topic 7: research methods Flashcards
demand characteristics
when participants know the aim of a research study and alter their behaviour accordingly (to give the results that the researcher wants)
what are experimental designs + list them
only had by experiments (not in observations), they are the different ways to use your participants
- independent measures design
- repeated measures design
- matched pairs design
what is an independent measures design + evaluate
where two completely different groups are assessed
time cost - have to re-explain, if you’re paying them then you might have to pay more people
sample size - needs the double the people of repeated measures
demand characteristics - no reason for demand characteristics
order effects - none
individual differences - most open to difference because there’s no matching between the groups
individual differences
any differences between the participants, they become extraneous variables
participants
your sample, the people that participated in your study
what is a repeated measures design + evaluate it
where the two groups used are identical because they are comprised of the same participants
time cost - if you’re paying them, then it could be cheaper bc there are less people
sample size - smallest
demand characteristics - highest chance because they will see variations of the IV (but counterbalancing is used)
order effects - possible as repeating the test many times may lead to improvement/fatigue
individual differences - least individual difference
what is a matched pairs design + evaluate it
where the two groups are matched on the most important variable (e.g. hair colour/height)
time cost - highest because everyone has to be tested individually and matched
sample size - needs the double the people of repeated measures
demand characteristics - potentially due to the prior tests for matching but unlikely
order effects - none
individual differences - in the middle of the other two
situational variables
anything about the test situation that affects the group
an extraneous variable
e.g. noise/temperature/lighting
participant variables
the individual characteristics that differ and ultimately decide how an individual responds to an environment
an extraneous variable
e.g. background differences, prior knowledge, health, anxiety, mood
how do you create a matched pairs design
- matched based on key variables (eg. gender + age)
- match with a relevant variable (eg GCSEs for memory)
- pretest to allocate (e.g IQ test)
- pair 1st and 2nd, 3rd and 4th etc.
- allocate one from each pair to each condition
what is the benefit of choosing a matched pairs design
it reduces individual differences which are extraneous variables
what is counterbalancing
used with repeated measures designs
1. split the participants into 2 groups
2. give each group IV1 and IV2 in opposite orders
this allows even distribution of order effects
what is randomisation
the process where it is decided whether participants should experience the experimental/control condition by chance when a repeated measures design is used
e.g. names in a hat/random number generator
what is a target population
the group/population that you want to study
your sample is then drawn from the target population
generalisability
the extent of the ability to apply the findings of your study to the general public
(SHS students liking to read ≠ all students liking to read)
random sampling
when all members of a population have equal chances of being selected
names in a hat:
- list out your population
- write all of the names on separate pieces of paper
- fold them and put in a hat
- randomly pick some for group 1
- randomly pick the same amount for group 2
random number generator:
- list out the population
- assign each participant a number
- set the generator to select 15 random numbers, these become group 1
- then eliminate those numbers/names
- set the generator to set a new 15 for group 2
opportunity sample
asking anyone who’s available and fulfils the criteria
(e.g walking into the common room and asking 4 girls and 4 boys)
self-selected/volunteer sample
advertise through newspapers etc. people just ask to be in the study
OR, people may not even know f it’s an experiment and choose whether or not to take part (e.g just leaving a ladder and seeing if people walk under or not)
stratified sample
using ratios and random selection from strata’s to form a mini population
e.g. if 80% of criminals are males and 200 total criminals, 160 should be males
consider things like gender/religion/ethnicity
it is v time consuming but very representative of the population
systematic sampling
- have a random list of your population
- number them
- decide how big your sample will be
- divide population by sample to get N
- use a random number generator to pick your starting person
- choose every nth person to get your sample
confounding variable
a variable that is confirmed to have interfered with the results
extraneous variable
any variable that interferes with a study
standardised procedures
anything that is kept the same
evaluate random samples
+ no researcher bias (they can’t determine who would be better suited)
- takes longer to plan (higher £)
- may accidentally select people with the same characteristics
- people may not want to take part/higher subject attrition
evaluate opportunity samples
+ don’t have to advertise; quick, cheap, easy to obtain
- may be biased, researcher may gravitate towards the more suited individuals
- the findings can’t be generalized to the target population (might have been mostly maths students in the common room at the time)
what is subject attrition
it is the loss of study units from a sample
evaluate self selected/volunteer samples
+ low subject attrition - people feel contracted and are usually committed
+ have to advertise but still minimal work
- biased; confident people often apply, people interested in psychological research often apply, people apply if they think they’d be good at it
evaluate stratified sampling
+ highly representative of population; lets you generalise results and avoid researcher bias 
- Very time consuming+difficult
- we can never fully represent the population (can’t include things like character)
- information can be difficult to find and finding it all and putting it into a random delete or can be v difficult/overwhelming
evaluate systematic samples
+ easy to construct, understand, compare and execute if already listed
+ cheap
+ avoids researcher bias
- assumes everyone is available
- can be difficult if there’s a v big population/no available list
what defines a piece of research as a true experiment
1) the researcher manipulates an IV and keeps all the other variables constant to measure the effect on the DV
2) participants can randomly be allocated to conditions
what are examples of true experiments
laboratory experiments
field experiments
what are examples of not-true experiments
natural experiments (not-trueal)
quasi experiments
what is a laboratory experiment
experiments carried out under controlled conditions; an artificial environment with tight control over variables
what is a field experiment
carried out in the p’s natural environment, but the IV is manipulated by researchers
what is a Quasi experiment
the IV being studied is a pre-existing difference between people (eg. age/gender)
a quasi experiment is often also a lab experiment
what is a natural experiment
when the IV occurs naturally; studying an island before vs after tv in terms of aggression level in kids
pros and cons of laboratory experiments
+ easier to replicate because of standardised procedures
+ allows control of IVs and extraneous variables; allows establishment of a cause and effect relationship
- low ecological validity; may produce unnatural behaviour, so findings can’t be generalised to a real life setting
- demand characteristics/experimenter effects may bias results, could become confounding variables
pros and cons of field experiments
+ higher ecological validity than lab experiment
+ if the study is covert, lower likelihood of demand characteristics
- less control over extraneous variables
- makes it difficult to replicate
pros and cons of natural experiments
+ ethical because it’s natural so you aren’t changing anything
+ very high ecological validity
+ lower likelihood of demand characteristics as p’s might not know they’re being studied
+ can be used when it could be ethically unacceptable to manipulate the IV (eg. stress)
- no control over extraneous variables that could bias the results; difficult to replicate
pros and cons of Quasi experiments
+ they can be used to study things that couldn’t be studied in a different way; eg. gender differences
— the two sample groups have characteristics that are unique to them that could become extraneous variables
since they’re often also lab experiments; can have high control but also higher chances of extraneous variables
objective
no bias is possible
subjective
bias is possible
validity
the extent to which a test measures what it claims to measure
ecological validity
the degree to which an investigation represents real life experiences
experimenter effect
the way in which the experimenter can accidentally influence the participant through their behaviour/appearance
control group
a group that doesn’t receive any manipulation of the independent variable, it can be used to compare with experimental groups
reliability
consistency (doesn’t mean that it’s correct)
what are ethics
a set of guidelines that should be followed by pyschologists that carry out research
what rights do participants in studies have
right to privacy
right to confidentiality
right to withdraw
informed consent
debriefing
what is the right to privacy
the right to not be studied anywhere that privacy would be expected; eg. toilets/at home
what is the right to confidentiality
participants have the right to expect that any information that they provide will be treated confidentially and not be identifiable as theirs if published
what is the right to withdraw
participants have the right to leave a study at any point and ask for their data to be fully erased
what is informed consent
when participants are made aware of all important aspects of research such as duration, purpose and potential risks / benefits and agree to being studied
what is consent
when a participant agrees to being studied but isn’t made fully aware of what the experiment entails
what is debriefing
a debrief is done after the study is over; participants should be given a general idea pf what the researcher was studying and why
if the participant was deceived they must be told so and given reasons as to why
any questions should be answered as fully as possible
why would you deceive a research participant
subject deception can increase scientific validity where having an honest reaction to the behaviour/attitude of the subject could produce biased results
what is the BPS code of Ethics in Psychological Research
a quasi-legal document produced by the BPS
it instructs UK psychologists on what behaviour is and isn’t appropriate when dealing with participants
it is built around; respect, competence, responsibility and integrity
what is a single blind technique
the participants isn’t told the aim of the study; it prevents demand charcteristics
what is a double blind technique
neither the researcher or participant know the study’s aim. it reduces both demand characteristics and investigator/experimenter bias
what are pilot studies
a small scale version of the final investigation
they are done to check that procedures, materials and measurement scales work and allows the researcher to modify the experiment if needed
it saves money in the LT as it is remedied on a small scale
what is standardisation
keeping the experiment identical for all participants
what is operationalisation?
it’s the process of making variables physically measurable/testable (variables such as stress, memory, intelligence)
it allows the construction of precise hypotheses
what is an aim
a general statement of why the study is being carried out
what is a null hypothesis
it states that the IV has no impact on the DV
what is an experimental hypothesis
a hypothesis that indicates that some difference will occur
can be directional/non-directional
non directional hypothesis
two-tailed hypothesis
acknowledges that there will be a difference between the conditions but not which direction it will lean in
what are the types of observations
controlled/naturalistic
covert/overt
participant/non-participant
direct/indirect
describe and evaluate a controlled observation
Watching and recording behaviour in a structured environment e.g. lab setting
+ high control of extraneous variables, making results less valid
– low ecological validity
describe and evaluate a naturalistic observation
watching and recording behaviors in a natural situation i.e. don’t create conditions
+ high ecological validity
– low control of extraneous variables, making results less valid
describe and evaluate a covert observation
participants are unaware of their behaviour being watched and recorded
+ high ecological validity because of natural behaviour
– ethical issues from a lack of informed consent
describe and evaluate an overt observation
participants may know that their behaviour is being watched and recorded, the research isn’t hidden
+ acceptable ethically because informed consent is given
– likely to have demand characteristics which decrease ecological validity
describe and evaluate a participant observation
the researcher observes whilst being a part of the group that is being observed
+ getting to know people could lead to subjective recording and observer bias
– can be more insightful and increase validity of findings
what are the types of observational designs
structured and unstructured
describe and evaluate a non-participant observation
the researcher is not a part of the group being observed while observing
+ the researcher can be more objective as they are less likely to identify with participants
– researcher may lose valuable insight
what is an indirect observation
when the behaviour is video recorded and observed and analysed at a later date
what is a direct observation
when the event is observed as it takes place
what is a structured observational design
the researcher quantifies what they are observing by using a predetermined list of behaviours and sampling methods
have to consider what behaviours you are interested in and then operationalise each one
categories have to be mutually exclusive
cant choose too many/few behaviours
too few - low reliability, validity and accuracy bc you miss out on other behaviours that do occur
too many - takes too long to locate each behaviour on a long list
what is an unstructured observational design
continuous recording where the researcher writes down everything that they see during the observation
often done at the start of research when unsure of what behaviours may occur
this lack of structure can cause you to miss small, influential behaviours whilst being side-tracked by bigger and more obvious ones
what are the sampling methods used for observations and evaluate them
- event - recording every time and event (behaviour) occurs, you have a behavioural list and have to tally every time is occurs, it gives frequency data
+ frequency data lets you know how often the behaviour occurs
– a lot could be happening at once, making it difficult to record all the behaviours and reducing reliability and validity
– don’t know when the behaviours occurred so cant identify trends/if the behaviours were triggered by something
- time - focuses on recording behaviours at certain points in time, eg. record what is happening every 5 minutes
+ more practical as less observations have to be recorded
+ can establish the sequence of behaviours
– if time slots are too large then behaviours will be missed, reducing validity
if time slots are too short then behaviours have to be recorded too regularly, making it impractical
describe and evaluate measures of central tendency
they are descriptive statistics that measure average and look at the middle of the data set
- mean
– an anomaly will interfere with the mean
+ it is representative as it makes use of all data values - median
– not representative of all data sets
+ not affected by anomalies - mode
– can have so many modes that it becomes pointless
+ useful if data is in categories
describe and evaluate measures of dispersion
range
– affected by extreme values
+ easy to calculate
standard deviation
tells us how much scores differ from the mean on average
used if the mean is used as the measure of central tendency and takes every value into account
a large standard deviation = scores are widely distributed, there are many scores occurring a long way from the mean and vice versa
it is used when your data has a standard bell value
+ not affected by extreme values
- can be hard to calculate
characteristics of bar charts
- used for non-continuous data, aka categorical/discrete
- bars can’t touch each other
characteristics of histograms
- show continuous data
- columns can touch each other
- all intervals are shown, even if there aren’t any scores within them
characteristics of scatter graphs
used because it shows a relationship between two variables
what is a self-report
a data-collection form where the p answers questions about themselves concerning their behaviour, thoughts or feelings
eg. questionnaire/interview
an advocate may answer instead if the p lacks capacity
outline characteristics of a questionnaire
they can be open, free rein in answering, or closed, limited no. of answers
formal eg. psychometric tests - IQ tests/personality tests
outline characteristics of interviews
can be unstructured/structured/semi-structured
unstructured: no predetermined structure but the interview does have a goal and sense of direction in mind. it usually involves open ended questions
- may lose focus
structured: has predetermined questions which they stick to; each p will be asked the same questions and in the same order
semi-structured: uses predetermined questions but with a flexible approach, the flow of following questions can be determined by the answers given
describe the peer review process
- author writes and submits article manuscript to journal
- journal editor send manuscript to expert reviewers to evaluate the quality of the research, write up and conclusions
- expert reviewers return manuscript to editor w/ change suggestions and recommendations on publishing or not
- editor reviews suggestions and returns manuscript to author for revision
- author revises manuscript and resubmits
- journal editor includes article in journal issue
what are the main aims of peer review
- validating the quality and relevance of research; assesses all elements (method, hypotheses, statistical tests etc. )
amendments are then suggested - to prevent scientific fraud (fake data)
- to allocate research funding to the most worthwhile causes
evaluate the peer review process
– publication bias; they want to publish snappy and attention grabbing headlines
– it can slow rate of change in a field; work is more likely to be accepted for publication is consistent with existing theories/the reviewers views
+ its anonymous (single blind, researcher is anon), which encourages honesty
+ promotes and maintains high research standards, spreading accurate knowledge throughout the field
how is test-retest reliability established
- repeat it with the same participants
- plot scores from first time on one axis of graph and second time on the other
- assess correlation strength with a Pearsons r test
- determine degree of reliability by comparing the score with the statistical table
why can using standard deviation be advantageous
SD isn’t easily distorted by a single extreme score
it takes account the distance of all scores form the mean
it doesn’t just show the distance between the highest and lowest scores
how can the validity of data be checked
can be done via face or concurrent validity
face validity - ask other people if your measurement method is a good measure
concurrent validity - give p’s a previously established version of your measurement method and compare with your method to see if the data sets correlate positively
how can reliability of observation data collection be established
use multiple observers and compare findings
what are the types of distribution
normal - symmetrical pattern of frequency data forming a bell shaped curve
skewed - asymmetrical spread of frequency data where the data clusters in one side
positive: lump at start
negative: lump at end
what are the economic consequences of research into psychopathology
workers can return to work
what are the economic consequences of research into attachment
mothers can return to work
working arrangements in families can be more flexible
what are the economic consequences of research into social influence
union strikes to make working conditions better
environmental campaigns to get companies to reduce waste and use of non-renewable energy
what are the economic consequences of research into memory
let to police using the cognitive interview which rescue a wrongful convictions and so less waste of money and space in jail
what is content analysis
a method of analysing qualitative data by turning it into quantitative data
- select a research question
- select a sample of pre-existing qualitative research
- code categories
- repeatedly listen or read
- tally to show which categories are most common
evaluate content analysis
+ allows statistical analysis of qualitative data
— purely descriptive and no explanatory power
— researcher bias when interpreting the data
outline thematic analysis
- researcher reads the data over and over again to familiarise themselves
- themes within the data emerge
- researcher reviews the themes and patterns to see if they can explain behaviour and answer the research question
- researcher categorises and defines each theme
- researcher writes up the analysis into their final report
evaluate thematic analysis
— researcher bias when interpreting the data
+ subjective so a range of theories can be applied
— researchers may find it difficult to know which themes to focus on
what are the components of a scientific report
- abstract: summary of key details of research report (ie. aim, hypothesis, method, results and conclusion) 150-200 words
- introduction: info on any related past research, gets more specific at the end when aim and hypotheses are presented
- method: design, sample method etc
- results: findings with inferential and descriptive stats, maybe a content/thematic analysis for qualitative
- discussion: considers what the findings mean for psychology and society and evaluation
- referencing: reference all sources used in the report
what if you’re told to ‘refer’ to something in a 16 marker?
deffo refer to it but yes you can mention other research!
compare primary and secondary data
primary gets the exact info needed and so fits aims and objectives
but it can be ££ and needs time and effort
secondary is cheaper and required minimal effort to collect
the data might be outdated and may e not reliable because the researcher wasn’t there when it was conducted so may be unsure about the results’ validity
why does multiple observers increase reliability
it lowers individual bias
what is done if correlation is significant
null is rejected!
what is qualitative data
descriptive data
at what correlation is reliability yayyy
generally 0.8
why is being unethical bad
it gives psychology a bad reputation and people are less likely to be in future research
what correlational test is used for ordinal data
spearman’s rho
what correlational test is used for interval/ratio data
pearson’s r
why would a scatter graph be used
to look at relationship between x and y
they display the relationships between co-variables
what is meta-analysis
the process where researchers collect and collage a wide range of previous research on a specific area to review it together and often stats analysis to get an overall conclusion
what does it mean for distribution if mean, mode and median are similar vs different
similar means that the data is normally distributed
if mean is greater than other two the data is positively skewed
if mean is lower than them the data is negatively skewed
how can ordinal data be identified
the difference between each score isn’t fixed and data can be ordered
how would you do a matched pairs design
- do the question
- give examples of what q you might ask
- pair pupils with similar recreational screen times
- place one pupil from each pair randomly into A/B
what’s a curvilinear relationship
where both variables increase with one another until a certain point where one variable keeps increasing but the other starts to decrease
this forms an inverted u on a graph
what is a disadvantage of meta analysis
publication bias: the researcher may intentionally not publish all of the data and choose to leave out negative results
this gives a false representation of what they were investigating
what are the features of science
paradigm: Kuhn said a paradigm shift shows scientific progress
bc psych has too much disagreement it is known as pre-science
theory construction and hypothesis testing: gathering evidence from investigations to test a hypothesis and deducing new hypotheses from an existing theory
falsifiability: Popper says this is key for scientific theories
replicability
objectivity and empirical method
evaluate psychology as a science
+ produces intuitive results against common sense
+ scientific methods are used in many research studies which gives them scientific creds
— a focus on experiments in artificial conditions has left us with little knowledge on how people behave in natural environments and so findings may only show behaviour in that environment
— subjective inferences of experiments
— lack of paradigms and paradign shifts
how do you assess internal reliability
split half method
randomly select half of the questions and put them in one form, then do the same for the others
the two forms of the same test are separately done and should yield the same sore with a correlation coefficient over +0.80
how can you improve reliability in a questionnaire
deselect or rewrite some items such as replacing open questions with room for misinterpretation, with closed questions
how can you improve reliability of interviews
use same interviewer or at least train them all to use the same questions that aren’t leading etc. baso make super similar
how do you improve the reliability of experiments
make sure that the method is precisely replicated and that all conditions are the same as expected
how can you improve the reliability of observations
make sure that behavioural categories have been fully operationalised and that they are measurable and self-evident
make sure categories don’t overlap
bc then people make their own judgements and these can differ
how can validity of experiments be improved
use a control group
single or double blind procedures
how can validity of questionnaires be improved
assure respondents that data will all stay anonymous
incorporate a lie scale within questions to assess consistency of responses and control for effects of social desirability bias
how can validity of qualitative methods of study be improved
triangulation
what are some factors to be considered when designing a questionnaire
clarity
emotive language, double barrelled questions, double negatives and leading questions bc they can cause biases affecting validity of results
think about the sequence of questions; start with easy and go on to harder ones
filler questions
pilot study to check that the questionnaire is suited
why is it good when people are very scientific in psychology
it increases its scientific credibility
outline how a researcher would obtain a stratified sample
- identify the strata within the given population
- calculated the required proportion from the stratum based on the proportion in the population
- select sample at random from each stratum
- use a random selection method eg. names in a hat
how is interval/ratio data converted into ordinal
order the scores from lowest to highest and give each a rank place, equal scores share a place
how can interval/ratio data be converted into nominal
categorise scores into groups for each condition
eg. scores over 40, scores between 40 and 60 etc.