did not see the film of the adult being violent, was his control group.
Medicines are usually tested in this way, with the control group receiving a dummy pill (placebo) that does not have any effect (but, of course, the control group does not know the medicines are placebos). Differences in the control and experimental groups can be assumed to be due to the medicine under test, provided the two groups are strictly comparable in the way that they are selected and treated during the study.
üActivity 4
We have described the control group in Bandura’s study as ‘the group who did not watch the film of the adult being violent towards the doll’. Consider whether this is a good idea. Should they watch no film at all, or is there a better way of conducting the study?
We need to make sure that any differences between control and experimental groups are really caused by the IV, so we need systematic ways of allocating participants to each group. There are three ways in which participants can be allocated:
This is the easiest design to understand. The participants are allocated to experimental and control groups randomly (as in Bandura’s study). However, there are problems with this design because, for instance, there may be a difference between the participants in the two groups. For example, we may end up with all the more timid children in one group. To overcome such problems, we can consider using the repeated measures design.
Here all participants go into both the control and the experimental groups, so that there are no differences between the two groups. However, the experience of being in one group may affect behaviour in the other. Such effects are known as order effects and can be dealt with by the use of counterbalancing. In some studies, it is just not possible to use this type of design.
üActivity 5
Try to imagine how the Bandura study could be conducted using the repeated measures design. Would it work as well? Compare your ideas with others in the group.
This method lies somewhere between the previous two. Different participants are used, but they are paired together so that each pair is as alike as possible for any factors that could affect the IV. One member of each pair is then allocated randomly to each of the control and experimental groups. Some experiments have even been conducted using identical twins as the matched pairs.
&
!SAQ 9
Compile a table that shows one advantage and one disadvantage of each of the three experimental designs.
Experiments involve the testing of a hypothesis in a controlled environment in order to see if predicted changes in behaviour occur. The results of the experiment are carefully recorded, and analysed to see if the hypothesis is proven true or not. The conclusion states whether or not our original hypothesis was proven. The results from experiments are considered to be more generalisable than results from other studies using non-experimental methods. However, many topics in psychology can’t be investigated using experiments because, for example, they are concerned with long-term influences on behaviour, or they would involve procedures that are unethical.
Natural experiments are situations where the independent variable changes in the normal course of events. For example, an isolated community might get TV for the first time. In a natural experiment, the researcher does not manipulate the independent variable. It changes naturally.
A field experiment is conducted in the natural environment of the participant, rather than in the controlled environment of the laboratory. The naturalistic environment provides more realism, but less control of variables. Remember, however, that ‘good’ controlled experiments, like Bandura’s, try to make the environment as natural as possible. This is particularly important when testing children.
&
!SAQ 10
Imagine an experiment to test the hypothesis that children who are taught their multiplication tables perform better in maths SATs tests. How would you test this using:
(i) a natural experiment, and (ii) a field experiment?
It is vitally important that research results are presented in ways that make it easy to understand what has happened and what the conclusions are. It is not enough to present tables of raw data. Results have to be analysed and presented meaningfully.
The main ways of presenting results are tables, graphs, histograms and pie charts. Whichever method is chosen, you should use clear labelling, and colour or shading where appropriate. If you have access to a computer with a spreadsheet package, you may want to use it to improve the presentation of your data.
¨ Tables show data with great accuracy, but they lack visual impact.
¨ Graphs are useful to show trends over time.
¨ Histograms (and bar charts) show the number of times something occurs.
¨ Pie charts show data as a percentage of the whole.
Generally, you should try to show your results in either a graph, histogram or pie chart wherever possible. Appropriate presentation of results is particularly important in your coursework.
Raw data doesn’t tell us very much about our research. The point of analysing our results is to turn a potentially confusing mass of data into meaningful information about our aim and/or hypothesis. The most basic way to analyse our data is to state the range: the difference between the highest and lowest scores. Psychologists are also interested in the average score, but there are actually three ‘averages’ in common use: the mean, median and mode.
Before we can accept any research findings, we need to know some details about how the participants were selected, which experimental design and method was used, and how valid were the tests. We should be very careful about generalising from small samples, which can easily be unrepresentative.
¨ Investigators must consider the ethical implications and psychological consequences for the participants in their research.
¨ Investigators should inform participants of the objectives of the research and gain their informed consent. If this is not possible, special safeguarding procedures must be followed.
¨ Withholding of information or misleading participants is unacceptable, if they are likely to object, or show unease, once they are debriefed and become aware of the deception. Intentional deception should be avoided whenever possible.
¨ In all studies where the participants are aware that they have taken part in an investigation, they must be fully debriefed, so that he/she understands the nature of the research.
¨ Investigators must emphasise the subject’s right to withdraw from the experiment at any time, should he/she so wish.
¨ All data obtained must be treated as confidential, unless otherwise agreed in advance.
¨ Investigators must protect participants from physical and mental harm during, or arising from, investigations.
¨ Studies based on observation must respect the privacy and psychological well- being of the people studied. Covert observation without prior consent is only permissible in situations where those observed would expect to be observed by strangers.
¨ Investigators must not give advice on psychological problems, but they may refer people to someone who can offer appropriate professional help/advice (e.g., a counselling psychologist).
¨ Investigators share responsibility for the ethical treatment of their participants with associates, employers or students, all of whom should comply with these guidelines.
üActivity 6
a) Draw up a checklist that you can use, in order to make sure your proposed piece of coursework is acceptable and conforms to ethical guidelines.
b) Apply this checklist to the Bandura study we have been considering.