Research methods Flashcards
Define aim
A general statement of what the researcher intends to investigate
Define hypothesis
Statement of what researcher believes to be true, must be operationalised - defined + measurable
Define extraneous variable
Nuisance variables that don’t vary systematically w/ IV. Any variable, other than IV that may have an effect on DV if not controlled
Define confounding variable
Change systematically w/ IV so can’t be sure if change to DV is due to CV or IV
Define single blind
PP don’t know aims so demand characteristics are reduced
Define double blind
PP + researcher don’t know aims so demand characteristics + investigator effect are reduced
Define independent groups
PP randomly allocated to diff groups where each group represents 1 experimental condition
Evaluate independent groups
(+) No order effects bc tested once
(+) Less likely to guess aim therefore behaviour more natural
(-) Pps are diff, act as CV - reduces validity
(-) Pp variable, more time + money wasted - need twice as much pp
Repeated measures
Same pp take part in all conditions of experiment. Order is counter balanced to avoid order effect
Evaluate repeated measures
(+) Pp variable is controlled as same person is used
(+) Fewer participants needed bc take part in all conditions - more economical
(-) Order effects acts as CV - reduces validity
(-) Pp may guess aim, change behaviour - reducing validity
Matched pairs
2 Groups of pp used but related by being paired on pp variable
Evaluate matched pair
(+) Reduces pp variable, controls CV - inc validity
(+) No order effect bc tested once
(-) Matching is not perf, time consuming + can’t be matched exactly
(-) More time + money spent bc need more pp
Describe a lab experiment
- Controlled env
- EV + CV are controlled
- IV is manipulated + effect on DV is recorded
Evaluate lab experiment
(+) EV + CV controlled so effects on DV is minimised
(+) Easily replicated due to standardised procedure
(-) Lack generalisability, controlled lab env is artificial
(-) Pp know they’re being studied so gives rise to demand characteristics
Describe a field experiment
- Natural setting
- IV manipulated + effect on DV is recorded
Evaluate field experiment
(+) More generalisability bc env is more realistic than lab
(+) High ext val bc pp don’t know they’re being studied
(-) More difficult to control CV so harder to establish cause + effect due to effects on DV
(-) Ethical issues if pps don’t give informed consent
Describe a natural experiment
Change in IV isn’t brought about by researcher but would’ve happened even if researcher hadn’t been there
Evaluate natural experiment
(+) Provide opportunities for research that may be impractical or unethical
(+) High ext val bc study real life issues
(-) Naturally occurring events happen rarely, reducing opportunities for research - limits generalisability
(-) Pp not randomly allocated to conditions, less sure whether IV affects DV
Describe a quasi experiment
- IV based on pre-existing diff btw people
- Variable not manipulated, already exists
Evaluate quasi experiment
(+) Carried out in controlled conditions so high internal validity
(+) Replication possible due to high control
(-) Pp not randomly allocated to conditions, less sure whether IV affects DV
(-) Causal relationship not demonstrated bc researcher doesn’t manipulate IV
Opportunity sample
Ask people most available + w/in area
(+) Quick - most convenient
(-) Bias - unrepresentative of target population as it draws from specific area
Volunteer sample
Advertise, pp select themselves
(+) Requires minimal input by researcher - saves time
(-) Bias sample - pp share similar traits, keen + curious
Random sample
Given no. then picked out
(+) unbiased - researcher has no influence over who is selected
(-) Time consuming
Systematic sample
Every nth person is selected from target popullation
(+) Unbiased - researcher has no influence over who is selected
(-) Takes time + effort to get complete list of population
Stratified sample
Subgroups are identified, relative percentages of subgroup in population are refelected in sample
(+) Highly representative of target population, generalisability is more likely
(-) Selected pp may still refuse so more like volunteer sample
What are the alternative forms of consent?
- Presumptive - ask similar group
- Prior general - agree to be deceived
- Retrospective - get consent after study
What are the diff btw correlations + experiments?
- Experiments, researchers manipulate IV + record effects on DV, correlation, no manipulation of variables so cause + effect can’t be demonstrated
- Correlation, influence of EV isn’t controlled, maybe 3rd variable
Evaluate correlation
(+) Useful starting point for research - measures how 2 variables are related + suggests hypotheses in future research
(+) Economical - no need for controlled env, less time consuming
(-) No cause + effect - maybe 3rd variable
When are observations usually used?
W/in experiments as a way of assessing DV
Evaluate observations
(+) Captures unexpected behaviour - insight into unplanned behaviour
(-) Risk of researcher bias - situation affected by expectations
Naturalistic observation
Takes place where target population usually occurs
(+) High ext val
(-) Low control
Controlled observation
Some control/manipulation of variables
(+) Can be replicated - standardised procedures
(-) Low ext val
Covert observation
Pp unaware they’re beig studied
(+) Demand characteristics are reduced - inc validity
(-) Unethical
Overt observations
Pp aware they’re being studied
(+) Ethical - pp given consent
(-) Demand characteristics - reduces validity
Participant observation
Researcher becomes part of group
(+) Leads to greater insight - inc validty
(-) Loss of objectivity - researcher identifies too strongly w/ pp, threatens validity
Non-participant observation
Researcher remains seperate from group
(+) More objective - less bias, inc validity
(-) Loss of insight - reduces validity
Behavioural categories - behavioural design
Target behaviour to be observed should be broken up into set of observable categories
(-) Difficult to make clear + unambiguous - shouldn’t overlap
(-) Dustbin categories - all behaviours should be in list + not 1 dustbin
Time sampling - observational design
Observation made at regular intervals
(+) Reduces no. observations - more structured + systematic
(-) May be unrepresentative - may miss important details outside time scale
Event sampling - observational design
Target behaviour is recorded everytime it occurs
(+) Records infrequent behaviour
(-) Complex behaviour is oversimplified + unrecorded - dec validity
Questionnaires
Pre-set list of written questions to which pp respond
(+) Cost effective - can gather large amounts of data quickly bc distributed to large no. of people
(+) Pp more truthful compared to interview bc less self-conscious
(-) May not always be truthful + respond to present themselves in +ve light - social desirability bias
(-) Often produce response bias - respondents reply in a similar way eg. always ticking yes.
What are the qualities of a good questionnaire?
- Avoid jargons
- Avoid double barelled Q
- Avoid leading Q
Closed questions
Respondent has limited choices
(+) Easier to analyse - produces graphs, easy to draw conclusions from
(-) Respondents are restricted - reduces validity
Open questions
Respondents provide own ans expressed in words
(+) Respondents not restricted - inc val
(-) Difficult to analyse
Structured interview
List of pre-determined Q asked in fixed order
(+) Easy to replicate due to standardised format
(-) Interviewees can’t elaborate
Unstructured interview
No set Q
(+) More insight - able to elaborate
(-) Difficult to replicate bc lacks structure + not standardised
Outline what a pilot study is and the aims of a pilot study
- Small scale trial run of research design before doing the real thing
- Find out certain things don’t work to correct them before spending time + money on real thing
Quantitative data
Numerical data
(+) Easier to analyse - can draw graphs + calculate averages
(-) Oversimplifies behaviour - eg. using rating scales to express feelings, individual meaning are lost
Qualitative data
Non-numerical data
(+) Greater external validity than quantitative data - provides researcher w/ more meaningful insight
(-) Harder to analyse - large amount of detailed info is harder to summarise
Primary data
First hand data collected for purpose of investigation
(+) Fits job - study designed to extract only data needed, info directly relevant to aim
(-) Requires time + effort - 2º can be accessed w/in minutes
Secondary data
Collected by someone other than person conducting research
(+) Inexpensive - requires minimal effort
(-) Qual may be poor - research my be outdated/incomplete, challenges validity of conclusion
Meta-analysis
Type of 2º data that combines data from large no. studies. Calculation of effect size
(+) Inc val of conclusions - eventual sample size is much larger than individual sample, inc generalisability
(-) Publication bias - may not select relevant studies, leaving out -ve/non-sig results
What are the measures of central tendency?
- Mean
- Mode
- Median
Mean
Arithmetic average
(+) Sensitive - includes all scores w/in calculation
(-) May be unrepresentative - large/small no. distorts no.
Median
Middle value
(+) Unaffected by extreme scores - representative of data set as whole
(-) Less sensitive than mean - not all scores included
Mode
Most frequent no.
(+) Relevant to categorical data
(-) Overly simple measure - may be many modes in data
What are the measures of dispersion?
- Range
- SD
Range
Diff. btw highest + lowest +1
(+) Easy to calculate
(-) Doesn’t account for distribution of scores
Standard deviation
Measure of average spread around the mean
(+) More precise than range - includes all values w/in calculation
(-) May be misleading - extreme values aren’t revealed
What is peer review + what are the aims of it?
- Before publication, all aspects of investigation are scrutinised by experts in the field, should be objective + unknown to researcher
- Funding: allocate research funding
- Validation of qual + relevance of research
- Improvements + amendments suggested
Evaluate peer reviews
(+) Protects qual of published research - minimises fraud + means research is highest qual
(-) Publication Bias - research not seen as +ve or headline grabbing may be ignored, creates false impression of current state of psy
Outline a case study
- Detailed + in-depth analysis of individual/group
- Tend to be longitudinal
- Qualitative data
Evaluate case studies
(+) Rich detailed insight - inc validity
(+) Enable study of unusual behaviour - eg. HM, help understanding of normal functioning
(-) Research bias - conclusion based on subjective interpretation of researcher, reduce validity
(-) Pp accounts are biased - prone to inaccuracy, low val
Outline content analysis
- Observational research
- People studied indirectly via communications eg. media
- Coding (counting no. times word is mentioned) may produce quantitative data
- Thematic analysis produces qualitative data
Evaluate content analysis
(+) Fewer ethical issues - data is 2º that already exists - no issues obtaining permission
(+) Flexible - produces qualitative + quantitative data
(-) Communication studied out of context - researcher may attribute motivations to speaker, reduces validity
(-) Lacks objectivity
How do you assess reliability?
- Test-retest - test same person twice on 2/more diff. occasions, results should be same
- Inter-observer
- Correlation coefficient: +0.8
Outline ways to improve reliability
- Questionnaire - replace open Q
- Interview - same interviewer
- Experiments - standardised procedure
- Observations - operationalise behavioural categories
Give 2 types of validity
- Ecological validity - findings generalise to everyday setting
- Temporal validity - findings should be consistent over time
How do you assess validity?
- Face validity - whether test looks like it measures what it should
- Concurrent validity - whether findings are similar to those on well established test
Outline ways to improve validity
- Experiment - control group + standardised procedure
- Questionnaire - lie scale (test for social desirability bias) + told that data is confidential
- Observations - operationalised categories
- Qualitative research - use no. diff sources
Outline type 1 error
- Null rejected
- Optimistic error
- Sig diff found even when it doesn’t exist
- Too linient
Outline type 2 error
- Null accepted
- pessimistic error
- Too stringent
