Upgrade to remove ads
Only $35.99/year

AQA A2 Psychology PSYA4 Research Methods

Terms in this set (46)

What makes a true experiment?
1. Direct manipulation of the IV to produce a change in the DV
2. Randomisation, ps are randomly allocated to conditions
3. Control extraneous variables, they should be eliminated
What are the three types of experiment?
Lab, field and natural
Lab experiments
-The IV is manipulated by the psychologist and extraneous variables are controlled.
-The direct manipulation establishes a cause and effect relationship
-Standardised procedures make the study easy to replicate
-Sophisticated equipment increases accuracy and reliability
-Lack of EV
-Ethics: deception (not told the true aim) but if they discover this then there could be demand characteristics.
Field experiments
-Conducted in a ps natural environment e.g. at home
-The IV is manipulated by the researcher
-Improves EV as artificial lab settings are avoided
-Ps may not have given informed consent to take part, but this does reduce the likelihood of demand characteristics
-Less control of other extraneous variables that they could become confounding variables
Natural experiments
-The IV is not directly manipulated by the researcher and so is naturally occurring then recorded to see the effects it has on the DV.
-Loss of control of confounding variables makes it more difficult to infer cause and effect
-Virtually impossible to replicate
What is correlational analysis?
A statistical technique which measures the relationship between two or more co-variables to see if there is a trend or pattern between them
This will either be a positive or negative trend
Correlational analysis involves the correlation coefficient
Measures the strength of the correlation
+1 = strong positive
-1 = strong negative
0 = variables are not correlated
Correlational analysis acts as a preliminary tool
For areas of new research. If there doesn't seem to be a general correlation it would waste both time and money.
Correlational analysis doesn't show a.....
Causation
Naturalistic observations
-By the researcher in a natural environment so that the IV is not manipulated
-high EV
-can't infer cause and effect
-low control over variables
-ethical issues
Controlled observations
-The researcher controls certain variables however natural behaviour is still observed
-Possible observer bias
-Still love EV
-High probability of demand characteristics occurring
Other types of observations
1. Ps observations, when the researcher participates
2. Non-ps observations, when the researcher doesn't take part
3. Disclosed obs, the ps know that they are being observed
4. Undisclosed, ps are unaware
5. Structured obs, such as a behavioural checklist. These generally provide quantitative data, gives a restricted view and data is not in great depth
6. Unstructured, recording behaviour as it happens, provides rich qualitative data but it is more difficult to analyse
How to overcome an observer bias?
-Use the double-blind technique, where the ps nor the experimenter know the aim of the experiment or what condition they are in
-Or inter-rater (inter-observer when referring to observations) identifies to what degree do different raters give consistent estimates of the same behaviour
Self-report techniques
-Questionnaires, can gather a vast amount of data in a systematic way from a large number of people
-Closed question, produce quantitative data
-placing in rank order
-attitude scales (strongly agree/disagree)
-liker scales (1=very important)
-Open questions, produce qualitative data
Tips for designing a questionnaire
1. Use short, easy questions, that are not leading
2. Use no ambiguous words
3. Avoid emotionally charged questions
4. Pilot the questionnaire
Positives and negatives of questionnaires
-Able to gather a large amount of information in a small period of time
-Highly replicable and standardised across the ps, making the results easy to compare

-Social desirability bias, ps may not provide truthful information in order to make themselves seem 'better'
-Ambiguous words could cause there to be some misinterpretation and therefore decrease validity of the results
Interviews
Are usually face-to-face interviews, flexible and natural, producing detailed qualitative data
-Structured, using a set of questions which produce more quantitative data
-Semi-quantitative, set of questions but some can be expanded upon
-Unstructured, with just some target points to discuss
Unstructured interviews
-They are flexible, allowing more complex issues to be explored
-Many types, that can be tailored to the requirements of a particular study to to generate a certain type of data
-Misinterpretation of info could lead to a bias
-Replicability is difficult though
Case studies
Are in-depth investigations of a single person, group, event or community
-detailed, rich qualitative info
-difficult to generalise to the wider population due to them being small and exceptions, makes them unrepresentative
-low reliability, relying on recalling past information
-impossible to replicate
What is an aim?
The intended outcome of an investigation, refers to what the researcher is trying to achieve
-Aims in non-experiemental conditions may be broad as qualitative data might make it difficult to generate specific aims
-Aims in experimental investigations where quantitative data is produced, the aim may be to test one or more hypotheses
What is a hypothesis?
A precise, testable statement
Types of hypothesis
1. Null, that the results obtained from the DV are due to chance and not due to the manipulation of the IV
2. Alternate, that the IV has produced the change in the DV
3. Directional, suggests one condition will have a more significant effect on the DV than the other
4. Non-directional, that the manipulation of the IV will affect the DV, but which way is unknown
Experimental designs - independent groups
Each group is allocated to a different condition, this reduces the likelihood of order effects such as practise and fatigue. There are also fewer demand characteristics as ps have less of an opportunity to guess the purpose of the investigation and change their behaviour accordingly. However, individual differences may alter results e.g. age, he alt, profession. May lead to confounding variables.
Repeated measures
The same ps are used in each condition of the experiment. Making it more cost effective and there being fewer extraneous variables between conditions due to using the same ps. Order effects are however more likely to occur but can be combatted using 'counter balancing', alternating the order in which pf perform in an experiment, e.g. A B B A
Matched pairs
Uses different ps in each condition but they are assessed and matched together. This design integrates in dependent groups and repeated measures together to create a happy medium with no order effects or individual differences. Yet it is difficult to match pairs on all relevant characteristics rather than by sex, or age for example.
Naturalistic observation sampling
-Time interval sampling e.g. observing behaviour every 10 minutes but behaviour may be missed.
-Time point sampling e.g. observing behaviour at given points in time.
-Event sampling e.g. observing behaviour at a complete event.
Behavioural categories
-Tally charts, are quick and easy to report a large amount of behaviours.
-Audiotape recordings, accurately record all information for later analysis, however it doesn't capture body language.
-Video, records all data in view but more demand characteristics if ps are aware of the camera.
-One way mirrors, ps cannot see the video recorder and so few/no demand characteristics however if ps are not informed then there could be ethical issues. Even if they were aware of the camera, they may act naturally anyway.
How to operationalise variables?
Identify the variable/s and how they are measured within the context of your study.
Pilot studies
Use a small sample of ps prior to the main study to:
-Make sure all equipment is suitable and working properly
-Ensure all instructions are clear, no ambiguous wording
-Make any necessary amendments
Extraneous variables
Are variables other than the IV that may affect the DV. If they are not controlled then they become confounding variables.
But, in some cases they cannot be predicted and are so difficult to control. A reduction of these makes the experiment more reliable.
Types of reliability
Reliability is the extent to which results or procedures are consistent.
-Internal reliability, how consistently a method measures within itself, can be tested using the split half method, where the test is divided into 2 and the results are correlated, a strong correlation indicates a strong reliability.
-External reliability, how consistently a method measures over time when it is repeated, using the test-retest method similar scores should be achieved.
-Inter-rater reliability, separate researchers observe and get the same/extremely similar results indicates a strong correlation and thus is reliable.
Types of validity
The extent to which something measures what it is supposed to measure.
-Internal validity, how accurately a test or piece of measuring equipment measures what it says to measure.
-External validity, can the findings be generalised outside of the study?
-Face/content validity, does it look like it measures what it is supposed to?
-Concurrent validity, do the results of a new test agree with the results of a more established test measuring the same thing.
BPS Code of Ethics
-Respect, psychologists should treat individuals with respect regardless of individual differences.
-Informed consent, told all details of the experiment before taking part so the ps can commit or opt out.
-Confidentiality, information mustn't be disclosed unless confirmed prior to the study or because it is required by the law.
-Deception, ps must not be misled or have information withheld from them.
-Right to withdraw, this should be made clear before the experiment for both them, and their results.
-Observational research, ps privacy must be protected. When they are unaware that they are being observed this should be done in areas they would expect others to observe them e.g. in a supermarket.
more.....
-Competence, all ethical issues must be considered, if they breach an issue then they must defend their decision to do this, e.g. Milgram claimed people were happy to participate and that the outcome outweighed the distress.
Responsibility
-Debriefing, risk of harm and any foreseeable threats should be eliminated.
-Understanding the implications.
-Protection from stress, assuring ps they they do not need to answer personal questions that may cause stress.
-Inducement, ps shouldn't be offered money to make them take risks.
-Professional advice, ps should be referred e.g. to a therapist if any problems that may have adverse side effects in the future are discovered.
-Non-human animals, animals shouldn't be in pain or discomfort.
Integrity
-Psychologists should be honest, accurate, clear and fair to all ps in all scientific activity e.g. recording and analysing.
-Ps should never feel forced into taking part.
-If psychologists feel a colleague is being unethical, their concerns should be raised and asked to reevaluate their research.
Conclusion on ethics
These guidelines can be difficult to apply and differ across cultures. In Canada, guidelines are more detailed and analyse long-term risks or benefits.
Fully informed consent isn't always obtainable as if a ps knew the intention of the investigation it may affect the outcome of the experiment e.g. Milgram.
By using prior general consent, ps are asked if they would mind being misinformed about the true purpose of the study.
Deception may be overcome by role play, tell ps the general nature of the experiment and ask them to role play as though they were naive.
Sampling
Sampling a large and representative number of ps from the target population.
-Random, every person has an equal chance of being selected. There are less issues of results being biased but it doesn't guarantee to be entirely representative.
-Opportunity, selecting anyone who is available at the time. This is quick and easy but it may result in some form of bias such as age or gender.
-Volunteer, ps have consciously selected themselves to take part in a study. This is convenient and eliminates a researched bias but the majority are unlikely to respond to an advert and so the sample is not representative of the target population.
Demand characteristics
Occur in a laboratory environment when the ps try to work out the true aim of the experiment and behave accordingly to either help the researcher or the sabotage them (the screw-you effect). Ps may act nervously if they are aware they they are being watched.
Using the single blind technique, the ps doesn't know the hypothesis or what condition they are in.
Investigator effects
Occur when the researchers' behaviour during the study can affect its results, consciously or unconsciously, this can lead to a self-fulfilling prophecy. The researcher may be committed to achieving a certain outcome and so analyse data in favour of this. The researcher may influence the ps when answering questionnaires or interviews e.g. nodding. Manipulation or fabrication of data to provide desired results.
Using a double blind test both the ps and the researcher are unaware what conditions they are in.
Levels of measurement
-Nominal, the simplest data with discrete categories e.g. male/female.
-Ordinal, categories can be placed in rank order but the intervals between each rank are not the same e.g. 1-10 for attractiveness, how big is the gap from a 5 to a 7?
-Interval, intervals on the same scale are equal because they are based on a standard unit of measurement e.g. temperature.
-Ratio, the true zero e.g. cm, kg.
Interpretation of quantitative data
-Graphs indicate any recurring patterns.
-Histograms are used when presenting data with interval or ratio levels of measurement. It shows the pattern of a whole data set.
-Bar charts, are used for displaying differences between data and can be used for nominal or ordinal data on the x axis. Nominal is presented in alphabetical order and ordinal in rank order.
-Frequency polygons, are used when you want to show 2 or more data sets on the same graph to directly compare results.
-Scattergrams, are used to show correlational relationships, they help to interpret the correlation coefficient. Show +ve or -ve and the strength of these.
-Tables, show the summary of the data and what the research has found. They help to make patterns in the data obvious. Data can be grouped and presented in a group frequency table.
Descriptive statistics
Summarise a set of quantitative data to one single value that makes the data easy to interpret. Each individual score may be lost using this.
Measures of Central Tendency and Measures of Dispersion.
M of CT
Include mean, median and mode.
-The mean, is powerful and uses all data but it can provide meaningless decimals and can be distorted by anomalies.
-The median, is used for ordinal data, when scores have been placed in numerical order. It only takes the middle value into account and can be unrepresentative in small data sets e.g. 2, 20, 39.
-The mode, is the most frequently occurring value of a data set. In some cases there are 2 modes (bi-modal). Used for nominal data. It is not influences by extreme scores and it gives exact scored. Not valuable with bi/multimodal data sets.
M of D
Show how far the scores vary within data sets, inc. range, interquartile range and standard deviation. This can be only used on interval or ratio data.
-The range is quick, easy and includes extreme values, however it doesn't tell us about the distribution of scores or individual values.
-The interquartile range measures the middle 50% of values when they are in numerical order, from the 25th to the 75th percentile. It doesn't take all values into account but it is more representative of the central, more typical values in a data set.
-The standard deviation measures how far the values in a data set are spread around the mean. It shows how consistently the IV affects the DV. If the spread is low, data is close to the mean. It is more difficult to calculate than the range but it powerful and takes all scores into account.
Presentation of qualitative data
Using coding frames or content analysis to count the number of times an item is mentioned. Researcher needs to ensure all data is objective, assessed using inter-rater reliability.
-Content analysis analyses data according to themes or categories. Organising qualitative data and reducing it to quantitative data.
-The researcher creates a coding frame established during the pilot study and ensures categories don't overlap. The number of times each theme occurs is then counted= easy data analysis and consistent. Also used to identify important information that is often omitted from speech. It is difficult to be completely objective, why chose certain categories and omit others? I-R reliability can be used to check the reliability of the frame and statistical procedures become possible as data is now quantitative.
Other sets by this creator