Aims and hypotheses
The aim of any study is a statement of what the researcher intends to investigate.
A hypothesis is a precise and testable statement about the expected relationship between variables. A hypothesis may be directional or non-directional:
- Directional- where it states the direction of the predicted difference or relationship between 2 conditions or 2 groups of participants.
- Non-directional- predicting that there will be a difference or relationship between 2 conditions or 2 groups of participants, without stating the direction.
IV & DV
IV- independent variable:
Is an event that is directly manipulated by the experimenter in order to observe its effects on the DV.
DV- dependent variable:
Depends in some way on the IV. Variables must be operationalised, i.e. defined in a way that they can be tested.
A set of procedures used to control the influence of factors such as participant variables in an experiment.
Repeated measures: Same participants in every condition being tested. + Good control for participant variables and fewer participants needed - Order effects and participants may guess the purpose of the experiment.
Independent groups: Participants are allocated to 2 (or more) groups representing different experimental conditions. + Avoids order effects and participants guessing the purpose of the experiment. - Needs more participants and there is no control of participants variables (can use random allocation).
Matched pairs: Pairs of participants matched on kay participant variables. One member of each pair is placed in the experimental group and the other member in the control group. + Avoids order effects and participant variables are partly controlled. - Matching is difficult and never totally successful.
Counterbalancing & Groups and conditions
Counterbalancing can be used to deal with order effects by ensuring that each condition is tested first or second in equal amounts.
- Some participants receive condition A then B, others received B then A.
- Or ABBA- All participants receive A B then B A.
Groups and conditions:
In an experiment using a repeated measures or matched pairs design each participant takes part in 2 or more conditions. The experimental condition (or conditions) contains the IV. The control condition provides a baseline measure of behaviour without the experimental treatment (IV), so that the effect of the experimental treatment may be assessed. In an experiment using an independent groups design, there is an experimental group and a control group.
Key terms 1
Demand characteristics: a cue that makes participants aware of what the researcher expects to find or how participants are expected to behave.
Extraneous variables (EV): any variable, other than the IV, which may potentially affect the DV and thereby confound the findings. Order effects, participant variables and situational variables may act as EVs.
Order effects: in a repeated measures design, an extraneous variable arising from the order in which conditions are presented. For examples, participants do better the second time beacuse they have had practice (practice effect).
Random allocation: allocating participants to experimental groups using random techniques.
Control of variables
'Control' refers to the extent to which any variable is held constant or regulated by a researcher. The IV is controlled to observe its effect. EVs are controlled so any effect can be attributed to tthe IV, for example:
- Participant variables such as ages, intelligence, motivation, etc. may explain why participants in one group do better.
- Situational variables such as time of day, temperature, noise, etc. may also explain group differences.
- Investigator effects, where an investigator unwittingly communicated information.
- Demand characteristics trigger a predictable response.
Methods of control include:
- Single blind technique- participants don't know the true aims of a study.
- Double blind technique- investigators and participants don't know the true aims of a study.
Observational design- Behavioural categories
It is necessary to devise objective methods to separate the continuous stream of action into separate behavioural categories, i.e. operationalise the target behaviour(s). This can be done using:
- Behaviour checklist- a list of component behaviours
- Coding system- individual behaviours are given a code for ease of recording
The behavioural categories should:
- Be objective- the observer should not have to make inferences about the behaviours.
- Cover all possible component behaviours and avoid a 'waste basket' category.
- Be mutually exclusive, meaning that you should not have to mark two categories at one time.
Observational design- Sampling procedures & Partic
In may situations, continuous observation is not possible because there would be too much data to record, therefore there must be a systematic method of sampling observations:
- Event sampling- counting the number of times a certain behaviour (event) occurs in a target individual.
- Time sampling- recording behaviours at regular intervals, e.g. every 20 sec.
Participant and non-participant:
In some observations, the observer is also a participant in the behaviour being observed (participant observation) which is likely to affect objectivity.
More often, the observer is not a participant (non-participant observation).
Observational design- Overt and covert; Structured
Overt and covert:
Participants who are aware of being observed (overt observation) may alter their behaviour, so validity is reduced. Making observations without a participant's knowledge (covert observations), such as using one-way mirrors, may raise ethical issues.
Structured and unstructured:
In a structured observation the observer uses behavioural categories and sampling procedures to structure the observations. In a controlled observation both the observations and the environment are controlled (structured); in a naturalistic observation only the observations are structured.
In an unstructured observation the observer records all relevant behaviour but has no system.
Questionnaires and interviews- Good questions
A 'good question' should be clear and unambiguous. It should not be biased in a way that might suggest particular answers to the respondent (i.e. a leading question).
Closed questions have a range of answers from which respondents select one.
- +: They produced quantitative data which is easier to analyse.
-: Respondents may be forced to select answers that don't represent their real thoughts or behaviour.
Open questions invite respondents to provide their own answers rather than select one of those provided. They tend to produce qualitative data.
- +: Can provide unexpected answers and rich detail, allowing researchers to gain new insights.
- -: More difficult to summarise answers because there may be such a wide variety of responses. This then makes it difficult to draw conclusions.
Questionnaires and interviews- Good questionnaires
Questions should be unthreatening and easy to answer. More challenging questions will be answers more truthfully one trust is established.
Filler questions may be used that disguise the true aims of the questionnaire/interview so that respondents are more honest.
Sampling techniques are important in obtaining a representative sample.
A pilot study enables questions to be tested on a small group of people. This means you can refine the questions in reponse to any difficulties encountered.
Questionnaires and interviews- Structured and unst
In a structured interview the questions are decided in advance and are the same for all the participants.
In an unstructured interview the questions are unplanned and usually guided by the respondents' answers.
In a clinical interview (semi-structured) the interviewer starts with a few standard questions but further questions develop in response to the answers given.
Key terms 2
Pilot study- a small-scale trial of a stud. Run to test any aspects of the design, with a view to making improvements.
Reliability- a measure of consistency. Internal reliability concerns consistency within a set of score or items. External reliability concerns consistency over time such that it is possible to obtain the same results on subsequents occasions when the measure is used with the same thing.
Validity- the extent to which a study and its findings are legitimate or true. Internal validity concerns whether a study has tested what is set out to test. External validity concerns the degree to which a research finding can be generalised to, for example, other settings, or other groups of people and over time. Any study that has low internal validity must lack generalisability and therefore also has low external validity.
Reliability- Experimental studies & Observational
Experimental studies- the reliability of an experiment can be determined through replication. To improve reliability extraneous variables should be controlled.
Observational studies- inter-observer reliability is the extent to which there is agreement between 2 or more observers.
- to assess internal reliability- correlate the observations of 2 or more observers. If (total number of agreements) / (total number of observations) > .80, the data has high inter-observer reliability.
- to improve reliability- observers are trained to use behavioural categories.
Reliability- Self-report techniques
Inter-interviewer reliability is the extent to which 2 interviewers produce the same outcome from an interview.
- To assess internal reliability- the split-half method can be used. Scores/responses on both havlves of a test should be the same/consistent.
- To assess external reliability- the test-retest method can be used when the same questionnaire/interview is repeated with the same respondent a few weeks apart.
- To improve reliability- remove questions which create inconsistency.
Validity- Experimental studies & Observational stu
Internal validity- the degree to which the observed effect was due to experimental maniputlation rather than factors such as extraneous variables. It is also effected by mundane realism and experimental realism.
External validity- the representativeness of the sample affects the ability to generalise the findings to other people and situations (e.g. ecological validity).
Internal validity- may be affected by an inadequate system of behavioural categories (e.g. not enough categories) and by observer bias (observers' expectations affect their objectivity. This can be improved by using more than 1 observer in more than 1 setting.
External validity- Naturalistic observations are likely to have high ecological validity but may have low population validity when the sample is limited.
Validity- Self-report techniques
Internal validity- may be affected by interviewer bias and social desirability bias.
Improving validity- use a more representative sampling method and better designed questions.
Assessing validity- compare the results with an established measure of the same thing (called concurrent validity).
External validity- the representativeness of the sample affects the ability to generalise the findings to other people and situations (population validity or ecological validity).
Ethical issues- Informed consent & Deception
Participants must be given comprehensive information concerning the nature and purpose of a study and their role in it. Using this information, participants can make an informed decision about whther to participate. From the researcher's point of view this may reduce the meaningfulness of the research because such information will reveal the study's aims and affect participants' behaviour.
This occurs when a participant is not tole the true aims of a study and what participation will involve. Thus the participants cannot give informed consent. From the researcher's point of view it might be argued that some deception is relatively harmless and/or can be compensated for by debreifing.
Ethical issues- Right to withdraw & Protection fro
Right to withdraw:
Participants should have the right to withdraw from a study if they feel uncomfortable in any way. They should also have the right to refuse the researcher permission to use any data they produce. From the researcher's point of view the loss of participants may bias the study's findings.
Protection from harm:
During a research study, participants should not experience negative physical effects, such as physical injury, or psychological effects, such as lowered self-esteem or embarrassment. From the researcher's point of view it may not be possible to estimate harm before conducting a study- however any study should be stopped as soon as harm is apparent.
Ethical issues- Confidentiality & Privacy
A participant's right to have personal information protected. The Data Protection Act makes confidentiality a legal right. From the researcher's point of view it may not be possible to keep information confidential because details of the study lead to individuals identification.
A person's right to control the flow of information about themselves. From the researcher's point of view this may be difficult, for example in a covert observation. If privacy is invaded, confidentiality should be protected.
Dealing with ethical issues- Debreifing & Ethical
A post-research interview designed to inform participants about the true nature of a study. This aims to restore participants to the state they were in at the start of the study. It may also be used to gain feedback about the procedures used in the study.
Concrete, legal documents that help to guide conduct within psychology. Establishes principles for standard practice and competence.
Dealing with ethical issues- Ethical committee & P
A group of people within a research institution that must approve a study before it begins. May consist of professionals and lay people. Considers how the researcher proposes to deal with any ethical issue that arises. Weighs up costs-benefit issues.
A method of dealing with lack of informed consent or deception. Asking a group of people who are similar to the participants whether they would agree to take part in a study. If this group of people agree to the procedures in the proposed study, it is presumed that the real participants would agree as well.
Key terms 3
Ethical issues: the dilemmas created by the conflict between the needs of researchers and the rights of participants. For example, in order to conduct meaningful research it may be necessary to deceive participants, but this affects their right to giving fully informed consent.
Attrition: the loss of participants from a study over time. This is likely to leave a biased sample, or a sample that is too small.
Random technique: method of selection that ensures each member of the population has an equal chance of being selected. For example, placing all names in a hat and drawing out the required number. Or by assigning each person a number and using a random number table.
Key terms 4
Sampling: the process of taking a sample. The technique used aims to produce a representative selection of the target population.
Target population: the group of people that the researcher is interested in. The group of people from whom a sample is drawn. The group of people about whom generalisations can be made.
Volunteer bias: a form of sampling bias. Occurs because volunteer participants are usually more highly motivated than randomly selected participants.
Selection of participants- Opportunity sample
A sample of participants produced by selecting people who are most easily available at the time of the study.
How? As people in the street, i.e. select those who are available.
+ The easiest method because you just use the first participants you can find, which means it takes less time to locate your sample than if using one of the other techniques.
- Inevitably biased because the sample is drawn from a samll part of the target population. For example, if you selected your sample from people walking around the centre of a town on a Monday morning, then it would be unlikely to include professional people (bebcause they are at work), or people from rural areas.
Selection of participants- Random sample
A sample of participants produced using a random technique such that every member of the target population has an equal chance of being selected.
How? Using a random technique such as placing all names in a hat and drawing out the required number.
+ Unbiased, all members of the target population have an equal chance of selection.
- The researcher may end up0 with an unrepresentative and therefore biased sample (e.g. more boys than girls) because the sample is too small.
Selection of participants- Stratified and quota sa
Groups or participants are selected according to their frequency in the population.
How? Subgroups (or strata) within a population are identified (e.g. boys and girls, or age groups: 10-12 years, 13-15, etc.). Participants are obtained from each of the strata in proportion to their occurrence in the target population. Selection is done randomly (stratified sample) or by another method such as opportunity sampling (quota sample).
+ More representative than an opportunity sample because there is equal representation of subgroups.
- Although the sample represents subgroups, each quota taken may be biased in other ways, for example, if you use opportunity sampling you only have access to certain sections of the target population.
Selection of participants- Systematic sample
A method of obtaining a representative sample by selecting every 5th or 10th person.
How? Use a predetermined system to select pkarticipants, such as selecting every 10th person from a phonebook.
+ Unbiased as participants are selected using an objective system.
- Not truly random unless you select a number using a random method and start with this person, and then you select every tenth person.
Selection of participants- Volunteer sample
A sample of participants produced by asking for volunteers.
How? Advertise in a newspaper or on a noticeboard.
+ Access to a variety of participants (e.g. all the people who read a newspaper) which may make the sample more representative and less biased.
- Sample is biased because participants are likely to be more highly motivated and/or with extra time on their hands (=volunteer bias).
Key terms 5
Quantitative data: represent how much, how long, or how many, etc. there are of something. Data that is measured in numbers or quantities.
Quantitative data analysis: any means of representing trends from numerical data, such as measures of central tendency.
Qualitative data: express the 'quality' of things. This includes descriptions, words, meanings, pictures, texts and so on. They cannot be counted or quantified but it can be turned into quantitative data by counting the data in categories.
Quantitative data analysis- Measure of central ten
Inform us of central (or middle) values for a data set, i.e. the average.
Mean: calculated by adding all the number and diving by how many there are. It can only be used with interval or ratio data. + makes use of the values of all the data. - can be misrepresentative of the data as a whole if there are extreme values.
Median: the middle value, suitable for ordinal or interval data. + not affect by extreme scores. - not as 'sensitive' as the mean because not all values are reflected.
Mode: 'most common' value. + useful when the data are in categories, i.e. nominal data. - not a useful way of descrbing data when there are several modes (i.e. a multi-modal set of data).
Quantitative data analysis- Measures of dispersion
Inform us of the spread of the data set.
Range: the difference between the highest and lowest score in a data set.
+ Easy to calculate, provides you with direct information.
- Affected by extreme values, doesn't take into account the number of observations in the data set.
Standard deviation: shows the amount of variation in a data set and assesses the spread ot the data around the mean.
+ More precise measure of dispersion becuase all the values of the data are taken into account.
- May hide some characteristics of the data, for example extreme values.
Quantitative data analysis- Correlation
A correlational analysis determines the extent of a relationship between 2 co-variables.
Zero correlation: co-variables not linked at all.
Positive correlation: co-variables increase together.
Negative correlation: as one co-variable increases, the other decreases. Usually a linear correlation is predicted, but the relationship can curvilinear.
Correlation coefficient: a number between -1 and +1 that tells us how closely the co-variables in a correlational analysis are related.
Quantitative data analysis- Visual display
Graphs and tables, such as:
Bar chart- the height of each bar represents the frequency of that item. The categories are placed on the horixontal (x axis) and frequency is on the vertical (y axis). Bar charts are suitable for words and numbers (nominal or ordinal/interval data).
Scattergram- a graphical representation of the relationship (i.e. the correlation) between 2 sets of scores.
Levels of measurement
Nominal: data is in separate categories, such as grouping people according to their favourite football team.
Ordinal: data is ordered in some way, for example asking people to put a list of football teams in order of liking. The 'difference' between each item it not the same, i.e. the individual may like the first item a lot more than the second, but there might only be a small difference between the items ranked second and third.
Interval: data is measured using unit of equal intervals, such as counting correct answers or using any 'public' unit of measurement.
Ratio: there is a true zero point as in most measures of physical quantities.
Qualitative data analysis
Categorising the data. These can be:
- Pre-existing categories- the researcher decides on some appropriate categories before beginning the research.
- Emergent categories- the categories or themes emerge when examining the data.
Using the categories to summarise the data.
- The categories or themes may be listed.
- Examples of behaviour within the category may be represented using quotes from participants or descriptions of typical behaviours in that category.
- Frequency of occurences in each category may be counted, thus qualitative data is turned into quantitative data.
- Finally, a researcher may draw conclusions.
Advantages and disadvantages of qualitative data
+: Provides rich details of how people behave because participants free to express themselves.
Gains access to thoughts and feelings that may not be assessed using quantitative methods e.g. closed questions.
-: Subjective analysis can be affected by personal expectations and beliefs- however quantitative methods are also affected by bias; they simply may appear to be objective.
More difficult to detect patterns and draw conclusions because of the large amount of data usually collected.
Advantages and disadvantages of quantitative data
+: Easier to analyse because the data are given in numbers that can be summarised using measures such as the mean and range, as well as simple graphs.
Can produce neat conclusions because numerical data reduces the variety of possibilities.
-: Oversimplifies reality and human experience (statistically significant but humanly insignificant).
Processes involved in content analysis
Content analysis is a form of indirect observation- indirect because you are not observing people directly but observing them through the artefacts they produce.
The process involved is similar to any observational study, the researcger has to make design decisions about:
1) Sampling method- what material to sample and how frequently (e.g. which TV channels to include, how many programmes, what length of time).
2) Behavioural categories to be used. These categories can be used in two ways:
- Quantitative analysis- examples in each category are counted.
- Qualitative analysis- examples in each category are described.
As with observations, if there is a team of researchers it is important to ensure that they are applying criteria in the same way by calculating inter-observer reliability.
Key terms 5
Cohort effects: one group of participants (cohort) may have unique characteristics because of time-specific experiences during their development, such as being a child during the Second World War. This can affect both cross-sectional studies (because one group is not comparable with another) or longitudinal studies (because the group studies is not typical).
Cross-cultural study: a kind of natural experiment in which the IV is different cultural practices and thr DV is a behaviour, such as attachment. This enables researchers to investigate the effects of culture/socialisation.
Cross-sectional study: one group of participants of a young age are compared with another, older group of participants, with a view to finding out the influence of age on the behaviour in question.
Hawthorne effect: the tendency for participants to alter their behaviour merely as a result of knowing that they are being observed.
Key terms 6
Intervening variable: a variable that comes between 2 other variables that is used to explain the relationship between those 2 variables.
Investigator effect: anything that the investigator/experimenter does that has an effect on a participant's performance in a study other than what was intended. This includes direct effects (as a consequence of the investigator interacting with the participant) and indirect effects (as a consequence of the investigator designing the study).
Investigator/experimenter bias: the effect that an investigator/experimenter's expectations have on the participants and thus on the results of the research study.
Meta-analysis: a researcher looks at the findings from a number of different studies in order to resach a general conclusion about a particular hypothesis.
Key terms 7
Longitudinal study: observation of the same items over a long period of time. Such studies usually aim to compare the same individuals at different ages, in which case the IV is age. A longitudinal study might also observe a school or other institution over a long period of time.
Participant effects: a general terms used to acknowledge the fact that participants react to cues in an experimental situation and that this may affect the validity of any conclusions drawn from the investigation.
Quasi-experiments: studies that are 'almost' experiments but lack one or more features of a true experiment, such as full experimenter control over the IV and random allocation of participatns to conditions. This means that they cannot claim to demonstrate causal relationships.
Key terms 8
Role play: a controlled observation in which participants are asked to imagine how they would behave in certain situations, and then asked to act out the part. This method has the advantage of permitting a study of certain behaviours that might be unethical or difficult to find in the real world.
Standardised procedures: a set of procedures that are the same for all participants in order to be able to repeat a study. This includes standardised instructions- the instructions given to participants to tell them how to perform a task.
Experiments- Lab, field, natural
Lab- IV manipulated to observe effect on DV, highly controlled.
+: can draw causal conclusion; extraneous variables minimised; can be easily replicated. -: contrived, tends to lack mundane realism; investigator and participant effects.
Field- More natural surroundings, IV directly manipulated by experimenter to observe effect on DV, some control.
+: can draw causal conlusion; higher ecological validity; avoids some participant effects. -: less control; may have demand characteristics.
Natural- IV not directly manipulated, p's are randomly allocated.
+: allows research where the IV can't be manipulated for ethical/practical reasons; enables psychologists to study 'real' problems. -: cannot demonstrate causal relationships; many extraneous variables; investigator and participant effects.
Correlational analysis & Case studies
Correlational analysis: co-variables examined.
+: can be used when not possible to manipulate variables; can rule out a causal relationship. -: people often misinterpret correlations; there may be other, unknown variables.
Case studies: detailed study of a single individual, institution or event. Involves many different techniques, e.g. interviews, psychological tests.
+: rich, in-depth data collected; used to investigate unusual instances of behaviour; complex interactions studied. -: lacks generalisability; may involve unreliable, retrospectice recall; researcher may lack objectivity.
Naturalistic observation: everything left as normal, all variables free to vary.
+: study behaviour where can't manipulate variables; high ecological validity. -: poor control of extraneous variables; observer bias, low inter-observer reliability.
Controlled observation: some variables controlled by researcher, e.g. the environment.
+: can manipulate variables to observe effects. -: less natural, reduced ecological validity; investigator and participant effects; observer bias, low inter-observer reliability.
Content analyis: indirect observation of behaviour based on written or verbal material such as interviews on TV.
+: high ecological validity because based on what people do; can be replicated easily because sources are publicly available. -: observer bias.
Questionnaires: set of written questions.
+ can be easily repeated, so lots of people can be questioned; respondents may be more willing to reveal personal information; does not require specialist administrators. -: leading questions, social desirability bias; biased samples.
Interviews: unstructured interviews where the interviewer develops questions in response to respondent's answers, conducted in real time.
+ more detailed information collected; can access unexpected information. -: social desirability bias, interviewer bias, inter-interviewer realiability, leading questions; requires well-trained personnel.