Aims: The aim of a study is what the purpose is of a piece of research. For example- to investigate if age affects memory.
Hypothesis: The prediction of what the results will be. This can be directional, in which the expected effect of a variable is stated, or non-directional, where a difference is predicted, but not the nature of the difference. For example:
- Directional: young people will do better in a memory test than older people
- Non-directional: there will be a difference between young and older people in performance on a memory test
The above hypotheses are known as the experimental/alternative hypothesis. There is also the null hypothesis, which predicts that there will be no difference between the conditions, for example ‘there will be no difference between young and older people in performance on a memory test’.
Directional hypotheses are used when previous research indicates what the results will be, for example if other studies have found that young people have better memories than older people, we can predict that the result will go the same way. If there is no previous research, or previous research is contradictory, and a non-directional hypothesis would be used.
Independent and dependent variables: In an experiment, the variable manipulated or changed by the researcher is the independent variable (IV). The result, which should be affected by the change in IV, is the dependent variable (DV). All other variables should be controlled as far as possible, so that it is the IV that affects the DV and nothing else. For the purposes of experiments, IVs and DVs must be operationalised- put into a form which is measurable. For example, ‘Age’ (the IV) could be operationalised as ‘participants between 20 and 25 years of age and participants between 60 and 65 years of age’. ‘Memory ability’ (the DV) could be operationalised as ‘the score on a test of memory’ or ‘the number of words successfully recalled’. This can be used in a hypothesis as follows:
‘Participants between 20 and 25 years of age will score more highly on a memory test than participants between 60 and 65 years of age’.
This is an operationalised hypothesis, and it is directional, in this case.
Variables & Control
As well as the IV and DV, other variables exist which potentially affect the results of experiments.
Extraneous variables: Variables other than the IV which may have an effect on the DV if not controlled for. For example, in the memory experiment, the intelligence and motivation levels of the participants may have an impact on their score on the memory test. The researcher should take steps to minimise the impact of these, for example by giving participants an IQ test beforehand and eliminating any particularly high or low scores from the sample.
Confounding variables: Variables other than the IV that have (or almost certainly have) had an effect on the DV. We know they have had an effect because they vary systematically with the IV. For example, in the memory experiment, if all of the young participants are given the memory test at 9am, and all of the older participants are given the test at 7pm, the time of day has acted as a confounding variable, as it has varied systematically with the IV. Therefore, any difference in the results of the two groups may be due to the difference in time of day, rather than the difference in age. The effect of this can be reduced or eliminated by testing both age groups at the same time of day.
Demand characteristics: These are clues which participants respond to when in an experimental situation, in which they try to guess the aim or intended outcome of a study and therefore change their behaviour accordingly. They are a form of participant reactivity (people not behaving naturally as they know they are being studied). The effect of these can be reduced by not revealing the aim of the study to the participants, or by using an independent measures design, so that participants only take part in one of the experimental conditions. For example, if participants are told the aim of the memory study, the young participants may try really hard on the memory test, as they may guess that this is the predicted outcome of the study. They act in ways that they think will please the experimenter. Alternatively, the older participants may try really hard on the test in order to prove the prediction wrong. Either way, the participants do not act naturally, so reducing the validity (correctness) of the results.
Investigator effects: These are any unwanted influences that the investigator/experimenter communicates to the participants which affects their behaviour. For example, being more encouraging towards the young participants in the memory test, as this is the expected result. These can be minimised by the use of standardised instructions, or the double-blind procedure.
Randomisation: This is a way of controlling for the effects of extraneous/confounding variables. Allocating participants to tasks, selecting samples of participants, and so on, should be left to chance as far as possible, to reduce the investigator’s influence on a study.
Standardisation: This is where the experience of an experiment is (as far as possible) kept identical for every participant, for example using standardised instructions. This reduces the effect of extraneous/confounding variables.
Single and double-blind: The single-blind procedure is when the participant does not know the aim of the study. This helps reduce the possibility of demand characteristics from affecting the results. Double-blind is when where the investigator who deals with the participants also does not know the aim of the study. This helps reduce the chance of investigator effects, as the investigator will not unconsciously communicate the aim to the participants.
Control groups: These are used for the purpose of comparison, often when testing the effects of a drug, for example. One group of participants (the ‘experimental group’) will be given the real drug, another group a placebo (fake drug). This can allow the researcher to directly compare the results of the two groups. If the experimental group improves then it is likely that this is because of the drug.
In a study, the population is the target group of people the researcher is studying (e.g. ‘males in their 20s’). The sample is the group of people selected to take part in the study, drawn from the target population. In order to select a sample, a sampling technique will be used:
Random: Each member of the target population has an equal chance of being selected. For example, using a random name or number generator, or picking names out of a hat.
- Evaluation: The researcher cannot influence the selection of participants, but the sample could, by chance, end up being unrepresentative.
Systematic: A participant is selected in a systematic way, for example, selecting every 10th person from the electoral roll. This is decided through a sampling frame, where a list of the target population is put in order.
- Evaluation: The researcher cannot influence the selection of participants, and it is likely to be representative.
Stratified: The make-up of the sample reflects the make-up of the target population. For example, if studying teachers, as there are more female than male teachers, there should be more female participants. If 60% of teachers are female, this means that in a sample of 20, there should be 12 female and 8 male teachers. Once these quotas are identified, the participants to fill them are selected at random from the target population.
- Evaluation: random techniques are used, so the researcher can’t influence the selection. The sample produced is representative of the target population, as it has been designed to be so. However, it is hard to represent all the ways in which people are different.
Opportunity: Participants are selected from whoever is most easily available. For example, standing in the street one afternoon and approaching passers-by to see if they want to take part.
- Evaluation: this is convenient, as it is much less time-consuming and costly than some of the other methods. However, there is a high chance of obtaining an unrepresentative sample, as large groups of the population have no chance of being involved. Also the researcher controls which participants are selected, which could lead to bias.
Volunteer: Participants put themselves forward to take part in a study. For example, a newspaper or internet advert is placed asking for volunteers, and people respond agreeing to take part.
- Evaluation: this is easy and convenient for the researcher, but it is open to volunteer bias, whereby only certain types of people (the type that put themselves forward) take part. This reduces the representativeness of the sample.
These are small-scale trial runs of an experiment. The purpose is to check that the procedure works smoothly and that there are no misconceptions. Any problems can be addressed and the procedures amended for the real study.
This refers to how participants are allocated to experimental conditions.
Independent groups: There are different participants in each condition. This would be appropriate for the memory experiment, as the participants need to be different ages in each condition.
- Advantages: participants are less likely to guess the aim of the study, and there are no order effects- effects arising from having completed two tasks, for example becoming more practised and doing better in the second condition, or getting bored/fatigued and doing worse in the second condition
- Disadvantages: there may be differences between the two groups of people- e.g. intelligence, age, gender, which may cause differences in the results. Twice as many participants are needed.
Repeated measures: One group of participants completes both/all of the conditions of the experiment. For example, to test the effect of listening to music on problem-solving ability, participants are given problems to solve whilst listening to music, then another set of problems to solve in silence.
- Advantages: there are no participant variables between the conditions, and fewer participants are needed
- Disadvantages: there are order effects which may influence the results. This can be addressed by counterbalancing- half the participants do one task followed by the second (A followed by B) and the other does the opposite (B followed by A). Also, participants may work out the aim of the study, like in the music example, and may change their behaviour.
Matched pairs: As independent groups, but participants are ‘matched’ on qualities relevant to the experiment. For example, in the music study, one participant is allocated to the ‘music’ condition, and another who is similar in terms of age, IQ and occupation is allocated to the ‘silence’ condition. This involves pre-testing participants on certain measures in order to match them up.
- Advantages: order effects and demand characteristics are less likely to have an impact, and participant variables are reduced
- Disadvantages: participants can never be matched perfectly, so there might still be some participant variables. Matched pairs is the most time-consuming and expensive design to use.
- Identify the experimental design used in this study.
- Your answer should include: Matched / Pairs
- Evaluate this experimental design.
- Your answer should include: Participant / Variables / Time / Cost
- ‘A psychologist wished to test the effects of an authority figure on participants’ willingness to obey orders. In one condition, participants were told by a man calling himself a professor, dressed in a white coat, to copy out a dictionary until instructed to stop. The second group of participants were chosen to be similar to the first group on several characteristics, and were given the same instruction by a teenage boy wearing a school uniform.’ Identify the experimental design used in this study.
- Your answer should include: Matched / Pairs
Structured or unstructured: The researcher may decide to focus on particular behaviours or actions when conducting the observation, and only record these- this is the structured method (more likely with several participants). Alternatively, they may just record everything that is going on- this is the unstructured method (more likely with fewer participants).
- Evaluation: Structured- the data is much easier to analyse to spot trends and make conclusions (it is likely to be quantitative). But, the data is in less detail as only certain things have been recorded. Unstructured- data is much harder to analyse, and there is more risk of observer bias (only recording things that fit with the observer’s preconceived ideas). But, the data in in more detail and is much ‘richer’.
Behavioural categories: This is the process of making a target behaviour measurable, by breaking it down into observable components. ‘Aggression’ could be broken down into shouting, hitting, punching and so on. This allows different observers to use the same checklist of behaviours, as they are not (or, should not be) subjective.
- Evaluation: To be useful, categories should be unambiguous and objective, as far as possible, and should cover all possible behaviours. Categories should not overlap (e.g. ‘hitting’ and ‘striking’ would be too similar).
Sampling methods: In unstructured observations, continuous recording is often used, where everything that happens is recorded. Because this is often not feasible, researchers usually use one of two sampling methods to record data. Event sampling is when a record is made of each example of a particular behaviour during the observation (e.g. counting the number of times someone shouts out). Time sampling is when behaviours are recorded in a specific timeframe (e.g. recording the behaviours shown during a one-minute time frame every five minutes).
- Evaluation: Event sampling- good for recording infrequent behaviours that may be missed otherwise. But, other important behaviours may be missed. Time sampling- reduces the number of observations needed, but the behaviour recorded may be less representative of the whole observation.
Inter-observer reliability: Usually there will be more than one person observing behaviours, to lessen the risk of observer bias affecting the results. Observers will use the same behavioural categories, and will observe the same behaviours in a pilot study, comparing their findings with the other observer(s). The results are compared, and the level of reliability is worked out by correlating the pairs of observations made. If necessary, behavioural categories can then be modified if inter-observer reliability is low.
Questionnaires: Can be designed with open or closed questions. In addition, questionnaires can use Likert scales, where the responder indicates their depth of feeling (‘1= strongly agree, 5= strongly disagree’); rating scales, where a number is chosen to indicate strength of opinion (‘how important do you think respect for authority is- 1=not important at all, 5= extremely important’); and fixed-choice options where the responder has to select from a pre-determined response which matches their opinion (‘why do people obey orders? (a) Uniform (b) Location (c) Proximity (d) Personality’)
Interviews: Most of these will be one-to-one with just the interviewer and interviewee, although sometimes they take place in groups. The list of questions to be asked will be drawn up, and the interview can be unstructured, structured or semi-structured. The interviewer aims to establish a rapport with the interviewee and make them feel comfortable.
Writing questions: To work effectively, questions should:
- Be free of technical language (jargon) and unfamiliar terms (‘do you think the agentic state is a valid explanation for obedience?’) as most people will not know what they mean
- Avoid emotive language and leading questions (‘do you think the disgusting habit of smoking should be banned?’) which guide participants towards the answer you are hoping for
- Not be ‘double-barrelled’- meaning two questions in one (‘do you think that smoking should be banned and anyone caught smoking sent to prison?’) as participants may agree with one half but not the other
- Avoid double negatives (‘are you not unhappy?’) as these can be confusing