Research Methods
- Created by: Claire Reynolds
- Created on: 18-10-13 08:50
Experimental Methods
Lab experiments
Where an IV is manipulated. Usually take place in an artificial environment so all extraneous variables can be controlled
+ High control over variables
+ Easy to replicate
- Can lack ecological validity
- Can lack experimantal validity
- Possible ethical issues
Experimental Methods
Field experiments
Where an IV is manipulated, but they take place in the participants' usual environment. Some control over extraneous variables
+ Improved ecological validity
+ Reduction of demand characteristics
- Difficult to replicate
- Less control over variables
- Possible ethical issues
Experimental Methods
Natural Experiments
IV isn't manipulated - naturally occuring one is used. Little control over extraneous variables. Takes place in a natural setting
+ High ecological validity
+ Few demand characteristics
+ Investigates situations not usually possible due to impractical/ethical reasons
- Very little control over variables
- Very hard to replicate
- Possible ethical issues
Experimental Methods
Correlational Analysis
Look to see if there is a relationship between two co-variables. this can obtain the direction (positive or negative) and the stength of any relationship/correlation.
+ Allows researchers to undertakes exploratory research into an unknown area to see whether a relationship exists
+ Possible to conduct when using other methods would be unethical
- Can't determine ccause and effect
- Correlations can only identify linear relationships
Observational Technique
Naturalistic Observations
+ Higher levels of ecological validity
- No control over extraneous variables
Controlled observations
+The researcher has greater control over the research situation
- The artificial situation may influence participants behaviour
Observational Technique
Participant observations
+ Researcher gets first hand insight into peoples natural settings
+ Higher levels of ecological validity
+ Rich, detailed data collected
- Researcher may become too involved, finding it difficult to remain objective.
- Presence of observer may change particpants' behaviour
Non-participant observations
+ Researcher is more likely to remain emotionally detached and therefore objective.
- The actual meaning of behaviour may not be clear from a distance
Observational Technique
Overt observations
+ More ehtical as participants know they are being observed
- Participants are aware they're being watched, so likely to show demand characteristics
Covert observations
+ Participants aren't aware they are being observed, so less likely to show demand characteristics
-Less ethical as participants don't know they're being observed.
Questionaires and Interviews
Questionaires
Questions can be:
Open - produce rich, detailed, qualitative data, but analysis can be difficult.
Closed - produce quantitative data, which can be analysed statistically, but may not be realistic
+ Large amounts of data can be collected quickly
+ Produce less investigator bias than interviews
+ Can be less time-consuming
- Social desireability bias
- Sample bias
- Ethical issues
Questionaires and Interviews
Interviews
Interviews can be structured, unstructured or semi-structured.
The behaviour of an interviewer towards a participant may influence responses
The appearance of an interview can influence how comfortable the participant feels
+ Can provide detailed information
+ Good rapport with participants
- Interpretation of data can be subjective
- Time-consuming
- Investigator effects
Case Studies
Case Studies
Involve a detailed study of a particular individual/group
-information is gathered from a range of sources, through a range of techniques
All case studies are very unique. Researchers tend to focus on one aspect of behaviour, collecting qualitative data.
+ Detailed data is collected
+ Allows researcher to gain insight into situations which couldn't be 'created' by researchers
- Replication is difficult
- Lack population validity
- Investigator bias
- Ethical issues
Experimental Design
Repeated measures
The same participants are used in each condition
+ No participant variables
+ Fewer participants required compared to other designs
- Order effects - boredom, practise, fatigue.
- Demand characteristics are more likely
- Different tests/matericals maybe required for each condition
Experimental Design
Independent groups
Different participants are used in each condition
+ No order effects
+ Lower risk of demand characteristics
+ Same tests/matericals can be used in all conditions
- Participant variables
- More participants are requires compared to repeated measures
Experimental Design
Matched pairs
Different participants are used in each condition, but participants are matched on age, gender etc
+ Fewer participant variables
+ No order effects
+ Lower risk of demand characteristics
+ Same tests/materials can be used in each condition
- Participants cannot be matched on every level therefore there are some participant variables
- Matching is difficult and time-consuming
- More participants are required than in repeated measures
Variables
Independent variable
The variable that can be changed/manipulated, usually by the researcher
Dependent variable
The variable being influenced by the independent variable, which can be measured
Extraneous variable
Variable other than IV which affects the DV
Co-variable
Two variables which are examined to see whether a correlation exists between them
Variables
Control of extraneous variables
Participant variables are related to the participants involved - only an issue when different particpants are used in each condition
Can be overcome with the use of repeated measures or matched pairs design
Situational variables are related to the research situation and may influence participant behaviour e.g. temperature, noise, time of day, demand characteristics etc.
Can be overcome with the use of standardised procedures
Demand characteristics can be overcome by using a single-blind design
Investigator effects can be overcome using a double blind design
Order effects can be overcome using counterbalancing
Reliability
Internal Reliability
A measurement of the extent to which something is consistent within itself
External Reliability
A measure of consistency over several different occasions. If an experiment is repeated, are the same results obtained?
Inter-observer reliability
When two observers are observing the same behaviour, are the results the same?
Validity
Internal validity
Concerns what goes on inside the study - the extent to which the changes seen in the DV are due to the IV
External validity
Concerns what goes on outside the study - the extent to which the findings can be applied to other settings (ecological) and other people (population validity)
Ethical Issues
Informed consent
Can be overcome by prior general consent, presumptive consent or obtaining fully informed consent
Deception
Can be overcome by prior general consent and debriefing
Right to withdraw
Can be overcome by reinforcing the right to withdraw
Confidentiality
Can be overcome by avoiding collecting personal details from participants
Protection from harm
Can be overcome by terminating research, debriefing or reinforcing the right to withdraw
Sampling techniques
Random Sampling
a sample where every menber of the target population has an equal chance of being selected for the sample - numbers are assigned to each member then pulled out of a hat
can you generalise? yes, but not if sample size is small
is it biased? no, researcher has no input
+ More likely to be representative than opportunity sampling
+ Less likely to be biased than opportuniy sampling
- More time consuming that opportunity sampling
- If taget population is large and sample is small, the sample may be unrepresentative
Sampling techniques
Opportunity Sampling
a sample which contains participants who are available to the researcher. Researcher decides on the type of participants needed and approaches anyone who appears suitable and willing until they have enough.
can you generalise? No, as only suitable people are used
is it biased? Yes, researcher selects participants
+ Relatively quick compared to random sampling
- Less likely to be representative than random sampling
- More likely to be biased than random sampling
Sampling techniques
Volunteer Sampling
A sample where participants volunteer to take part in research. Researcher advertises and participants respond by answering the advert or replying to a postal questionnaire.
can you generalise? No, not representative
is it biased? Depends how participants are selected if response is larger than sufficient numbers needed
+ Easier to conduct than random sampling
- More likely to be biased than random sampling
Data Analysis
Measures of cantral tendency
provide an average score for the data
Mean
+ Most sensitive of all measures of central tendency
- Affected by outliers
- Not appropriate when category data has been collected
Median
+ Not affected by outliers
- Doesn't take all scores into account
- May be unrepresentative if there are few scores
Data Analysis
Mode
+ Not affected by extreme scores
+ Can be used with all types of data
- Not useful when there are many or no modes
- Doesn't take all scores into account
Data Analysis
Measures of dispersion
provide information about how the values in a set of data are spread out
Range
+ Gives basic indication of the spread of scores
+ Easy to calculate
- Distorted by extreme scores
- Doesn't indicate how scores are grouped around the mean
Standard Deviation
+ Most accurate measure of deviation
- Can be difficult to calculate
- Cannot always be calculated depending on data
Data Analysis
Correlational Analysis
The strength of the relationship between two variables is measured by a statistical value between -1 and +1 called the correlational coefficient.
- the closer the coefficient is to +/- 1, the stronger the relationship. The closer it is to 0, the weaker the relationship.
- A coefficient of 0 indicates no relationship
Data Analysis
Qualitiative Data
involves collecting any analysind peoples meanings, experiences and descriptions. Data is usually in the form of verbal/written descriptions. It can be colected and presented in the following ways:
- content analysis
- coding units- involves clearly labelling or identifying different elements of the behaviour a researcher wants to study.
- categorising - grouping similar items (coding units) together
- quotations - word for word quotes which typify what others have said during research
+ Data collected allows meaning behind data to be analysed
+ Conclusions drawn can be assessed for reliability and validity
- Data analysis tends to be subjective, so data may be interpreted differently
- Can be very time-consuming to analyse large amounts of data collected
Data Analysis
Content Analysis
a form of indirect observation- researcher isn't observing people directly but the artefacts they've produced, e.g. books,magazines, TV programmes, films, diaries, songs etc.
Sampling: researc her must decide what to sample, e.g. newspapaers- make a decision based on political leaning, target reader etc, or TV programmes - taking representative sample, times, advert slots etc.
Behavioural categories: Involves deciding what categories can be used and how they'll be dealt with, there are two options:
- quantitative analysis- examples in each category are counted
- qualitative analysis- examples in each category are described
Researcher must consider inter-rater reliability
+ High ecological validity
+ As long as sources are kept, reliability can be assessed
- Possibilty of observer bias
- Cultural bias
Designing Investigations
What to include:
Methos
Aim/hypotheses
Variables
Extraneous variables
Sample
Ethics
Procedure
Materials
Analysis of data
Sampling techniques
A target population is the target group that a researcher wishes to study and to which they generalise their findings
A sample is a group taken from the target population that is studied. The sample should:
- be representative so that generalisations can be made to the target population
- not be biased by the researcher selecting certain participants as this prevents making generalisations to the target population
Ethical considerations
Ethical issues
**** when there is a dilemma between what the researcher wants to do in their psychological study, and the rights and best intrests of the participants
BPS Code of Conduct
Respect - includes expectations relating to informed consent, privacy/confidentiality and the right to withdraw
Competence - psychologists ensure high standards in thgeir research
Responsibility - psychologists have a responsibility to their clients, the general public and to psychology.
- refers to standards relating to protection from physical and psychological harm and also debriefing
Integrity - psychologists should be honest and accurate in reporting their research
- they should report examples of misconduct by other psychologists to the BPS
Ethical considerations
Dealing with ethical issues
- Adhering to the BPS code of conduct
- Discussing research proposals with an ethics comittee to decide whether procedures are ethical
- Conducting a cost-benefit analysis whereby potential costs (participant distress etc) are weighed against potential benefits (valuable results)
- If a psychologist does conduct unethical research, the BPS can enforce punishments, e.g. preventing researcher from practising psychology
Ethical considerations with non-human animals
Why do researchers study non-human animals?
- Animals may be studies because they are fascinating in their own right
- Animals offer the opportunity for greater control and objectivity in research procedures. Most of the behaviourist theory was established using animals
- We can use animals when it's not possible to use humans - they can be exposed to various procedures and events not possible with humans
- humans and animals have sufficiently similar physiology to justify studying non-human animals and generalising findings to humans
Ethical considerations with non-human animals
Objections to using non-human animals
- Singer (1990) believes we have no moral right to elevate the rights of one species above another - this is committing 'speciesism'
- Animals' rights: Regan (1984) believe that the traditional scientific position on animal research treats animals as 'renewable resources' rather than organisms of value with rights
- Animals are different to humans - so it's not possible to generalise findings from research on animals to humans, which some researchers attempt
Ethical considerations with non-human animals
Constraints based on use of animals in psychological research
- Legal constraints - the animals act (1986) states that animal research must:
- take place in laboratories licensed for animal research
- be part of an approved research project
- be carried out by properly trained people who have a personal license to conduct such research
- The Three Rs
- Reduce the number of animals used
- Refine reseach methods to minimise suffering
- Replace animal research with computer simulations, brain scanning etc
- BPS Guidelines for psychologists working with animals - psychologists are advised to show committment to legislation and the 'Three R's'
Reliability
Reliability
concerned with consistency. It refers to how much we can depend on a measurement.
Types of reliability
- Internal reliability - is a measure of the extent to which something is consistent within itself. In a questionnaire/interview, all the questions should be measuring the same behaviour. In an experiment, all participants should be treated the same way.
- External reliability - is a measure of consistency over several different occassions. If an investigation is repeated, the same results should be obtained.
- Inter-rater reliability - is a measure of the similarity of two sets of results from the same investigation. When two observers are involved in an observation, their results should be the same.
Reliability
Assessing reliability
Reliability can be assessed in the following ways:
-
- Split half method which assesses internal reliability. Data from one half of the test (e.g. questionnaire) is correlated with the data from the other half. If they are similar, they will show a strong positive correlation and the test has internal reliability
- Test-retest method which assesses external reliability. They study is repeated with similar or the same participants. If similar results are obtained, the study has external reliability.
Reliability
Improving reliability
Reliability can be improved by:
-
- using standardised procedures
- training all observers in the coding/rating system in observations
- ensure all questions are carefully worded/written
- take more than one measurement from each participant and record the average score
- conduct a pilot study
Validity
Validity
is concerned with whether the researcher is measuring what he/she is claiming to measure
Types of validity
-
- Internal validity is concerned with what goes on inside the study. Whether the DV is due to the IV. Factors that threaten internal validity
- poorly operationalised variables
- extraneous variables
- demand characteristics
- experimenter bias
- External validity is concerned with what goes on outside the study. It refers to the extent to which the findings can be generalised beyond specific research study - to other settings (ecological validity) and other people (population validity)
- Internal validity is concerned with what goes on inside the study. Whether the DV is due to the IV. Factors that threaten internal validity
Validity
Assessing validity
Validity can be measured in the following ways:
- Face validity involves getting an expert to look at the study and confirm that it does measure what it claims to measure
- Predictive validity refers to whether the findings will predict future performance. e.g. GCSEs predict A- Levels. However, only applicable if they correctly predict.
- Construct validity assess whether a study really measures the 'construct' it's supposed to be measuring, e.g. do IQ tests really measure intelligence? They don't measure social ability, so do not have construct validity.
- Concurrent validity involves comparing a new method or test with an already well established one. A strong positive correlation suggests they're measuring similar characteristics.
Validity
Improving validity
- Internal validity - only in experiements
- use scales or measurements that had already been assessed as having validity
- ensure studies are carefully controlled - all variables except the IV need to be kept constant
- use single-blind technique
- use double-blind technique
- External validity
- use everyday tasks to improve ecological validity
- use a representative sample to improve population validity
Application of scientific method
Science is generally seen to be a trustworthy body of knowledge and the method of obtaining that knowledge
Features of a science
-
- Objectivity - scientists aim to be objective and conduct research free from opinion or bias
- Replicability - the ability to check and verify scientific information. It should be possible to replicate scientific investigations so that other researchers can check any conclusions made. If similar results are obtained when studies are repeated, there is confidence in research findings. If not, this implys flaws/lack of control in the method used and they cannot therefore be used in theory construction.
- Theory construction - science aims to collect facts and use them to generate theories - which are used to explain and make predictions about behaviour
- Use of empirical methods - science uses empirical methods of observation and measurement that rely on direct sensory information
- Hypothesis testing
Application of scientific method
The scientific method
1. Researcher identifies a problem
2. A hypothesis is proposed (must make a prediction and be testable)
3. A study is designed to test the hypothesis
4. Results are analysed to see if they support the hypothesis
5. Modify and repeat process in light of whether the hypothesis was supported or not
6. Develop a theory
Application of scientific method
Is the scientific method appropriate for psychology?
Scientists critisise the scienctific method because:
- People cannot be studied in the same way as physical phenomena- people interact with researchers, inanimate objects don't
- Conventional research tends to involve studying people in artificial environments - so therefore the behaviour observed is unnatural
- The scientific method emphasises controll, so tries to study one variable in isolation. This can be impossle in humans - e.g. we cannot study memory in isolation from a persons feelinds/past experiences
- Focus on empirical methods means people are treated as passive participants whereby their thoughts/feelings are ignored - making data collection superficial
- Some claim objectivity is impossible - researcher cannot put aside their beliefs/values when designing a study or analysing data from it
Peer review
Peer Review
the process by which psychological research papers, before publication, are subjected to independent scruitiny by other psychologists who are experts in the same field. Peers consider research in terms of it's validity, significance and originality. It's central to validating new knowledge.
1. Research is conducted
2. Researcher prepares a manuscript of the study and sends it to a journal
3. Editor of the journal sends the manuscript to experts in the topic, who undertake peer review.
4. The experts assess all aspects of the study and then return the manuscript to the editor, with comments and reccommendations.
5. Based on these comments and reccommendations, the editor decides whether the research should be:
-
- accepted for publication
- revised and resubmitted for peer review
- rejected
Peer review
Why is peer review important?
- It's difficult for authors and researchers to spot every mistake in a piece of work. Showing work to others increases the probability that weaknesses will be identified and addressed.
- It prevents dissemination of irrelevant findings, unwarrented claims, unacceptable interpretations. personal views and deliberated fraud.
- Peer reviewers judge the quality and significance of the research in a wider context.
- This process ensures published reseach can be taken seriously because it's been scrutinised.
Peer review
Problems with peer review
- in a small number of cases, it doesn't detect fabrication or miscommunication
- research that doesn't 'fit' with previous work is often seen as suspect and wiill be rejected
- peer review is subject to bais by the reviewer. There is some evidence of institutional bias (favouring prestigious universities/male researchers)
- Peer review tends to favour research that has supported the research hypothesis. If it's rejected it, it tends to be discarded.
Comments
No comments have yet been made