Research Methods 1&2 Flashcards
Aim
a statement of what the researchers intend to find out in a study
Hypothesis
a precise and testable statement of prediction about outcome
Null
no difference between groups
Experimental
predict difference between groups
Directional
suggest direction of results used if previous shows one thing
Non-directional
suggests no direction of results normally has no previous research or has conflicting research
IV
variable that researcher manipulates
DV
variable that researcher measures
Extraneous
variables that could effect the IV that is not the DV
Confounding
when extraneous variables not controlled can damage validity
Situational
features of a situation that could effect results
Order effects
order taken for tests can effect results
Counterbalancing
ABBA
Participant variables
individual differences
Investigator Effects
experimenter unconsciously conveys to ppts how they should behave, called experimenter bias
experimenter is often unaware of the influence which they are exerting, but they have an influence nevertheless
Double Blind Design
ppt and conductor blind to aims of hypothesis
Inter-rater/observer Reliability
independent raters, rate and compare
single observers may miss important details or may only notice events that confirm their hypothesis
2 observers should carry out a experiment so their is no bias and compare results and discuss differences, correlating each pair of results and an overall figure is made
observe behaviour at same time
Demand Characteristics
if ppts guess the experiments intention change behaviour to suit aims
they may think they are doing this to be helpful - but its not actually helpful
Independent groups
different ppts take part in different conditions
i.e. group A listen to music while revising group B do not, results are compared
usually, ppts would be randomly allocated to each condition
Repeated Measures
every ppt takes part in every condition (results are compared to themselves)
Matched Pairs
different but similar ppts are used in each condition
they are matched on variables relevant to the study
monozygotic twins can be good for this as they are genetically perfectly matched
Population
refers to a large group of individuals who the researcher may be interested in studying
Target Population
group of people who take part in the research
Sample
desired sub-group of the population to be studied
Random Sample
all ppts have an equal chance of being selected
What should an aim be
clear and detailed
outline the purpose of the study
What shouldnt an aim be
a question
Operationalise
to make measurable
What a good hypothesis should include
all conditions groups/ comparisons
prediction …..will be
should be directional or non-directional
operationalised
Internal Validity
are we measuring what we intend to measure ,only thing effecting DV is IV
Examples of Situational Variables
heat
time of day
order effects e.g. two tests in a row might be fatigued or might get better, solution is splitting group and having one do the test in AB order and the others in BA order
Examples of Participant Variables
mood
IQ
anxiety
concentration
Controlling Investigator Effects
double blind deign, both the ppt and conductor are blind to the aims of hypothesis
experimenter may be unaware which variable the ppt is involved in
inter-rater reliability, independent raters rate same behaviours and then check agreement
Screw-you Effect
deliberately try to hinder the experiment
Controlling Demand Characteristics
single blind design - the ppt is unaware of which condition they are in or the researcher aims, this prevents the ppts from seeking clues about the aims and reacting to them
deception - lying about the aims of the study and/ or using distracting questions
Extraneous or Confounding
when extraneous variables are not controlled by the experimenter, they become confounding variables that can damage validity of the experiment
Independent Groups Strengths
avoids order effects (such as practice or fatigue) as people participate in one condition only
avoids demand characteristics as people are less likely to guess aims
Independent Groups Limitations
more people are needed than with the repeated measures design (more time consuming + expensive)
differences between participants in the groups may effect the results, for example variation in age, gender, social behaviour
Repeated Measures Strengths
as the same ppts are used in each condition ppt variables are reduced
fewer people needed
Repeated Measures Limitations
there may be order effects, this limitation can be controlled using counterbalancing
demand characteristics, ppts guess aims and change behaviour
Matched Pairs Strengths
reduces ppts variables because the researcher has tried to pair us the ppts so that each condition has people with similar abilities and characteristics
avoids order effects
Matched Pairs Limitations
if one ppt drops out you lose two ppts data
very time consuming to find closely matched pairs
impossible to match people exactly even identical twins
Four Types of Experiments
lab, field, natural, quasi
Laboratory Experiment
conducted in highly controlled environments (not always a literal lab)
could be in a classroom in which conditions have been controlled in
Laboratory Experiment Strengths
uses a standardised procedure and is easy to replicate
high control of extraneous variables, allows cause and effect to be established
highly controlled makes accurate measurements possible
Laboratory Experiments Limitations
settings may not reflect real life and therefore no ecological validity
demand characteristics could be an issue
ppts know they are being studied which could pose a problem because of the Hawthorne effect
could be time consuming and expensive compared to other methods
Natural Experiments
when the researcher takes advantage of a naturally occurring independent variable
iv would have occurred despite if the experimenter were interested or not
setting for this type does not have to be natural
Natural Experiments Strengths
this type of study might be especially useful in studying phenomena that would be unethical to manipulate
Natural Experiments Limitations
can only be used when IV occurs naturally
ppts cannot be randomly allocated to these groups, there is strong likelihood of confounding variables
researcher has no control over IV as it will have occurred in everyday life
Quasi Experiments
has an IV based on an existing difference between people, it has not been manipulated it just exists
experimenter does not control IV
can occur in lab conditions
Quasi Experiment Strength
often conducted in controlled conditions and so shares the strength of a lab study
Quasi Experiment Limitation
can only be used when IV varies naturally
ppts cannot be randomly allocated to these groups, there is strong likelihood of confounding variables
Field Experiments
conducted in a real, every day setting in the real world
experimenter still manipulates IV in this type of experiment
Field Experiments Strengths
high ecological validity because it is done in the ppts environment
if study is covert much less chance of DC happening
Field Experiments Limitations
more difficult to control extraneous variables in this type of experiment making it harder for other researchers to replicate
Generalisation and Bias
ideally the sample that is drawn will be representative of the target population so that generalisation of findings becomes possible
difficult to achieve due to diverse nature of individuals
Random Sample Methods
need a list of whole target population
1 - put these names/ numbers in a hat
2 - random name generator choose relevant number or name
Random Sample Strengths
is free from researcher bias (no influence over whose selected)
Random Sample Limitations
time consuming and can be difficult to conduct
no guarantee they are representative
Opportunity Sample
those people who are available at the time and are willing to take part
Opportunity Sample Strengths
it is a quick and easy/ convenient way of choosing ppts
Opportunity Sample Limitations
researcher bias - could approach people who look helpful
unrepresentative of target population
Volunteer Sample
ppts put themselves forward to be part of the sample
Volunteer Sample Strengths
easy and requires minimal input, so less effort and time consuming for reseacher
Volunteer Sample Limitations
volunteer bias - type of person
Systematic Sample
every nth person of the target population is selected to take part
Systematic Sample Strengths
avoids researcher bias, once system has been established the researcher has no influence over who is chosen
Systematic Sample Limitations
difficult to achieve, time consuming
no guarantee it will be truly representative
Stratified Sampling
the researcher identifies the different types of people that make up the target population and work out the proportion needed for the sample to be representative
Stratified Sample Strength
avoids researcher bias
designed to be representative of the population
Stratified Sample Limitation
stratification is not perfect
difficult to have complete list of general population
Sampling - representative, time consuming, biased
most representative to least representative
stratified, random, systematic, opportunity, volunteer
most time consuming to least time consuming
stratified, systematic, random, volunteer, opportunity
most biased to least biased
opportunity, volunteer, stratified/ random/ systematic
BPS
the British Psychological Society is the representative body for psychology and psychologists in the UK, and is responsible for the proportion of excellence and ethical practice in science, education, practical applications of the discipline
the principles are: respect, competence, responsibility, integrity
if researchers break code of conduct they face losing their job
Ethical Issues
informed consent
right to withdraw
privacy
confidentiality
protection from harm
deception
Informed Consent
involves making ppts aware of the aims of the research, the procedures, their rights and also what their data will be used for, only then can the ppts make an informed judgement of whether they wish to participate
Right to Withdraw
this is available for ppts after they find out everything about the investigation and are completely informed
Privacy
ppts have the right to control information about themselves, this is the right to privacy
Confidentiality
if privacy is invaded then confidentiality should be protected, it refers to our right to have any personal data protected
Protection from Harm
ppts should not be placed at any more risk than they would be in their daily lives, and should be protected from psychological and physical harm
Deception
means deliberately misleading or withholding information from ppts at any stage of the investigation
Ideal Consent Form
what will they be asked to do and why
things they may experience
how there data will be used
time duration
remind them of right to withdraw
give options
check box for participation
give opportunity to ask questions
signature, name, date
over 16 can sign it under has to be parent
Alternative ways of getting consent
presumptive consent - ask a similar group of people if agreed, consent of ppts is presumed
prior general consent - ppts agree to take part in study they are consenting to being deceived
retrospective consent - asked after having already taken part
Dealing with Ethical Issues (deception and protection from harm)
ppts offered a full debrief of experiment, contains true details of experiments, aims, conditions, what data will be used for and given right to withdraw data, ppts may require therapy researcher should provide
Case Studies
case studies are in depth investigations of a single person, group event or community, typically data is gathered from a variety of sources and by using several different methods
Case Studies Strengths
provides detailed information
provides insight for further research
permitting investigation of otherwise impractical or unethical situations
Case Studies Limitations
providing little bias for generalization of results in the wider population
researchers own subjective feelings may influence the case study
difficult to replicate
time-consuming and expensive
Pilot Studies
a pilot is a small scale run of the actual investigation and may involve only a few ppts
they are run for: experiments, self-report measures, observations
Pilot Studies Evaluation
used to test the experiments effectiveness and make improvement before the real thing
allows them to identify issues early then rectify them saving time and money
any part of the study could be measured e.g. validity, how long it takes, if instructions are too complicated
Naturalistic Observation
takes place in the setting or context the target behaviour would usually occur
Controlled Observation
there is some control over variables to observe effects and also control extraneous variables
Participant Observation
the observer joins in the scenario being observed
Non-participant Observation
the observer is separate from scenario being observed
Covert Observation
the ppts dont know that they are being observed
Overt Observation
the ppts know that they are being observed
A03 Naturalistic vs Controlled
ecological validity - naturalistic because it happens in a natural setting high EV
extraneous variables - controlled, less variables
replicability - can be replicated in controlled setting
A03 Covert vs Overt
ethics - in overt situations people know they are being filmed
DC - covert you wouldnt know you are being observed therefore behaviour wouldnt change
A03 Participant vs Non-participant
more insight - participant, would understand ppts feelings
objectivity - non-participant, will be more objective
Behavioural Categories
when a target behaviour is categorised into components that are observable and measurable, something you can see no interpretation
categories must be exclusive and not overlap
Event Sampling
count every time a behaviour occurs during observation
Time Sampling
count behaviours that occur in smaller time frames e.g. every five minutes count for 1 minute
Time Sampling Evaluation
could miss infrequent behaviours
less effort for researcher
Event Sampling Evaluation
can catch infrequent behaviours
more effort for researcher
Questionnaires
set of written questions - designed to collect information
can discuss what people think or feel
always pre-determined
can be quantitative or qualitative
used to asses DV
Quantitative
numerical - closed
Qualitative
non-numerical - open
Writing Good Questions
clarity - do not use double negatives
jargon - refers to technical terms that are only familiar to those in a specialised area
bias - do not use leading questions which make one answer more attractive than the other
analysis - questions need to be written so they are easy to analyse (open questions - hard to analyse, closed questions - easy to analyse)
Likert scale
respondents can indicate the extent to which they agree or disagree with a statement, usually five levels
Rating Scale
a means of assessing attitudes or experiences by asking a respondent to rate statements on a scale of 1 to 3 to 5
Fixed Choice Option
questions with pre-determined number of answers
Questionnaire Strengths
once you have designed a questionnaire you can use them for lost of people cheaply and quickly
respondents may feel comfortable to reveal personal information
can be completed without the investigator being present
reduces experimenter bias which can be found in interviews
no special experience needed to hand out questionnaires
questionnaires with open questions can provide unexpected answers which can lead to further research
Questionnaire Limitations
can only be filled out by those who read and write
questionnaires using closed questions, limit response however easier to summarise
social desirability bias - answering questions in a way so you are deemed socially acceptable dont have to be honest
acquiescence bias when a ppt just ticks and agrees with all questions
Structured Interviews
has pre-determined questions, face to face questionnaires
no deduction from original questions
Unstructured Interviews
new questions developed during the course of the interview
interview may begin with general aims
more like a conversation
Semi-structured Interviews
many interviews fall in between
job interview is semi-structured
list of questions worked out in advance but interviews are also free to ask follow up questions when they feel appropriate
Structured Interview Evaluation
+ can be easily repeated because questions are standardised
+ different people can be compared due to the standard questions
+ answers are easier to analyse due to being predictable
- comparability if interviewer is different or behaves differently
- interviewer bias
Unstructured Interview Evaluation
+ more detailed information can be obtained
- requires interviewer with more skills to develop questions
- questions may lack objectivity due to quick creation
- more expensive due to need for experienced interviewer
- interviewer bias
Design of Interviews
record interviews to make sure all information taken down, visually or audially
schedule standardised for ppt to reduce effect of interviewer bias
an interviewer can effect answers by asking leading questions or hinting at the answers
body language - open interested, closed uninterested
Effect of Interviewer
presence of an interviewer who is interested in the ppts answers may increase amount of information provided and vice versa
what are descriptive statistics
they show patterns and trends in data
what are the measures of central tendency
average - mean/median/mode
what are the measures of dispersion
spread - range/standard deviation
what are inferential statistics
they tell us of the likelihood that iv affected the dv
p < 0.05 this is the conventional levels, 5% or less likely that a chance factor affected dv
95% or more likely iv affected dv
when can probability be lower
in drug trails it can be p < 0.01
1st step in choosing a stats test
test of difference (experiment between two conditions) or association - correlation
2nd step in choosing a stats test
experimental design
independent groups - unrelated
matched pairs - related
repeated measure - related
3rd step in choosing a stats test
level of data you are using
nominal/categorical
ordinal data - can be ranked difference between rankings is not standardised
interval - ranked with standardised gaps, published scale e.g. height, weight, temp
a test for: nominal data with independent groups
chi2
a test for: ordinal data with independent groups
mann-whitney
a test for: interval data with independent groups
unrelated t-test
a test for: nominal data with matched pairs or repeated measures
sign test
a test for: ordinal data with matched pairs or repeated measures
wilcoxon
a test for: interval data with matched pairs and repeated measures
related t-test
a test for: nominal data with a test of association
chi2
a test for: ordinal data with a test of association
spearmans
a test for: interval data with a test of association
pearsons
acronym for stats test
milk - mann-whitney
will - wilcoxon
spoil - spearmans
unless - unrelated t-test
refrigerated - related t-test
properly - pearsons
which test wont always give you the calculated/observed value
sign test
type 1 error
error of optimists
accept experimental, reject null
when p value is too big
false positive
type 2 error
error of pessimists
accept null, reject experimental
when p value is too small e.g. p<0.01
false negative
types of reliability
internal reliability
external reliability
internal reliability
the extent to which something is consistent within itself, we asses internal reliability by using the split half method where we compare scores from one half of the test with scores on the other half
external reliability
the extent to which a test is consistent over several different occasions if the same interview was conducted with the same people one week and the next the outcome should be the same
test-retest reliability
testing the same group of patients on more than one occasion and checking that there is significant positive correlation between the two sets of scores, sufficient time must be allowed between tests so that the ppts dont remember the answers
inter-observer reliability
establishing reliability in observational research by using more than one observer to rate or code the different types of behaviour, two observers making recordings at the same time avoids research bias
ensuring reliability with the experimental method: standardisation
in an experiment make sure that the conditions are the same, everyone receives standardised instructions, everyone does the same task, all extraneous variables are controlled. this is to ensure that the research can be replicated
ensuring reliability with the experimental method: test/retest reliability
checks for variation in measurement taken by a single person with a single measurement tool for the same type of behaviour if the correlation between the two sets of scores is high then the measurement tool has good retest reliability
how to improve reliability
taking more than one measurement from each ppt and calculating an average score
pilot studies - these ensure that the measure works and that all concerned can use the apparatus
training researchers properly: particularly when a number of researchers are collecting data, this improves inter-rater reliability, a pilot study can also help standardise procedures and data recording techniques
operationalising terms precisely so that all observers categorise behaviours in the same way
validity
does it measure what it is meant to measure
types of validity
external validity
- ecological validity
internal validity
external validity
can the results be generalised from the setting in which they are obtained
ecological validity
do the results reflect a persons real life behaviour, research carried out in a artificial environment like a lab does not
internal validity
does the experiment really measure the effects of the iv on the dv
further threats to validity
social desirability bias, investigator effects, order effects
how to control internal validity
consent, not informed consent (deception), independent groups design
ways of assessing validity
face validity
concurrent validity
predictive validity
face validity
do the questions look like they measure what the researcher intended to measure, are they obviously related to the topic
concurrent validity
check for this with a new research tool against an established tool, if the new research tool gives similar results than it has concurrent validity
predictive ability
check for this by assessing two sets of data at different times e.g. look at the correlation of an initial diagnosis of mental state compared to the actual state over time, if the predicted behaviour at the initial diagnosis has occurred, the diagnosis has validity
ways of improving validity and reliability
experimental research
questionnaires
observations
qualitative methods
how are validity and reliability related
a study may be reliable but not valid
however if a measurement is not reliable then it can not be valid
what should be included in a psychological report
title, abstract, introduction, aim, hypothesis, experimental/alternative hypothesis, method, procedure, results, discussion, references, appendix
what is in the abstract
this is always written last because it is a summary of the report about 150 words, abstracts provide an efficient way of gaining information without having to read an entire study, they contain all the essential information that a researcher would need to know about replicating the study
what is in the introduction
this tells everyone why the study is being carried out and the commentary should form a funnel of information, first there is a broad coverage of the background research, then the general research has been covered, the focus becomes much narrower finishing with the main researcher/research area you are hoping to refute. this then leads to the aim and the hypothesis
what is the aim
this covers exactly what it is you are hoping to find out and how
what is the hypothesis
this is a short testable statement, the independent variable must be operationalised and the dependent variable measurable, justification must be given for whether the test is to be directional or non-directional
what is the alternative hypothesis
going against the normal hypothesis
what is the method
this section is split inro sub-sections
design: experimental, non-experimental method used -laboratory/field/natural or correlation or survey method
design type - independent groups, related measures, matched pairs, correlation, observation, interview, IV, DV, EV, use of counterbalancing/ measures taken to avoid bias
ethical issues, participants, relevant details of target population - age/socio-economic status, gender, etc
relevant details of sample population, sampling method used, allocation to conditions, report of those who dropped out, materials, description of equipment used and how to use it
what is the procedure
this is a step by step guide of how the study was carried out - when, where, how, including details of how variables were operationalised, instructions to ppts must be standardised to allow replication
what is the results
this ection contains only a summary of the data, all raw data and calculations are put in the appendix, this generally starts with a section of descriptive statistics - measures of central tendency and dispersion. graphical representations of the data must also be clear and properly labelled and referred to in the text. one the summary statistics have been explained, there should be an analysis of the results of any inferential statistics, including observed values, how these related to the critical table value, significance level and whether the test was one or two tailed, this section finished with the rejection or acceptance of the null hypothesis
what is the discussion
begins with a statement of the findings and how these relate to the original hypothesis, all results are reported even if they do not fit in the hypothesis or science would not progress, all studies have flaws, so anything that went wrong, or the limitations of the study are discussed together with suggestions of how it could be improved if it were to be repeated . suggestions for alternative studies and future research can also be explored, the discussion ends with a paragraph summing up what was found and assessing the implication of the study and any conclusion that can be drawn from it
what is the refernces
every researcher cited in the text must be fully refernced using the harvard system, referncing is time consuming but is important in order to prevent plagarism
