Research MethodsHow do you choose which method
is best for your study?
Experimental Methods• Laboratory experiment:
Increases reliability, but reduces external validity. It tends to be high in internal validity.
• Field experiment: Increases Ecological validity, but
Extraneous variables can be a problem.
• Natural experiment: No causal conclusions can be made.
Reduced validity = no controls
Non-Experimental Methods Questionnaire and Interview: Open ended question = rich in detail,
however difficult to analyse. Problems with social desirability.
Observational: Rich in detail, high ecological validity however possible observer bias.
Correlation: Lots of extraneous variable = low internal and external validity. No cause and effect can be made.
Sampling MethodsYou need to understand the implications of choosing the appropriate method for sampling
Sampling MethodsTarget Population: The population is a large
sample of people or information from which you draw a proportion to study.
Population Validity: The people chosen for the study
are relevant to the hypothesis
•Random sampling • has high population validity
•Opportunity sampling • There is a high chance the
sample will be biased, leading to low Population validity
•Volunteering: Only a certain type of people respond to adverts and therefore there is bias leading to low population validity
• Stratified samples: A quota of individual is taken to represent the population dynamics. If there is 40% men and 60% women in the target population, then the sample must contain the same = more representative
• Snowball Sampling • Limited sample ask those you • do have to direct you to others • Limited selection of population, • therefore bias.
•Reliability or consistency refers to whether the research can be repeated with similar findings.
• If the findings can be consistently replicated then the outcome is seen as reliable.
• There are three types of reliability• External• Internal• Observer
Reliability• External: To assess external reliability, you use the
test-retest method.
• Internal: To assess internal reliability, you use the split-half method.
•Observer• To improve observer
reliability you train the observers, ensuring That they understand each category.
Inside the experiment: Where changes in DV due
to IV, or something else?
Was the task representative (mundane realism)?
Are the findings meaningful?
Outside the experiment:
• Can you generalise the findings to other people?
• To other settings?
• Are the findings useful?
Validity
Internal External
Validity
Laboratory Vs Field• Laboratory: If the tasks in a lab
experiment are not true to life, like in memory; then it’s low in validity.
• Field: Sometime field experiments don’t represent real life; i.e. Hofling’s nurse study in obedience.
Observational techniques: Internal validity is low due to observer bias. Ecological validity may be high.
Self Reporting techniques:Assessing internal validity
Face validity: does it measure what it claims to measure.
Concurrent validity: Compare with a previous questionnaire or test.
Ethics
Research needs human participants, therefore it needs to maintain participants confidence in the research methods.
Ethical Issues
• Deception: Only acceptableWhen the integrity of research would be compromised.
• Informed Consent: Participants need significant information about a study.
• Protection of participants from psychological harm: Humiliation, embarrassment and loss of dignity or self-esteem
Ethical Guidelines• Only Guidelines: There are not hard and fast rules but
only guidelines.
• Implemented:
• The BPS impose penalties for not following ethical guidelines
It is sometimes difficult to monitor all research. Psychologists must monitor their colleague work.
Non-Human animals• Animals are used to
increase control of an experiment and when unethical to use humans.
• Sentient beings: • Do animals have conscious to feel pain?
• Speciesism: Peter singer argued that animal are discriminated like sexism or racism.
Ethical Guidelines• The BPS published guidelines
for animal research. The Animals Scientific Procedures Act (1986) requires an license beforeresearchers can use animals.
• Russell and Birch (1959). The three Rs Reduction–Replacement and Refinement
What is Science
Research is more than just subjective ideas
Major Features of ScienceObjective: Expectations
don’t affect results. Control and Replicability
Empiricism: Info from direct observation or experiments what do the images tell us about empirical evidence?
Theory construction: Record facts and construct theories.
The Scientific Process• Induction: Conducting observation, devising hypothesis then
drawing conclusions and a theory
•Deduction: • Making a theory based on • observation. Testing it to • support the theory.
• The Hypothetico-• Deductive model: • Karl Popper (1935) Theories – Generalise – Falsified.
Application of scientific methods
The presentation of verifiable knowledge as distinct from common
Sense
scientology - Google Videos
• Pseudoscience: Miller ( 1983) argued psychology isn’t a true science.
• Pre-science: Kuhn (1962) Psychology has a number of approaches (Paradigms = a shared set of assumptions.)
• Objectivity: Heisenberg (1927) argues that even experiments in physics change the particles they are measuring.
Is science appropriate for Psychchology?
• Idiographic approach: R.D.Laing (1965) people are individuals and not subject to generalisation.
• Nomothetic = study of groups which is more scientific.
• Reductionist: Behaviour is Reduced to the simplest explanation, this tell us little about real behaviour.
• Determinist: cause and effect = can oversimplify relationship.
Constructing a Report
The role of peer review.
Conventions for reporting a
Psychological report
Peer Review• Allocation of funding: Government needs to
decide who gets funding. Needs advice from public body reviews.
• Publication: Peer review helps monitor research before it is published.
• Assessing Universities: Research assessment exercise RAE rates each university.
Template for Reports
• Research reports tend to have a fixed structure.
•Abstract: A short summary
• Introduction: A review of • previous research leading to the aim of the
project
•Method: A detailed description of procedures
Template for Reports• Results: The statistical
data, including tables, measure of central tendency and graphs.
• Discussion: Of outcomes and suggestionsfor future work and conclusions.
• References: Full details of journal articles and books used.
Evaluation of the use of Reports
• Unachievable ideal: Finding experts to do the peer review is difficult, under-qualified.
• Anonymity: Rival competition may do the review in secret = open reviews are best.
• Publication bias: Editors have preferences for more positive reviews to heighten sales.
Paradigm shifts Responsibility: Psychologists shapesocial policy. Burt (1955) discredited on genetic resource. At the time he was promoting 11+ for brighter children
Preserving the status quo: We are afraid of change and resistant to large shifts.
Normal Science: One view is dominate through small change and it take a paradigmshift for the next
view to replace it. Kuhn Paradigm Shift
Design your own study• There may be a 12 mark question in the exam asking you to
design your own study from a small scenario.
• Activity: Get into Groups of 4
• Take one of the Scenario from page 16 – 17, each group will take a different Study.
• Complete the activity
• Two from each group will move to the next group and describe how they designed their study. The two remaining members will then explain how their group designed the study
Inferential Analysis
The probability that the results could have arisen by chance.
Significant Difference• What we aim in an experiment is to demonstrate that there is
a sufficiently large (significant) difference between samples.
• We use a test to state that the probability of the results being due to chance is greater than 0.5%
• Nominal: Data is placed in categories.• Hypothesis: Female students will get more A* grades than
male students.
• Ordinal: Data that is put in order. i.e. lining up the class members in accordance with their height.
• Interval:
Males Females Total
B grades 4 4 8
Below B 3 5 8
Total 7 9 16
• Level of Significance for a one-tailed test• 0.1 0.05 0.024 0.01
• Level of significance for a two-tailed test• 0.20 0.1 0.05 0.02
• df: • 1 1.64 2.71 3.84 5.41• 2 2.6 4.6 5.99 7.3• 3 4.3 6.25 7.82 8.9• 4
Rational for Choosing a Test• 1: What is the level of data: Nominal, Ordinal or Interval
• 2: Are you looking at a Correlation or a Difference
• 3: What design does the study use: Repeated Measures or Independent groups Design
• Nominal = Chi Squared Test• Correlation = Spearman’s Rho Test• Independent groups design = Mann-Whitney U Test• Repeated Measures/Matched Pairs = Wilcoxon T Test