Independant Variable - some event which is directly manipulated by an experimenter in oder to test its effects on another variable
Dependant Variable - what you measure in the experiment
Extraneous Variables - any other variable (aprats from the IV) that may potentially affect the DV and thereby confound the results. For example:
- Participant Variables - eg gender, age, intelligence etc.
- Situational Variables - eg time of day, temperature, noise etc.
- Investigator Variables - eg language (formal/accent/long words) etc.
If they have affected your results they are known as confounding variables.
The IV and DV must be operationalised.
Takes place in a carefully controlled environment but the researcher manipulates the IV.
- you are able to use technical equipment to make your results more accurate.
- replication is easy so its easier to test reliability
- you reduce extraneous variables and it gives you a better chance of a causal link
- less realistic - not measuring natural behaviour so is hard to generalise
- demand characteristics - participants try to guess what behaviour is wanted causing unnatural behaviour.
Takes place in a natural environment but the researcher is still manipulating the IV.
- behaviour is more natural (easier to generalise)
- reduced demand characteristics
- measurements not as accurate
- more time consuming and expensive
- more likely to have extraneous variables to confound results
- harder to replicate
Takes place in a natural environment and the IV varies naturally
- you can carry out research where you cannot ethically change the IV
- Easier to generalise results - very high internal validity
- very little demand characteristics
- very little control
- can be many extraneous variables
- replication may be impossible
- sample bias
- ethnic problems
A way of seeing if there is a relationship between two variables. No IV or DV. Both variables are measured without manipulation.
Correlational co-efficient - number between 1 & -1 that expresses strength of the relationship.
A co-efficient of 0 means there is no relationship
A positive correlation occurs when both variables increase together.
A negative correlation occurs when one value decreases as the other increases.
- Allows predictions to be made
- It can tell you where further research needs to be carried out
- Allows research to be carried out without manipulation
- You cannot establish a causal connection
- It only works for linear relationships not curvy linea
The precise measurement of naturally occurring behaviour. Uses behavioural catagories which are clearly defined statements about what observers are looking for.
Training may be required all observers need to record in the same way. This achieves inter-rater reliability (consistency between observers)
Naturalistic observation - takes place in natural environment (watching children play in park)
Controlled observation - artificial environment (the strange situation)
Participant observation - the person doing the observation takes part in the behaviour (dinner ladies in a school)
Disclosed/overt observation - the people know you are watching them
Undisclosed/covert observation - the people do not know you are watching them
Observer bias - when the observer sees what they are looking for (eg belives girls are more aggressive so observations reflect that)
Blind procedure - participants dont know the conclusions that will be drawn from the experiment
Double blind provedure - recorders/observers dont know either
- more likely to see natural behavior, high external validity
- very practical method for studying of children and animals
- covert - people wont change their behavior
- you can only infer the causes of behaviour rather than make causal connections
- you can get observer effects if overt (change behaviour)
- observer bias
- you miss other catagories if not identified in catagory at start
- hard to control
- ethical issues
- rely on memory - participant observation
Questionaires and Surveys
These are usually concerned with opinions and attitudes.
You can use closed or open questions to collect data which is appropriate to your study
Closed questions - easy to quatify but can be artificial and lose detail, limited range of answers (tick boxes)
Open questions - will provide rich data about human experience but are time consuming anddifficult to analyse, word format, qualitative data (opinions). Questions must avoid technical terms, leading questions, vagueness, ambiguity, assumptions, emotive language and double barrelled questions.
abmiguity - interpreted in more than one way.
Questionnaires and Surveys
- closed questions - you can statistically analyse easily large amounts of data
- easy to quantify
- reduces investigator effects
- less time consuming (CQ)
- easy to replicate/compare
- you can identify faults
- Design problems can lead to bad results
- Social desirability bias (we want to be judged as good person especially if not anonymous)
- Low response rate
- Biased samples based on who is willing to fill them out
A question and answer session that takes place face-to-face.
Structured/formal - pre-determined questions i.e. a questionnaire that is delivered face-to-face with diviation from the original questions
Unstructured/informal interviews - less structured, new questions are developed as you go along
- avoids misunderstandings
- better for dealing with complex issues
- Time consuming
- Expensive - interviewer needs to be appropriately qualified
- Ethical issues
- Harder to collect larger amounts of date/analyse it
- Social desirability bias
- Interviewers affect - age, sex, dress etc
- Interviewees don't have the verbal skills to express themselves accurately
This is an indepth study of an individual or place. Sources of data can include diaries, interviews, observations and experiments.
- Clive Wearing
- Shallice and Warrington's study of KF
- Curtis' study of Genie
- Informed consent (of every aspect of study)
Give them the right to withdraw and follow the BP's code of conduct
- You can collect rich, detailed data
- Allows us to get to the essence of human behaviour
- Can be used to show consequences of rare experiences
- They can challenge existing theroies and lead to new developments
- If your looking at one person, findings can be hard to generalise
- Researcher bias
- Ethical issues
- If relying on memory, data may be distorted
Aims and Hypotheses
The aim should include a general statement of why the research is taking place, what is being studied and what the study is trying to achieve.
A testable statement about behaviour that a researcher aims to support by the use of an objective test.
They describe behaviour in general not just the behaviour of the participants in the experiment.
It should be in present tense.
It is written at the start of the research process before the results are known
Be specific about the variables (operationalise)
Aims and Hypotheses
Alternate/experimental hypotheses - predicts a significant difference in the dependant variable as a result of the manipulation of the independent variable
Null Hypotheses - will predict no significant difference in the DV and the IV. Suggests your results could have occurred by chance.
Directional - indicates which way the DV will change
Non-Directional - suggests there will be a difference but it could be and increase or a decrease of the DV
The choice of which to use will depend on:
- how sure you are on the outcome
- whether you have previous experience/research
Aims and Hypotheses
- Include the IV (bothe conditions) and the DV
- operationalise both variables
- decide if you want it to be directional or non-directional
- use the word 'significant'
Possible Frame works
Directional - People who (IV 1st condition) do significantly better at (DV) than those who (IV 2nd condition)
Non-Directional - There is a significant difference in (DV) between people who (IV 1st) and people who (IV 2nd) .....OR..... Any difference between (IV 1st) and (IV 2nd) when (DV) is due to chance
Aims and Hypotheses
Directional - There is a significant positive/negative correlation between (variable A) and (variable B)
Non-Directional - There is a significant correlation between (variable A) and (variable B)
WHEN 'RELATIONSHIP' IS WRITTEN IN AN EXAM QUESTION IT MEANS WRITE A CORRELATIONAL ANALYSIS
You use the same participants for both conditions - minimum of 24 participants
- no group differences (eg age, gender, ability etc)
- fewer participants required
- Order effects - results may be affected when an activity is repeated by fatigue or practice (controlled by counter balancing)
- Time consuming
- can lose participants between conditions
- demand characteristics (screw you effect)
Counter balancing - half of the participants do one condition first and the other half does the other condition first - then swap
Uses different participants for each condition - minimum of 12 participants
- no order effects
- reduced chance of demand characteristics - participants are less likely to guess what behaviour is expected
- need more participants
- Group differences could create confounding variables
Uses different participants but matches them on any characteristics which could be important (age, gender, ethnicity, memory)
- no order effects
- reduced chance of demand characteristics
- reduced group differences
- matching is difficult
- participants may guess the aim of the experiment as the IV may be noticeable
- time consuming
A pilot study is a trial run of an experiment to make sure everything works the way you had planned. It is a chance to test you procedure before you carry out the main experiment. This could save you wasting time and money.
Examples of things you may need to change
- instructions if participants don't understand them
- poorly worded questions theat have been misunderstood
- inaccurate measuring equipment
- behavioural categories that are not clear enough or that are missing
- stimulus materials that don't work (eg word lists have unfamiliar words)
Reliability refers to how consistent or trustworthy your results are. The results would be the same id you were to repeat the experiment.
Test-retest method - carry out your test, then do it again
Split-half test - (usually used for questionnaires) same question asked twice in different ways, then correlate
Alternate test - try to measure things in a different ways using different tools
To improve reliability
- training to make sure everyone is consistent
- behavioural categories
- take more than one measurement from each participant (prevents flukes, find average)
- use a pilot study
- standardise the way researchers collect date (Inter-rate (observer) reliability) everyone does the same job
- check data very carefully (and that it makes sense)
The ability of the study to test the hypothesis that it was designed to test.
Are your results affected by;
- demand characteristics
- confounding variables
- poor measurement
- investigator effects (unintentional influence of the investigator)
Improving internal validity
- reduce demand characteristics - single blind technique
- reduce investigator effects - double blind technique
Can you generalise your results
1) Ecological validity - can your results be applied outside your experiment setting?
2) Population validity - can your results be generalised to other populations?
Improving external validity
- ecological validity - make your experiment more natural
- population validity - improve your sampling techniques
The more ecologically valid your research is the more likely you are to have confounding variables - you cannot control everything.
Informed consent - we should be maintaining peoples trust
- you should always get informed consent first if possible ...OR... prior general consent (explain similar levels of the experiment) ...OR... presumptive consent (ask a group of people similar to your group of participants and presume they would answer the same)
Deception - people have the right to expect that they will not be misled into behaviour that may be damaging to them
- should follow the British Psychological society's coed of conduct
- get your research reviewed by ethical committee
- if you have deceived someone you should debrief them
- give them the right to withdraw
Right to withdraw - we cannot force people into things in a free society
- make sure they know they have the right to withdraw
- people need the right to protect themselves from harm
Confidentiality - people have the right to expect they will not be named in a report
- take out ALL personal data e.g. name, age, address
Invasion of privacy - researchers do not have the right to pry into peoples lives without permission
- you should not observe people without permission except in public places
Protection from psychological harm - people should not be expected to be harmed during research
- give them the right to withdraw
- ethics committee
- follow code of conduct
- if harm occurs debrief then give follow up care/help
Conduct - people have the right to expect that the researchers are properly qualified or supervised by qualified people
- ensure/check researchers are fully qualified/supervised
In order to carry out research we need participants who are supposed to represent your target population (the group of individuals a researcher is interested in the population you will apply your research to)
The lottery method, pull names out of a hat etc.
- all members of the target population have equal chance of selection
- the researcher may end up with a biased sample (more girls than boys)
- time consuming
- hard to get the necessary information (all blondes in the UK)
As people walk by you in the street i.e. those who are available.
- the easiest method because you just use the first participants you find which takes less time to locate your sample than if using another method
- inevitably biased because the sample is drawn from a small part of the target population eg. highstreet on monday morning in a city wont include professionals or people from rural areas
Advertise in a newspaper or on a notice board willing participants into your responsee
- access to a variety of participants which would make the sample more representative and less biased
- sample is biased because participantsare likely to be highly motivated and have plenty of free time
Types of Data
- objective, less detailed, precise, reliable
- quick to analyse (especially with a computer)
- often used for research on behaviour
- can be examined as descriptive statistics
- subjective, imprecise, unreliable, detailed, rich in meaning, experimental, time consuming to analyse
- used for beliefs, opinions, attitudes
- uses smaller samples but larger amounts of data are produced
- difficult to present
- often aims to confirm pre-existing categories or find emerging categories in the data. From these the conclusions are drawn
Graphs and Charts
Bar Charts - these show discrete data which has categories. The columns MUST have a space between them
Histograms - these show continuous data. The columns do not have a gap between them
Scattergraphs - these illustrate correlational analysis. Each cross on the graph represents one participant
Bar chart Histogram Scattergraph
Measures of central tendency - ways of looking at the average
- makes use of the values of all the data in one final calculation
- it can be misrepresentative of the data as a whole if there are extreme values
- it cannot be used with nominal data
- useful when data are in categories, i.e. nominal data
- not a useful way of describing data when there are several modes
- not affected by extreme scores
- not as 'sensitive' as the mean because not all values are reflected in the median
Nominal - the data are in separate categories, such as grouping people to their favourite football team
Measures of Dispersion
- easy to calculate
- can be distorted by extreme values
- tells you nothing about how the scores are grouped or how many observations are in the data set
Standard deviation - shows the average variation from the mean
- takes every score into account
- may hide some of the characteristics of the data (eg. extreme values)
- harder to calculate
Analysis of Qualitative Data
Identify common themes in your data and then count how often they occur. This can convert qualitative data to quantitative date as it is a frequency count - this allows statistical analysis. You can choose your own themes or see what ones emerge from the responses.
- mathematical way of analysing data
- indirect form of observation
- counting specific occurrences eg. words, themes
- interpretations are often biased
Presentation of Qualitative Data
The catagorised data can be listed with an indication of how often they occurred. Typical examples of behaviour within the categories can be given. Quotations can also be used to illustrate the data and add meaning. They should be typical responses.
Researchers must code in the same way to ensure reliability, therefore coding units should be strictly operationalised.