Research Methods Flashcards
Falsifiability
The possibility that a statement or hypothesis can be proved wrong via testing
Objectivity
Measurement of data is not affected by researcher expectations
Objectivity means reducing individual differences to make it more scientific
Replicability
Recording procedures carefully for another researcher to repeat and verify the original results, increasing validity of findings
Empirical methods
Methods which rely on direct observation or testing, like brain scans or blood tests
Paradigm
A shared set of assumptions about a subject matter of a discipline and the methods appropriate to its study
Paradigm Shift
Scientific revolution where the accepted paradigm is questioned and this gains popularity to the point the contradictory evidence becomes a new, more dominant approach
Steps of induction (theory construction) OHSCT
Observations
Testable hypothesis
Conduct a study
Draw conclusions
Propose a theory
Steps of deduction (theory construction) OTHSC
Observations
Propose theory
Testable hypothesis
Conduct a study
Draw conclusions
How should a hypothesis be tested?
Using systematic and objective methods to determine support or rejection
Experimental methods:
Lab
Field
Natural
Quasi
Lab experiment
Uses a carefully controlled setting and standardised procedure to measure how the IV affects the DV
The participants are aware they are in it but may not know the aims
Lab experiments Strengths
Greater control over variables
Easily replicable so results can be tested and compared
Standardised procedure increases internal validity
Lab experiments Weaknesses
Demand characteristics- when participants figure out the aims they can alter behaviour
Artificial environment lacks ecological validity
Field experiement
Uses control of lab experiments but in real-world settings. The IV is still deliberately manipulated and the DV measured
Usually participants unaware
Field experiment
Strengths
High ecological validity due to being in real-life settings
Reduced chance of demand characteristics- naturally environment=more natural behaviour
Field experiment weaknesses
Less control as it is harder to manipulate the IV and record the DV
Ethical problems- participants may be unaware they are observed
No controls so more difficult to replicate
Natural experiment
Study of a naturally occurring situation, no influence over the situation but observes individuals and circumstances
Natural experiment advantages
High ecological validity- natural settings, high relativity to real life behaviour
Less chance of demand characteristics- less likely to adjust behaviour as they are unknowingly being observed
Natural experiments disadvantages
Ethical issues- participants may be unaware they are participating in the study(deception)
No control over extraneous variable so affect of IV are not always clear
Quasi experiments
Naturally occurring IV, e.g. gender, age, disorder, control (DV still measured)
Quasi experiments
Advantages
Allows for comparison between types of people- no manipulation is carried out but results show differences
Can be done in lab- DV tested in a lab so high control/ replicability
Quasi experiments
Disadvantages
lab setting- low ecological validity
lack of random allocation-participants cant be randomly allocated so there may be confounding variables
Experiments demonstrate what
Cause and effect
Independent variable
Variable that is manipulated
Dependent variable
Variable that is measured
Operationalisation
Process of making an abstract construct/variable to something measurable
Extraneous variables
Anything other than the IV that affects the DV. These variables can be controlled by the experimenter
Confounding variables
Variables that aren’t controlled and affect the results
Research aims
Stated intentions of what questions are planned to be answered
Hypothesis
A formal, unambiguous statement of what is predicted
Main features of a hypothesis
Contains conditions of the IV and expected DV outcome
Be operationalised and measurable
Directional hypothesis
States whether the DV outcome is expected to be greater or lesser, positive or negative
e.g group A, who used a mnemonic will score significantly higher than group B who didnt
Non-directional hypothesis
Doesnt state the direction of the DV, just that there is a difference
e.g. there is a significant difference between the scores from group A and group B
Null hypothesis
A prediction of no difference between the two IV conditions on the outcome of the DV
Reliability is…
consistency
Internal reliability
each person in a study is treated the same way
External reliability
Same/similar results found after repeated test
Test-retest reliability
Test the same person twice
same sample, same test, ensure a time gap
Inter-observer reliability
Compares observations from different observers- reduces bias
Correlational reliability
should exceed +80 on Pearsons correlation coefficient which tells us if IV has enough influence
Validity is….
Accuracy/ representativeness
Internal validity
is it measuring what its meant to measure?
External validity
is it generalisable beyond the experimental setting?
Ecological validity
Population validity
Temporal validity
Realistic setting?
Applicable sample?
Does it stand “the test of time”?
Face validity
Surface level, does it look like it measures what it should?
Concurrent validity
Are findings similar to those on a well established test
2 tests correlating similarly as a check of accuracy
3 ways to improve validity
Representative sample
Larger sample size
Realistic setting
Independent groups
Two groups exposed to different conditions
Random allocation and the DV is measured for each and compared
Independent groups advantages
No order effects
Data collection is easier and faster
Independent groups disadvantages
Finding different participants- 2x
Risk of individual differences affecting
Repeated measures
One group takes part in each condition then compare results
Repeated measures advantages
Controlled variables- same group of pp
Fewer people- economic advantage
Repeated measures disadvtanges
Order effects- boredom fatigue from repeats
Demand characteristics
Matched pairs
Recruit and group based on relevant characteristics then treat like independent variables
Matched pairs advantages
Reduces participant variables
Less order effects- one condition
Matched pairs disadvantages
Time-consuming matching
Less economical especially with a pre test
Demand characteristics
Cues that might indicate study aims to pp and cause change in behaviour
Investigator effects
When a researcher unconsciously influences the outcome of research
Single blind
Double blind
Participants are not aware
Participants and experimenters both not aware
Control group
A group in which a variable is not being tested
Confederate
A secret participant which makes the cover story more real and capture naive reactions
Randomisation
Standardisation
Randomly assigning subjects to many potential influences that cant be controlled
Keeping everything the same= fair
Pilot study: what, why, advantages
Small study to test research protocols, strategies, data collection instruments in prep for a large study
Why: tests aspects for a larger indepth investigation
Adv: identify potential problems and fix, avoids time/resource wasting
Opportunity sampling
Participants selected based on availability
Opportunity sampling
Advantages and disadvantages
+ Fastest way, reduces time and costs
- researcher bias(select participants for desired result)
-unrepresentative(easy access)
Random sampling
Eaxh sample has an equal chance of being selected
Random sampling
Advantages and disadvantages
+Avoids researccher bias
+easy way of sampling
-unrepresentative of all minority groups
-difficult and time consuming
Stratified sampling
Dividing subjects into subgroups/strata based on shared characteristics
Stratified sampling
Advantages and disadvantages
+Representative of target pop
+Random chosen within each stratum
-Not every characteristic will be represented
-Time consuming to establish strata then select
Systematic sampling
Select participants of the population at regular intervals
Systematic sampling
Advantages and disadvantages
+Avoids researcher bias
+If list already exists of targets=quick
-Unrepresentative sample
-Big target pop, too hard
Volunteer sampling
Researchers seek volunteers to participate
Volunteer sampling
Advantages and disadvantages
Easy sample to collect-self presented
Reach a large no of potential participants
Generalisability to target population?- Volunteers have diff characteristics
Volunteer bias- friendlier, more freetime
informed consent
permission to use participant data and selves, should be informed on anything that could affect willingness
could also be parental consent if the child is too young
dealing with informed consent when its broken
retrospective- gives consent for data use after debrief
presumptive- similar group is assumed
prior general- give permission to take part where they may be deceived
informed consent right to withdraw
have to have it and tell if you use any of the above methods
deception
no lies but sometimes it is avoidable to keep pps naivety
cost benefit analysis should be used to decide if its acceptable
dealing with deception if broken
pps receive an immediate debrief in full detail
protection from harm
risk or stress
protection from extremely damaging to physical or psychological state
protection from harm dealing
debrief
right to withdraw
counselling
what should be in a consent farm
info about the study- aim and procedure
right to withdraw
confidentiality
opportunity to ask questions
statement for them to sign
written debrief form
thanks and explain the purpose of the study and predictions
remind of confidentiality and if decepton is used explain why
suggest counselling if necessary and right to withdraw
inform if they want any info about results they should contact and ask them to not talk about it when its still running
and provide names if anyone wishes to contact
qualitative data
data representing non numerical info, language associated
qual data collection
interviews
diaries
lab notebooks
questionnaires
qual data adv and disadv
+more detailed to explain human complexity
descriptive nature allows more analysis
-time consuming
harder to analyse as categorising is harder
quantitative data
data from measures of values, expressed as numers
quant data collection
experiments
surveys
polls
probability sampling
quant data adv and disadv
+easy analysis
consistency due to its objectivity
-difficult to explain complex issues
hard to analysis correlation
primary data
original data from researchers own research
primary data evaluation
+accuracy
reliability
- lost and time consuming
researcher bias
secondary data
info collected and available for use by researchers who did not create it themselves
secondary data evaluation
time efficent- analysis is quicker and cheaper
- lack of accuracy- incomplete or inaccurate data
not up to date- lack of applicability
nominal data
qualititave values, usually tallied
eg. age, ethnicity
ordinal data
scaled/ ordered data, subjective ratings usually 1-5
scales, ranking
interval data
ranked data with equal intervals and units
eg. temperature, precipitation, time
ratio data
same as interval but with an absolute zero/baseline
eg. distance, weight, cash
measure of central tendency
how sets of results are measured and analysed
eg. mean, mode, median
measure if dispersion
how spread out the data is in a set of results
eg. range, semi interquartile range, standard deviation
mean corresponds with which data type
interval/ratio
mode coresponds with which data type
nominal
median corresponds with which data type
ordinal
normal distribution
naturally occuring symmetrical bell curve
more in middle few on side
68% on one side of standard deviation, 95% within 2 SD of mean
negatively skewed distribution
more at the higher end, outliers at the lower endp
positively skewed distribution
more scores at the lower end, outliers at the higher end
probability
how likely something is to happen between 0 and 1
conventional use of p<0.05 significance level
likelihood of behaviour happening is equal or less than 5%
any higher= significance
lower= hard to achieve, only needed for medical research
challenging other research
psychologists adapt more stringent significace levels, 0.02/0.01
proof
does not exist in psychology research unless 100% accuracy is found, 0.05 is enough to show support
type one error
belief that significant difference/correlation is found but it doesnt exist
type two error
belief that no significant difference/correlation is found but there is one
avoiding type 1 or 2 errors
stringency, increasing the sample size
inferential statistics
stat test is done to determine whether a difference found in results is statistically significant and not by chance
when to use a sign test
nominal data
checking for differences
repeated measures experimental design
matched pairs experimental design
how to calculate sign test
state null and alternative hypothesis
represent each pair of data with plus of minus to work out S value
look up critical value of S by knowing N and whether it is two tailed or one tailed
compare critical t compared- if calculated is equal or less than value is significant
what do formal writeups always include
title
abstract-summary paragraphs
introduction- backgrounds and hypothesis aim
method-designs, participants, materials
results- data types of stats, tests done for significance
discussion- analysis of data, support of hypothesis
references
appendices- forms, stats
