Extraneous variables effect the internal validity of the study.
- Situational variables: associated with the situation e.g. temperature, lighting, time of day. They are controlled using standardised procedures, ensuring conditions are set up in exactly the same way.
- Participant (ppt) effects: associated with research ppt e.g. age, intelligence, gender. These include demand characteristics - cues in a study that make ppt aware of what is expected of them and change their behaviour - and social desirability bias - when ppts behave in a certain way to present themselves in a desirable way to researcher. Controlled through experimental design e.g. repeated measures, randomly assigning ppts to conditions or a single blind design, field experiments or covert observations.
- Investigator effects: occur when researchers behaviour or characteristics influence research. Controlled by using standardised instructions to prevent leading questions or using double blind design.
Operationalisation, Aims and Hypotheses
Operationalistation = Defining how we are going to measure or manipulate variables. Variables must be in a form that can be tested and allows for replication.
- e.g. For memory you might use the number of words recalled correctly. For aggression you might list various behaviours that can be measured (hitting, biting, kicking).
Aim leads to hypothesis (a clear statement of what researcher believes to be true). It is a precise and testable statement of relationship between two variables.
Hypothesis can be directional (one-tailed) or non-directional (two-tailed).
- e.g. People recall more words correctly during a memory test when it is done under quiet conditions rather than with loud noise. (Directional)
- e.g. There will be a difference between the number of words recalled correctly during a memory test when it is done under quiet conditions and when it is done with loud noise. (Non-directional)
Types of Experiments
- Conducted in controlled environment - artificial, no extraneous variables.
- IV is directly manipulated by researcher.
- High control of variables, minimising extraneous variables - Can be confident that the change in IV is causing the change in DV.
- High replicabilty - easier to repeat so other researchers can check reliability of results.
- High control makes situation artificial - difficult to generalise from lab to real life settings, lacks ecological validity.
- Demand characteristics - people may behave differently in a lab to a real life setting so ppts may try to guess aims of study and change their behaviour.
Types of Experiments
- Conducted in more natural setting
- Researcher manipulates IV and ppts can be allocated to conditions.
- Less demand characteristics - usually unaware they are in an experiment.
- Higher external validity - conducted in natural setting so findings can be generalised to similar settings.
- Less control of extraneous variables - decreases internal validity, researchers cannot be confident that change in IV causes change in DV.
- Ethical issues - e.g. deception as ppts unaware they are in an experiment.
Types of Experiments
- Conducted in natural environment.
- IV is not directly manipulated by researcher.
- Ppts cannot be randomly allocated to conditions - there will be biases in different groups.
- Natural setting - behaviour more likely to be genuine, increase external validity.
- Allows study of effects of IV that would be unethical to manipulate e.g. stress or hours spent in daycare.
- No random assignment of ppts to conditions - individual differences.
- Many extraneous variables - cannot establish cause and effect.
Repeated Measures Design
- Same ppt used in both conditions
- Ppt variables eliminated as each ppt acts as their own control
- Less time needed to find ppts as this design uses few ppts.
- Order effects - e.g. fatigue/boredom may lead to poorer performance or practice may lead to improved performance.
- Increased demand characteristics - ppts guess aim of experiments and change behaviour.
- Order effects controlled through counterbalancing - ensures each condition is tested 1st and 2nd in equal amount.
Independent Groups Design
- Each ppt only tested in one condition.
- Ppts must be allocated to conditions randomly e.g. names out of a hat.
- No order effects.
- Reduced demand characteristics.
- Can use same stimulus in all conditions.
- No control of ppt variables - differences between conditions caused by using different ppts not IV.
- Needs twice as many ppts.
- Ppts randomly allocated to conditions using matched pairs or repeated measures design.
Matched Pairs Design
- Ppts matched closely with another ppt and members of each pair randomly allocated to either one or another of conditions.
- No order effects.
- Good way of controlling ppt variables.
- Time consuming to match ppts exactly.
- More ppts required.
- Everyone in target population has equal chance of being selected.
Method: Every member of target population identified. Random sampling technique used e.g. manual selection - every name put onto separate pieces of paper and placed into container. Researcher selects however many slips to make sample size. Container should be shaken between each draw, slips folded in the same way and selector shouldn't be able to read slips of paper.
- More representative than other methods - equal chance of being selected so findings can be generalised to population.
- Even if sample is random it may not truely be representative of population.
- There are practical limitations with obtaining a random sample.
- Selected by using people who are most readily available.
Method: Researcher approaches people and asks them to take part in research. Whoever is available and willing to participate.
- Convenient if there are no names available to get random sample.
- High chance sample will be biased - drawn from small part of target population therefore not representative and has low population validity so hard to generalise findings.
- Ppts self-select, they volunteer to take part in the research.
Method: Researcher advertises research e.g. in a newspaper or noticeboard and people who respond become the sample.
- A particular type of person is likely to volunteer for research - high chance of bias so cannot generalise to target population leading to low population validity.
A small scale trial of research design, run before doing the real thing.
To check procedures and materials e.g. ppts may not understand instructions, may guess what experiment is about, may get bored because there are too many questions.
Saves time and money in the long run.
Reliabilty and Validity
Reliability = whether a test/results are consistent.
- Researchers should be able to measure/test something time after time and get similar results.
- Replication = when an experiment is repeated using standardised procedures to see if the findings are the same, therefore reliable.
Validity = whether the test/measure is measuring what it claims to measure. Internal and external validity.
- Internal: what goes on inside a study e.g. whether the IV produced the DV. To gain high internal validity researchers must control extraneous variables, minimise investigator effects and demand characteristics.
- External: whether the results can be generalised beyond the study itself e.g. other situations (ecological validity), other people (population validity), other cultures (cultural validity) or other times (historical validity).
Measures of Central Tendency
Mean: Adding up all values and diving by the number of values.
- Advantages: Makes use of all the values of data.
- Disadvantages: Can be misrepresentative of the numbers if there is an anomaly.
Median: Middle value of an ordered list.
- Advantages: Not affected by extreme scores.
- Disadvantages: Not as sensitive as the mean as not all values are reflected in the median.
Mode: Most common value.
- Advantages: Useful when data is in categories.
- Disadvantages: Not a useful way of describing data when there are several modes.
Measures of Dispersion
These tell us how spread out the data is.
Range = Difference between highest and lowest numbers, obtained by subtracting lowest value from highest value.
Standard Deviation = A statistical measure of the spread of data around the mean. The larger the standard deviation, the more the data is spread out around the mean. This shows there is greater variability in individual scores, so care must be taken when interpreting the data.
Informed Consent - Ppts must be told what they will be doing, how long it wil take and why they are doing it so they can provide informed consent.
Deception - Ppts should not be deceived unless absolutely necessary. If deception is required, great care and careful consideration must be given to the project.
Debriefing - After research is complete, ppts must be debriefed and informed of true motivations of investigation. Must be given a chance to ask questions.
Right to Withdraw - Ppts should be free to leave experiment and take their data with them at any time.
Confidentiality - Any information and data provided by ppt must be anonymous e.g. using initials.
Protection from Harm - Physical and psychological safety and well-being of ppt must be protected. Ppts should leave study in exactly the same state as when they began.
Watching what people do.
Controlled observations in laboratories to naturalistic observations in a natural environment.
- Aim of controlled observations is to control variables that might influence behaviour. Naturalistic observations aim to produce data with higher ecological validity.
Observations can be covert (ppt unaware they are being observed) or overt (aware they are being observed).
Behavioural categories are used to collect information in observations to record particular instances of behaviour. Checklist should be: objective, cover all possible areas and be clearly defined and precisely stated so the observer knows what is meant by each behaviour.
Ethical issues: informed consent, deception (if covert), confidentiality, right to withdraw.
- Hard to decide how to categorise behaviour as some are subjective e.g. aggression.
- Demand characteristics may occur (overt or controlled observation).
Asking people questions about their experiences and beliefs, face-to-face.
- Unstructured Interviews: These have little structure, interviewer guides discussion and encourages interviewee to talk freely.
- Structured Interviews: Standard set of questions asked in same fixed order to all interviewees. Not very good for indepth information.
- If open questions are used, rich, detailed information can be gathered.
- As researcher present ppts may be more honest and feel like they cannot lie.
- As researcher present, misunderstandings can be clarified and ppts less likely to leave questions out.
- If open questions used the data can be hard to analyse and draw conclusions.
- More time consuming and expensive than questionnaire - needs specialist training for interviewer.
- Increased risk of investigator effects as researcher is present - lower validity.
- Social desirability bias as ppts may want to present themselves in a good light - lower validity.
Questionnaires can include open questions or closed questions (fixed choice).
Advantages of open questions:
- Allow respondents to express themselves so provide rich, detailed data - hard to analyse.
Advantages of closed questions:
- Produces quantitative data - easy to analyse, but may not give enough detail.
- Easier for respondents to fill in.
Questions and instructions need to be easy to understand. Carrying out a pilot study first is useful so you can make any changes if necessary.
Strengths and Weaknesses of Questionnaires
- Lots of data can be collected relatively quickly and cheaply compared to an interview which needs specially trained interviewers.
- Researcher doesn't need to be present - investigator effects are reduced.
- Questionnaires offer ethical way to collect data e.g. when researching stress.
- Anonymous - ppts more likely to be truthful in answers.
- May be problems with leading questions and social desirabilty bias - lower validity.
- Sample bias - certain kinds of people fill in questionnaires (literate individuals who are inclined to spend time filling them in).
- Ppts may give untruthful answers - especially in relation to sensitive issues - lower validity.
- May be ambiguity/misunderstanding which cannot be explained as there is no researcher present.
Detailed and in-depth study of a single individual or group of people.
Using information from a range of sources - interviews, psychological tests, observations, written records and experiments.
Generally longitudinal (following an individual or group for an extended period of time).
- Produces rich, meaningful data (qualitative).
- Offers high levels of ecological validity.
- Usually one-off situations - difficult to replicate, so difficult to establish reliability of data.
- Specific setting/population - difficult to generalise results beyond individual/group being studied, low population validity.
Representation of Qualitative Data
Observational studies, questionnaires and interviews can produce qualitative data (non-numerical).
Content analysis: A way of analysing data such as test/images using themes and categories - concerned with converting qualitative data into quantitative data.
E.g. Using content analysis on diary entries, researcher would create a checklist of different categories which included behaviour/words they were interested in. Then read through the diary entries and count/tally the number of times each category appeared and compare with other diary entries.