Research methods Flashcards
experimental method
changing independent variable to measure effect on dependent variable
aim
a statement outlining the purpose of an investigation
hypothesis
a clear, testable statement stating relationship between variables being used
directional hypothesis
states the direction of the difference or relationship
levels of the IV (independent variable)
to test the effect of the IV, there needs to be different conditions. different conditions are known as levels
operationalisation of variables
the process of defining variables in a way which makes them measurable i.e. referring to specific observable behaviours.
extraneous variables
any variable that isn’t the IV that will affect results if not controlled. two types: participant and situational variables
confounding variable
an EV that affects independent and dependent variables, it’s difficult to tell if results are due to CV or a IV
demand characteristics
cue from a researcher which might affect participants behaviour. please-u effect = acting in a way to please researcher. screw-u effect = acting in a way they think will sabotage the study
investigator effect
anything investigator does which has an effect on participants performance in a study other than intended
randomisation
the use of chance in order to control for the effects of bias when designing materials and deciding the order of conditions
standardisation
using exactly the same formulised procedures and instructions for all participants in a research study
laboratory experiment
conducted in a highly controlled environment
evaluation of lab experiments
+ high control over EVs and CVs so you know that any change in DV is due to IV, giving it high internal validity
+ replication is more possible because of amount of control, making it more valid
- artificial so lack generalisability and have low external validity
- pts are aware they’re being tested so may act unusually
- don’t represent the everyday experience, low mundane realism
field experiment
an experiment conducted in the participants’ natural environment. natural, more everyday setting
evaluation of field experiments
+ higher mundane realism
+ pts may be unaware they’re being studied, more valid + authentic behaviour, high external validity
- no control of EVs and CVs so difficult to establish link between IV and DV
- precise replication is often impossible
- ethical issues of having no consent from pts - invasion of privacy?
natural experiment
no control over the IV and can’t change it.
the IV is always natural but not always the setting - could be in a lab.
DV may also be naturally occurring.
evaluation of natural experiments
+ allows us to study real world issues as they occur, high external validity
- naturally occurring events are rare, reducing opportunities for research, also limiting scope for generalising
- pts may not be randomly allocated, so less sure if IV affected DV
- if conducted in lab = lacks realism
- demand characteristics may be an issue
quasi-experiment
the IV is based on existing differences (e.g. age or gender) so can’t be controlled
random sampling
everyone has an equal chance of being chosen. everyone should have random numbers assigned to them and chosen using a random selection method
systematic sampling
a systematic formula is used – every 50th person is picked from the phone book/school register
stratified sampling
different strata or subgroups are identified in the target population. what percentage of the whole target population does each sub group form? a random sample is taken from each subgroup so that the sample has the
same percentages as the target population
opportunity sampling
the researcher selects people from the target population who happens to be there at the time, and who are willing and able to take part.
volunteer sampling
asking for people to volunteer if they are able to take part in the study
what is the purpose of a sample?
to be representative of a target population, meaning that the results can be generalised to the whole population
what are the 6 types of observation?
naturalistic, controlled, covert, overt, participant, and non-participant
naturalistic vs controlled observation
naturalistic is observed in environment where behaviour would normally occur. controlled is in a structured environment where some variables are controlled
covert vs overt observation
covert is when participants behaviour is watched without consent. overt is with consent
participant vs non-participant observation
participant is when the researcher becomes a member of the group they are observing. non-participant is when the research remains separate from the group whose behaviour they are watching.
evaluation of natural vs controlled observation
natural observations have high ecological validity + external validity, they can be generalised but they are difficult to replicate. controlled observations can be repeated to check reliability but they’re not as easy to generalise
evaluation of covert vs overt observations
covert observations don’t have issue of demand characteristics, meaning data collected has higher internal validity, but are less ethical as participants can’t give informed consent. overt are more ethically acceptable but demand characteristics can make behaviour unrealistic
evaluation of participant vs non-participant observations
participant studies mean that the researcher can get more in-depth data as they are in close proximity but they do risk ‘going native’ and losing objectivity. non-participant studies allow researcher to maintain objectivity but lose the chance of more in depth data
what is the British Psychological Society (BPS) and what do they do?
representative body that promotes excellence and ethical practice in psychological studies
when do ethical issues occur?
when there is a conflict between the rights of participants and the goals of the study
what are the 4 main ethical concerns?
informed consent, confidentiality, deception and protection from harm
informed consent
made aware of the aims and procedure of the study and their right to withdraw. in form of a consent letter, if participant is under 14, parent/guardian has to give consent
evaluation points of informed consent
- explaining exact aims can make studies meaningless and artificial.
- naturalistic observations don’t need consent as they are in a public area where you can expect to be watched.
- Menges (1973) found 97% of American studies hadn’t given participants all information
deception
deliberately misleading or withholding information. participants cant give informed consent if deceived. BPS allow some deception if scientifically justified. they can also obtain general consent with knowledge that deception will take place
protection from harm
risk of harm to participants can’t be higher than risk of their everyday lives. counselling should be offered if subjected to embarrassment or stress. a debrief should also be given after studies finish.
protection from harm evaluation
it’s difficult to accurately assess, and predict what participants will find distressing or embarrassing.
confidentiality
right to privacy, meaning personal information can’t be shared. all data collected should not use names. pts can’t be identifiable from data published (e.g. from characteristics, place, time). pts should be warned if data isn’t going to be completely anonymous
pilot study
small scale trial run of study to check procedure and make final tweaks - prevents money being wasted on research that has too many issues
single-blind procedure
participants are unaware of the aim + conditions. this controls demand characteristics
double-blind procedure
researcher is also unaware of the aim + conditions. this reduces potential bias
control condition
sets a baseline with no manipulation of the independent variable. it can be used to compare to the experimental condition to know the effects are due to the IV rather than other variables
what are the two types of observational design?
structured or unstructured.
what is an unstructured observational design?
writing down everything observed, rich in detail, good for small groups
what is a structured observational design?
selecting observational categories that are being looked for (operationalisation), sampling, then recording behavior either each time it is observed or which behavior occurs in a pre-decided time frame.
inter-observer reliability
single observers may miss certain behaviors or only record ones that fit their hypothesis. to fix this, observers should familiarize themselves with categories and also discuss any differences between amount of behaviors observed. can be calculated to give exact figure
structured vs unstructured evaluation
+ structured makes recording of observations easier and more systematic
+ structured produces numerical data = easier to analyse and do statistical tests
+ structured means less risk of observer bias
- unstructured provide more rich data with more detail
behavioral categories evaluation
+ makes data collection more structured and objective
- must not; require further interpretation, be ambiguous, be too broad, overlap with each other
a pilot study should be conducted to check validity of these categories.
what are the three different types of experimental designs?
independent groups, matched pairs and repeated measures
what is independent groups?
separate different groups of people participate in each condition of the experiment
evaluation of independent groups
+ no order effects and less chance of demand characteristics
+ relatively easy to organise
- participant variables are more likely (different skills/traits can affect results)
what is repeated measures?
all participants complete all conditions of experiment
evaluation of repeated measures
+ no participant variables
+ fewer participants needed to carry out experiment
- order effects and demand characteristics
what is matched pairs?
participants only take part in one condition of the experiment
but they are matched to the participants in the other
condition(s) for all key participant variables.
evaluation of matched pairs
+ reduces participant variables
+ no order effects and less chance of demand characteristics
- matching takes time and can be difficult
- some participant variables could still be overlooked
how can independent groups be improved?
randomly assigning which groups complete each condition, reduces likelihood of participant differences through researcher bias
how can repeated measures be improved?
counter-balancing - splitting group in half to get half participants to do tasks in opposite order to other half of participants. reduces order effects and demand characteristics
self report techniques
interviews and questionnaires
acquiescence bias
tendency to agree no matter content of the question (following a pattern in answers on questionnaire)
social desirability bias
presenting in positive light
evaluation of questionnaires
+ qualitative and quantitative data
+ quick and cheap
+ large sample
- social desirability bias
- sample bias to people who are willing and able to complete
- possibility of misunderstanding a question
evaluation of interviews
+ rich and detailed information
+ chance to ask for clarification if misunderstood question
+ qualitative and quantitative data
- time consuming
- demand characteristics
- not appropriate for all topics or all people
types of closed questions
fixed choice questions
likert scales (agreement on numerical scale)
rating scales (value that represents strength of feeling - e.g. enjoyment)
designing an interview
you need a:
standardised procedure to reduce researcher bias
notes being taken or video being recorded
one to one or group
quiet place
list of questions/topics to cover
neutral questions at first to establish rapport and allow to relax
reminder of right to withdraw and confidentiality
what to avoid when writing questions
- leading questions
- double-barelled questions
- double-negatives
- emotive language
- overuse of jargon
correlation
relationship between two variables where changes in one variable go along with changes in the other
positive correlation
as one goes up so does the other
negative correlation
as one goes up, the other goes down
coefficient of correlation
-1 is perfect negative correlation
1 is perfect positive correlation
evaluation of correlation
+ don’t require manipulation of variables
+ can make predictions that can be tested in experiment
+ high ecological validity, results come from real life
- can’t show cause and effect
- don’t reflect curvilinear relationships
- extraneous variables could be causing changes
quantitative data
represents how much, how long, how many, etc. there are of something, measured in numbers or quantities (objective)
qualitative data
can’t be readily counted, expressed in words, may include description of thoughts, feelings, and opinions (subjective)
evaluation of quantitative data
+ allows for statistical analysis and comparisons to be made
+ objective and scientific data so easier to establish reliability of results
- numbers produced without context as to why behaviour happened
- reducing people to numbers (not taking experience into account)
evaluation of qualitative data
+ in-depth and rich so increases external validity of findings
+ helps us understand why people behave in a particular way
- interpreting data is up to researcher bias and subjectivity
- difficult to interpret and make statistical comparisons
primary data
original data collected specifically for the purpose of the investigation
secondary data
data collected by someone other than the person who is conducting the research
meta-analysis
large scale combination and analysis of secondary data from many studies with the same research questions
evaluation of primary data
+ designed so it fits the aim and hypothesis of the study
- lengthy and therefore expensive process to collect the data
evaluation of secondary data
+ easier and cheaper to access someone else’s data
- may not fit needs of the study or may be of poor quality (lacking temporal validity or being incomplete)
evaluation of meta-analysis
+increases validity due to large sample size
- prone to publication bias, researchers don’t select studies with negative or non-significant results (file drawer problem)
things to do when drawing graphs
- clearly label axis
- draw with pencil and ruler
- linear scales
- take up as much space as possible
- title explaining graph
what type of data do line graphs show and what is their purpose?
continuous data that isn’t grouped, and they track the changes over periods of time and compare data
what type of data do bar charts show and what is their purpose?
discrete data and allows to compare the categories
what type of data do scatter graphs show and what is their purpose?
discrete and correlational data, allows observation of relationship between two variables
what type of data do histograms show and what is their purpose?
continuous grouped data that has to put in data and allows us to see distribution of data
one-tailed test
directional hypothesis
two-tailed test
non-directional hypothesis
null hypothesis
no differences between conditions
the three questions to ask to know what statistical test to use?
- test of difference or correlation?
- experimental design (related or unrelated)
- type of data (nominal, ordinal, interval)
validity
measure of truth and accuracy, also about generalisation
internal validity
clear cause and effect
external validity
able to be generalised
ecological validity
how generalisable the settings of the study are to real life settings
mundane realism
how realistic the stimuli is to real life
reliability
how consistently a method measures something, the ability to repeat and obtain the same results
external reliability
consistency when repeated
internal reliability
items are consistent within themselves (questions in a questionnaire)
ways to increase validity
test-retest, replicate, split half, inter observer
test-retest
interviews + questionnaires, same person different occasions, correlated and greater than 0.8 -> reliable
split half
compare half qs with other half of qs to check similar difficulty
replicate
should obtain same results when repeated if standardised procedures are used
inter observer
compare observation between observers, check interpretation of behaviour is same, helps overcome researcher bias
face validity
looks promising that tool measures what it’s supposed to
population validity
how representative sample is of the population
temporal validity
if research findings apply across time
concurrent validity
if measure is in agreement with pre-existing measures to measure the same thing
triangulation
gathering evidence from different sources
interpretive validity
researcher has to use direct quotes to show their interpretation matches participants reality.
why are psychological reports written?
for clear communication between researchers, in a conventional manner to aid understanding
order of sections
abstract, introduction, method, results, discussion, references
economical implications of psychological research
- gender pay gap increased/reduced
- burden on NHS
- more/less taxation
- less sick pay
- increased productivity
- cutting edge research = funding and investment from overseas
why is peer review important?
Prevent plagiarism, methodology, validity, integrity, significance
evaluation of peer review
+ helps prevent scientific fraud
+ £5.8 bil spent by gov. on research due to high research ratings
+ promotes accurate information
- conflict of interest, researchers may negatively review others work
- file drawer effect, only statistically significant results are published
- slow process
example of importance of peer review
Andrew Wakefield claimed MMR vaccine caused autism in children, they were false claims with no evidence yet people believed him and therefore many children died
statistical tests mnemonic
Simon Cowell Curiously Wants More Singers Receiving Unanimous Praise
questions to ask when deciding stat test
- correlation or difference
- type of test (related or unrelated)
- type of data (nominal, ordinal or interval)
