Research Methods Flashcards
(33 cards)
The Experimental Method
- Aims are stated
- There are two or more levels of the independent variable (IV), manipulated by the experimenter
- The effect is measured on a dependent variable (DV), which is operationalised
- Extraneous variables are controlled and procedures are standardised
- The hypothesis states the relationship between the IV and DV
- Casual conclusions can be drawn
Control of Variables
- Confounding variables vary systematically with the IV, thus any changes in the DV may be due to a confounding variable instead of the IV
- Extraneous variables are nuisance variables and make it harder to detect change in the DV
- Mundane realism is the extent to which features of a study mirror the real world
- Generalisation-findings from a study may lack generalisability if the materials or environment lack mundane realism or if participants know they are being studied
- Validity= legitimacy, genuineness
- Internal Validity- enhanced by control of confounding variables, high mundane realism
- External Validity- generalising to other situations (ecological validity), people (population validity) and historical periods (historical validity)
Hypothesis
- Directional hypothesis states more, less, higher, lower etc
- Non-directional hypothesis does not state the direction of the difference
- State direction if indicated by past research
Pilot Study
- A Pilot Study is a trial run with similar participants to test procedures and amend them if necessary
Confederates
- Confederates are directed by a researcher to play certain roles in a study
Sampling
- A small group of people selected from a population
- Opportunity sample= recruit those easily available
+ easy because participants are there- biased
- Random sample= using a random technique e.g. lottery method or a random number generator
+ unbiased- takes time
- Stratified sample= identify relevant subgroups, randomly select appropriate proportion from each subgroups
+ proportional, representative and unbiased- time consuming
- Systematic sample= selecting every nth person
+ unbiased- not truly random
- Volunteer sample= people respond to an advertisement
+ variety of participants- volunteer bias
Ethical Issues- Researcher:
- Informed consent= may give away the aims of the study
- Deception= acceptable when information withheld, less acceptable then dishonest
- The right to withdraw= biases the sample
- Protection from psychological and physical harm= difficult to guarantee because it is unpredictable
- Confidentiality- publication of findings may reveal the identify even if anonymous
- Privacy- hard to protect when participants studied without their awareness
Ethical Issues- Participants:
- Informed consent= basic human right based on knowing what is involved
- Deception= prevents informed consent, may distrust psychologists in the future
- The right to withdraw= compensates for situations involving deception
- Protection from psychological and physical harm= risks should be no greater than everyday life
- Confidentiality= a legal right (the data protection act)
- Privacy= even in a public place people may not wish to be observed
Dealing with ethical issues:
- Ethical guidelines= tell psychologists what is acceptable
+ rules and sanctions approach offers clarity- can not cover everything
- closes off discussion
- absolves researcher responsibility
- Cost- benefit analysis= judged from the participant’s perspective or perspective of society
- can not predict costs/benefits beforehand
- just creates new dilemmas (Baumrind)
- Ethics committees= approve studies based on cost-benefit considerations
- Punishment= may be barred from work as a psychologist
Dealing with specific ethical issues:
- Informed consent= sign a form, presumptive consent, the right to withdraw
- -> participants may not understand what is actually involved
- -> presumptive consent is a hypothetical agreement is different from real agreement
- Deception= debriefing is required and the right to withhold data
- -> debriefing can not turn the clock back, the harm is already done
- The right to withdraw= part of informed consent
- -> participants may feel that they can not
- Protection from harm= researcher can stop the study
- -> may only be apparent with hindsight
- Confidentiality= maintain anonymity
- -> people may work out identity
- Privacy= only acceptable in a public place
- -> lack of universal agreement about what counts as public
Experimental design:
- Repeated design- each participant is tested twice/ experiences both levels of the IV
- order effects (e.g. practice or boredom)
- may guess the aims of the study
+ can counterbalance
- Independent groups- each participant is only tested on one level of the IV
- participant variables not controlled
- need more participants
+ can use random allocation
- Matched pairs- each participant is paired with another participant, each pair receives both levels of the IV
- matching takes time
- may not account for all variables that matter
+ good compromise
- Counterbalancing- each condition is tested first or second in equal amounts, could be AB or BA or ABBA
Laboratory Experiments:
Laboratory Experiments- study with an IV and DV, conducted in the controlled environment. The IV may be contrived and thus reduce the mundane realism.
+ high internal validity
- usually aware of being studied
- operationalised of IV or DV may have low mundane realism
- participants may feel uncomfortable in the setting
Field Experiment:
Field Experiment- study with an IV and DV conducted in a more natural environment. The IV may be contrived and thus reduce mundane realism
+ usually not aware of being studied
+ more natural setting, so more relaxed
- Operationalisation of IV or DV may have low mundane realism
- difficult to control extraneous variables
- difficult to debrief
Natural Experiment:
Natural Experiment- the IV is natural insofar as it varies whether or not the researcher is there. The DV may be measured in a lab.
+ used where the IV can not be manipulated for ethical or practical reasons
+ can study ‘real’ problems
- can not demonstrate casual relationships
- random allocation is not possible, so low internal validity
- can only be used with existing conditions
- participants may be aware of being studied
- Operationalisation of DV may have low mundane realism
Quasi Experiment:
Quasi Experiment- the IV is not a variable, it is a condition that exists such as age or having an external locus of control
Problems with Experiments:
Demand Characteristics- cues in an experimental situation that convey hypothesis to participants, creating expectations of how to behave
Investigator effects- unconscious cues from investigator that affect participant’s performance other than what was intended, including indirect effects e.g. investigator experimental design effects
Single blind or double blind trials- minimise these problems
Observational Techniques:
Naturalistic observations- everything is left as it is normally, in an everyday setting
+ realistic picture, high in ecological validity
- little control, which makes conclusions difficult
Controlled observations- researcher regulates aspects of the environment
+ useful for focusing on particular behaviours
- behaviour may be less natural
Overt observations- participants aware that they are being observed, whereas not aware during a covert observation (may be informed afterwards)
- affects naturalness of behaviour, demand characteristics
Participant observations- observer is part of group being observed, whereas in non-participant observation observer watches from a psychological and probably physical distance
+ more objective
- likely to be covert, so ethical issues
Observational Design:
Unstructured observations- observer records everything
Structured observations- a system is used
Behavioural categories- target behaviour divided into individual behaviours that can be clearly identified e.g. smile showing teeth rather than being happy
Sampling procedures- record events (event sampling) or record behavioural categories at time intervals (time sampling)
Self-report Techniques:
Questionnaire- data collected through written, fixed questions
+ collect data from large number of people
- people may be more willing to reveal confidential information than in an interview
- can only be filled in by people who can write and have the time, biased sample
Structured interview- questionnaire delivered face-to-face or over phone; in real-time
+ can be repeated exactly (as can a questionnaire) good for making comparisons between people
+ good for analysis
- may lack comparability between interviewers
- interviewer bias
Unstructured questionnaire- may start with some predetermined questions and further questions developed in response to answers given
+ more detailed information
- more issues with interviewer skills and bias, more expensive because of training needed
- in-depth questions lack objectivity
- interviewer bias
Self-report design:
Clarity of questions- avoid ambiguity, double negatives, double-barreled questions
Bias- avoid leading questions and social desirability bias
Analysis- use closed or open questions
Open Questions
+respondents can expand answers, more detail provided
+ unexpected answers, new insights
- respondents may not give full answers
- may be difficult to analyse
Closed Questions
+ mainly quantitave data produced, easier to analyse and draw conclusions
+ procedures can be easily repeated, to confirm findings
- no causal relationship demonstrated
- people may act on erroneous causal conclusions, which may be dangerous
- apparent correlations may be due to intervening variables
- measurement of variables may lack validity or poor sampling may mean low generalisability
Other considerations- filler questions, leave anxiety-provoking questions until later, consider sampling technique, run a pilot study
Recording interviews- writing down answers may make interviewee feel evaluated
Effect of interviewer- non-verbal cues (e.g. crossing arms), listening skills (e.g. regulating comments)
Questioning skills in unstructured interview- how and what questions to ask (e.g. not too probing or repetitive)
Correlations:
Correlations- an analysis of association between continuous co-variables
Positive correlation- dots on scattergram from bottom left to top right, co-variables increase together
Negative correlation- top left to bottom right, as one co-variable increases the other decreases
Zero correlation- no significant association
Correlation Hypothesis can be directional (positive or negative correlation) or null (zero correlation)
A correlation coefficient is a number between +1 and -1 to show the strength of the correlation
Evaluations:
+ can investigate trends, which may justify further investigations or may rule out any causal link
+ procedures can be easily repeated, to confirm findings
- no causal relationships demonstrated
- people may act on erroneous causal conclusions, which may be dangerous
- apparent correlations may be due to the intervening variables
- measurement of variables may lack validity or poor sampling may mean low generalisability
Meta-analysis:
Meta-analysis= data analyzed from a review of studies sharing the same aim/hypothesis, effect size (dependent variable) may be produced
+ results from number of studies increases validity
+ balances out contradictory results and gives an overall figure
- studies may not be comparable
Case Study:
Case Study= a detailed study of a single individual, institution or event using many different research methods
+ in-depth data overlooked using other methods
+ good for studying rare events/ behaviours
+ complex interactions can be studied
- unique characteristics
- may involve recollection of past events
- there is generally no ‘before’ comparison
Content Analysis:
Content Analysis- form of indirect observation of the artefacts people produce
+ based on real events, high ecological validity
+ easily replicated if data is public
- may suffer from observer bias
