Research Methods Flashcards
Aim
General statement of what the researcher intends to investigate
Hypothesis
Clear and precise statement stating relationship between variables investigated
Prediction of what will happen
Directional hypothesis (one-tailed)
Clear difference is made, e.g., people who drink energy drink become more talkative than people who don’t…
Non-directional hypothesis (two-tailed)
States a difference but the nature of difference is not specified, e.g., people who drink energy drink differ in terms of talkativeness…
When to use a one-tailed
Findings of previous research studies suggest particular outcome
When to use a two-tailed
No previous research/findings from earlier studies
IV
Some aspect of the environmental situation that is manipulated by the researcher/changes naturally so DV effect can be measured
DV
Variable that is measured by the researcher. Effect caused by IV
Levels of IV
Control condition = leaving things how they are, e.g., no energy drink/drinking water
Experimental condition = changing circumstances, e.g., drinking energy drink
Operationalisation
Clearly defining variables in terms of how they can be measured (numerical), e.g., after drinking 300ml of energy drink ptpts say more words in the next 5 min…
Extraneous variables
Any variable other than the IV that may have an effect on the DV if it isn’t controlled; do NOT vary systematically with IV, e.g., lighting
Confounding variables
Any variable other than the IV that may have an effect on the DV; DO vary systematically with IV
Demand characteristics
Any cue from the researcher/situation that may be interpreted by ptpts as revealing the purpose (leads to behaviour change)
Investigator effects
Any effect of the investigators behaviour on the research outcome, e.g., leading questions
Randomisation
Use of change in order to control for the effects of bias when designing materials; deciding order of conditions
Standardisation
Using exactly the same formalised procedures and instructions for all ptpts in a research study
Independent groups
Two separate groups in two separate conditions
Repeated measures
Only one group of ptpts taking part in both conditions
Matched pairs
Two separate groups matched on certain qualities
+/- of Independent groups
+ Order effects
- Individual differences
+/- of Repeated measures
+ Ptpt variables controlled (fewer needed)
- Order effects
+/- of Matched pairs
+ Order effects not an issue
- Ptpts cannot be matched exactly; time-consuming
Lab experiment
Controlled environment - researcher manipulates IV and records effect on DV
Field experiment
Natural setting - researcher manipulates IV and records effect on DV
Natural experiment
Change in the IV is not brought about by researcher but would’ve happened regardless
Quasi experiment
Almost an experiment but lacks key ingredients: IV has not been determined by anyone - variables simply exist, e.g., being old/young
+/- Lab experiment
+ High control over extraneous variables; replication more possible
- May lack generalisability
+/- Field experiment
+ Higher mundane realism
- Ethical issues - consent
+/- Natural experiment
+ High external validity
- Naturally occurring event may only happen rarely
+/- Quasi experiment
+ Often controlled conditions - high control over extraneous variables
- Cannot randomly allocate ptpts to conditions
Random sampling
All members of target population have equal chance of being selected (lottery method)
Systematic sampling
Every Nth member of the target population, e.g., every 5th pupil on a school register
Stratified sampling
Researchers divide subjects into subgroups based on characteristics they share. Once divided, each group is randomly sampled
Opportunity sampling
Researchers select anyone who happens to be willing/available
Volunteer sampling
Ptpts select themselves to be part of the sample
Random s +/-
+ Free from researcher bias
- Time-consuming/difficult to conduct (complete list may be difficult to obtain)
Systematic s +/-
+ Avoids researcher bias
- Fairly unrepresentative, e.g., could be an all-male sample
Stratified s +/-
+ Avoids researcher bias
- Identified strata cannot reflect all the ways that people are different
Opportunity s +/-
+ Less time-consuming
- Unrepresentative; researcher bias
Volunteer s +/-
+ Minimal input from researcher - less time-consuming
- Volunteer bias
BPS code of ethics
Instructs psychologists about what behaviour is and isn’t acceptable when dealing with ptpts
Respect, competence, responsibility; integrity
Informed consent & way of dealing
Making ptpts aware of aims, procedures & rights
* consent form/signature
Protection from harm & way of dealing
Ptpts should not be placed at risk
* counselling
Deception & way of dealing
Deliberately misleading/withholding info
* debrief; right to withhold
Confidentiality & way of dealing
Right to control info/remain private
* anonymity, e.g, initials
Pilot study
Small-scale version of investigation before the real one
Allows problems to be identified
Single-blind procedure
Researchers do not tell ptpts if they are being given a test/control treatment
Double-blind procedure
Neither the ptpt nor the experimenter know who is receiving a particular treatment
Single-blind procedure +/-
+ Avoids demand characteristics
- Experimenter bias
Double-blind procedure +
+ Prevents bias/placebo effect
Naturalistic observation
Watching and recording the behaviour in the setting it would normally take place
Controlled observation
Watching and recording behaviour in a structured environment, e.g., lab
Covert observation
Ptpts are unaware their behaviour is being recorded and watched
Overt observation
Ptpts are aware their behaviour is being recorded and watched
Participant observation
Researcher who is observing is part of the group being observed
Non-participant observation
Researcher observes from a distance; is not a part of the group
Naturalistic obs +/-
+ High external validity
- Replication can be difficult
Controlled obs +/-
+ Easy replication
- Low mundane realism
Covert obs +/-
+ Natural behaviour is recorded (high internal validity)
- Ethical issues (no consent)
Overt obs +/-
+ Ethically acceptable (consent)
- Demand characteristics
Participant obs +/-
+ Can be more insightful
- Researcher may lose objectivity
Non-participant obs +/-
+ Can be more objective
- May lose some valuable insight; observer bias
Structured observation
Researcher quantifies what they are observing using predetermined list of behaviours/sampling methods
Unstructured observation
Continuous recording - researcher writes down everything they see
Structured obs +/-
+ Quantitative data
- Not much depth of detail
Unstructured obs +/-
+ Rich in detail
- Qualitative (difficult to analyse; can’t be used in stats test table)
Behavioural categories
Breaking target behaviour into components/checklists, e.g., affection: kissing, hand holding, smiling
Must be measurable/observable
Time-sampling
Recording behaviour within a pre-established time frame
Event-sampling
Counting a number of times a behaviour occurs
Time-sampling +/-
+ Reduces number of observations (less time-consuming)
- Unrepresentative of whole observation (behaviour may be missed)
Event-sampling +/-
+ Behaviours are not missed
- More time-consuming
Questionnaires: open questions
Does not have a fixed range of answers (respondents free to answer in any way they wish) - qualitative
Questionnaires: closed questions
Fixed number of responses - can be qualitative/quantitative
Questionnaires +/-
+ Cost-effective & straightforward to analyse
Can gather large amount of info quickly
- Demand characteristics
Response bias - social desirability
Structured interviews
Pre-determined set of questions asked in a fixed order (q+a pattern)
Structured interviews +/-
+ Straightforward to replicate
- Not possible for interviewers to deviate from topic
Unstructured interviews
Works like a conversation - no set questions; an aim that a certain topic will be discussed
Unstructured interviews +/-
+ More flexibility - interviewees can go in depth
- May lie for social desirability/problems of replication
Semi-structured interviews
Falls between structured & unstructured, e.g., a job interview. Set list + free-flowing
Likert scale questionnaires
Respondent indicates their agreement with a statement of usually 5 points, e.g., ‘zombie films can have educational value, 1- strongly agree, 5- strongly disagree’
Rating scale questionnaires
Respondents identify a value representing their strength of a feeling about a topic, e.g., ‘how entertaining do you find zombie films? 1- very, 5- not at all’
Fixed choice option questionnaires
Includes list of possible options; respondents required to indicate those that apply, e.g., ‘why do you watch zombie films: tick all those that apply’
Interview schedule
List of questions intended to cover
Should be standardised for each ptpt to reduce the effect of investigator bias; interviewer will take notes
Designing interviews
Usually interviewer and single ptpt
Can be a group
Interviewer should conduct in a quiet room away from others
Writing good questions: overuse of jargon
Simple and easily understood, e.g., ‘do you agree that maternal deprivation in infanthood leads to affectionless psychopathy?’
Writing good questions: emotive language & leading questions
Attitudes towards a topic are made clear through ways of phrasing; leading questions can result in unreliable answers
Writing good questions: double-barrelled questions
Contains two questions in one. Issue ? respondents may agree with one half and not the other
Writing good questions: double negatives
Unstraightforward ways of asking questions, e.g., ‘I am not unhappy in my job (a/d)’
Correlation
Illustrates strength and direction of an association between two/more co-variables
Positive correlation
As one co-variable increases, so does the other
Negative correlation
As one co-variable increases, the other decreases
Zero correlation
No relationship between the co-variables
Curvilinear relationship
As one variable increases, so does the other but only up to a certain point as one variable begins to increase and the other begins to decrease, e.g., Yerkes-Dodson Law
Correlations +/-
+ Quick/economical to carry out
Secondary data can be used in correlational study
- Difficult to establish a cause-and-effect
Third variable problem
Qualitative data
Displayed in words; non-numerical
Qualitative data +/-
+ Depth of detail; allows ptpts to develop opinions
- Difficult to analyse; make comparisons with other data
Quantitative data
Displayed numerically; not in words
Quantitative data +/-
+ Can be analysed statistically/converted to graphs
- Lack of depth in detail
Primary data
When info is obtained first hand by the researcher for an investigation
Primary data +/-
+ Targets the exact info which the researcher needs so the data fits their aims
- Time-consuming/expensive
Secondary data
Info is collected by someone other than the researcher but used by them
Secondary data +/-
+ Data is accessed so requires minimal effort to collect
- Data may be outdated/incomplete/unreliable
Meta-analysis
Researcher combines results from many different studies; uses all the data to form an overall view of the subject investigated
Meta-analysis +/-
+ More generalisability (larger amount studied)
- Publication bias (file drawer problem)
Measures of central tendency
Mean, median; mode
Mean
Total of all values divided by number of values
Mean +/-
+ Good for interval data/makes use of all values
- Influenced by extreme scores so can be unrepresentative
Median
Arrange data from lowest to highest then find central value
Median +/-
+ Good for ordinal data/not affected by extreme scores
- Not as sensitive as mean; doesn’t use all data
Mode
Most frequently occurring value in a set of data
Mode +/-
+ Useful for nominal data
- Not useful when there are several modes
Measures of dispersion
Range & standard deviation
Range
Minus lowest score from highest score
Range +/-
+ Easy to calculate
- Does not use all data/affected by extreme values
Standard deviation
Low SD = more data is clustered close to the mean hence there is less data spread
Standard deviation +/-
+ Precise measure where all data values are considered
- Difficult to calculate; affected by extreme values
Bar charts
Describes data divided into categories
Histograms
Represents that we are dealing with continuous data
Scattergrams
Used to show associations between co-variables
Line graphs
Points are connected by lines to show change of values
Normal distribution
Symmetrical pattern forming a bell-shape
Skewed distribution
Spread of frequency that is not symmetrical; all data clusters to one end
When only can a sign test be used?
- Looking for a difference not an association
- Using a related experimental design
- Nominal data
How to conduct a sign test
- Convert data to nominal
- Subtract score 2 from score 1. If less than = -, if more than = +, if the same = leave
- Add up + and -
- Less frequent sign = S
- Compare calculated value with critical value
Rules for a sign test
If S is less than or equal to critical value = significant difference
If S is more than or equal to critical value = no more significant difference
Standard level of significance
0.05
Peer review
Assessment of scientific work by experts in the same field to make sure all research intended to be published is of high quality
Main purposes of a peer review
- Knowing which research is worthwhile so funding can be allocated
- Validate relevance and quality to prevent release of fraudulent research
- Suggest possible improvements
Weaknesses of a peer review
Anonymity might mean rival researchers can be easily criticised
Publication bias
Can be difficult to find an expert
Implications of psychological research for the economy: Psychopathology
Treatments: CBT and REBT for depression, drug therapy, and OCD
Economy: Workers able to return to work
Implications of psychological research for the economy: Attachment
Role of the father: Fathers can take on role of PCG
Economy: Mothers can return to work/maximised income
Implications of psychological research for the economy: Social
Social change: Minority influence, appealing to NSI, disobedient models
Economy: Health campaigns, environmental campaigns, unions strike
Implications of psychological research for the economy: Memory
EWT: How leading questions/PED affect EWT
Economy: Led to police using cognitive interview reducing wrongful convictions
Case studies
Detailed study into the life of a person covering backgrounds
Builds a qualitative case history
Case studies +
Detailed (in depth insight)
Forms basis for future
Case studies -
Not generalisable
Time-consuming/difficult to replicate
Content analysis
Studying behaviour indirectly by studying things we produce, e.g., TV ads/newspapers
How to conduct a content analysis
- Identify hypothesis
- Create coding system, e.g., 1= m, 2 = f
- Gather resources
- Conduct analysis, record data in table
- Analyse data which is descriptive/qualitative
- Write up a scientific report
Content analysis +
Strong external validity as already in real world (high mundane)
Easy replication
Content analysis -
Observer bias
Content of choice to analyse can be biased by researcher
Internal reliability
How consistent something is within itself
External reliability
Consistent results are produced regardless of when the investigation is used/who administers it
Split-half method
Randomly select half of questions; put them in one form then do the same for others
Test-retest
Researcher administers same test on same person different occasions
Inter-observer reliability
Extent to which there is an agreement between two or more observers involved in observing behaviour
Eliminates subjectivity bias
Improving reliability in questionnaires
Replace open questions with room for misinterpretation with closed, fixed choice alternatives
Improving reliability in interviews
Use the same properly trained interviewer & follow a structured interview
Improving reliability in experiments
Lab experiments - strict control over procedural aspects such as conditions tested in
Improving reliability in observations
Making sure behavioural categories have been properly operationalised & that they are measurable/don’t overlap
Internal validity
Whether outcomes observed are due to the manipulation of the IV and not any other factor
External validity
Factors outside the investigation; is generalisable
Ecological validity
Generalisability to other places/settings
Population validity
Generalisability to other people
Temporal validity
Generalisability to other eras
Face validity
Whether it measures what it is supposed to
Concurrent validity
Extent to which a psychological measure compares to an existing measure
Predictive validity
How well a test can predict future events/behaviours
Improving validity in experimental research
Using a control group - able to assess whether changes in the DV were due to effect of the IV
Improving validity in questionnaires
Lie scale - assess consistency of answers & control social desirability bias
Improving validity in observations
Ensuring not to overlap behavioural categories; use covert observations
Improving validity in qualitative methods
Direct quotes from ptpts & different sources for evidence, e.g., diaries/interviews/observations
Nominal data
Categories, e.g., male/female
Ordinal data
Ranks, e.g., low income/middle income/high income
Interval data
Precise, e.g., temperature in C/F
Interval data
Precise, e.g., temperature in C/F
Statistical test table
Test of diff
Ind grou Rep m/Mat p Test of a/c
N Chi-sq Sign test Chi-sq
O Mann W Wilcoxon Spear rho
I Unrel t Rel t test Pearsons r
Significance
How sure we are about a correlation/difference existing
If significant, we reject null hyp and accept alt
Probability
How likely it is for an event to happen
0 = stat impossibility
1 = stat certainty
Usual = 0.05
Type 1 error
Incorrect rejection of a null hypothesis which is actually true (false positive)
Type 2 error
Failure to reject null hyp that is false (false negative)
Null hypothesis
No relationship between two variables being studied
Paradigm
Set of shared ideas; assumptions within a scientific discipline
Paradigm shift
Significant change in central assumptions within a scientific discipline
Shows progress in science
Theory construction
Gathering evidence from direct observation
Should be able to make diff hyp from a theory
Deduction
Process of deriving new hypotheses from an already existing theory, e.g., episodic buffer in 2000
Falsifiability
Theory cannot be considered scientific unless it allows itself to be proven untrue
Hypothesis-deductive method
Process of formulating hypotheses that can either be proved/disproved by experimentation
Replicability
Extent to which scientific methods and their results can be repeated by other researchers across other contexts
Objectivity
All possible biases from the researcher are minimised so they don’t influence the research process
Empirical method
Evidence is collected through making direct observations; direct experiences
Psych as a science (+)
- Intuitive results produced against common sense
- Scientific methods used in many research studies giving scientific credibility
- Findings positively impact society, e.g., CBT
Psych as a science (-)
- Subjectivity
- Not all research is generalisable, e.g., case studies
- Psychologists often make inferences rather than directly measuring it