research methods Flashcards
non-experimental research methods
observations
self report techniques-questionnaire
interviews
correlation
eg. experimental research methods
lab experiments
field experiments
natural experiments
quasi experiments
self report
person asked or explained their own feelings, opinions, behaviours or experiences-related to given topic
lab experiments
highly controlled researcher manipulates IV + records effect on DV
lab experiments-strengths
establish causes + effect
easy to replicate
remove extraneous variables
lab experiment-weakness
demand characteristics
low ecological validity
field experiment-strengths
higher ecological validity
less demand characteristics-less artificial
field experiment
takes place in natural everyday setting-researcher manipulates IV + records effect on DV
field experiment-weaknesses
not possible control + eliminate extraneous variables in field so impact on DV
difficult to replicate-in natural environment-not same if replicated
natural experiment
takes place in natural setting
IV not manipulated by researcher to have an effect on DV
natural experiment-strengths
higher ecological validity
less likely to demonstrate demand characteristics
natural experiment-weaknesses
not possible control + eliminate extraneous variables in field so impact on DV
difficult to replicate-in natural environment-not same if replicated
quasi experiment
IV based on existing differences in ppl-no one has manipulated variable it simply exists
quasi experiment-strengths
highly controlled-establish cause + effect-if lab
high ecological validity-if natural/field
quasi experiment-weakness
if lab-low ecological validity
demand characteristics
have confounding variables
naturalistic observation
natural setting
Ps in own environment + interference-kept to minimum
can be observed or done secretly
naturalistic observation-strengths
high in ecological validity
less demand characteristics
naturalistic observation-weaknesses
cannot control extraneous variables
difficult to replicate
controlled observation-strength
easy to replicate
easy to check reliability of findings
unwanted extraneous variables eliminated
controlled observation
highly controlled researcher manipulates variables + observes Ps behaviour
controlled observation-weakness
demand characteristics
low ecological validity
covert observations-strengths
no demand characteristics
allows to explore behaviour-private or secretive eg.criminal behaviour
covert observations
observations done secretly
covert observations-weaknesses
ethical issues eg.lack of informed consent
difficult to record behaviour w/x being discovered
overt observations
observations done openly
overt observations-strength
fewer ethical issues-Ps know that their taken part in an observation
researcher can found out more info about them-find out reasons for Ps actions
overt observations-weaknesses
behaviour not natural-observer/investigator effect-can lead to demand characteristics
researcher might find it difficult to recruit Ps willing to take part
double blind procedure
researcher assistant-doesnt know full aim
so can’t give clues to Ps
record observation in less bias way
participant observation
observer gets involved + Ps in behaviour of group observed
can be done overt or covert
participant observation-strengths
researcher will have fuller understanding of actions of group
Ps will have natural behaviour
participant observation-weakness
researcher becomes more of P than observer-difficult to be objective + step back about observation
difficult to record behaviour w/x being discovered
non-participant observations
researcher follows group around but doesn’t get involved
non-participant observations-strengths
researcher is not interfering w/behaviour being observed
able to remain objective
extraneous variables-situational variables
aspect of research situation that might influence Ps behaviour
non-participant observations-weaknesses
might not fully understand actions of group
presence of observer can change behaviour of group
confounding variables
uncontrolled extraneous variables that have affected at least 1 of condition
researchers could not be sure whether differences in homework performance was due to presence of music or intelligence
extraneous variables-participant variables
characteristics or traits of Ps that may affect results
investigator effects
unwanted influence of investigator on DV
eg. personality, gender, age of researcher
researcher may also be biased when selecting/allocating Ps + when recording their data
randomisation-strengths
minimise effect of extraneous/confounding variables
prevents investigator effects in allocation of Ps + reduces unconscious bias
counterbalancing
used to deal w/order effects when using repeated measures design
Ps sample is divided in half w/1 half completing 2 conditions in 1 order + other half completing conditions in reverse order
standardisation
all Ps should be subject to same environment, information + experience
ensure this all procedures + instructions are standardised + kept same
behaviour categories
behaviour checklist w/different behaviour categories
behaviour categories-strengths
tallies= easy to quantify data + use graphs-compare to qualitative data
more scientific + objective way of caring out observation-standardised way
easy to replicate + check for reliability
time sampling
observing at different time intervals
eg.1h observe, 1h not observe
strength-reduces no. of observations made
weakness-but could miss important info
behaviour categories-weaknesses
lack of inter-observer reliability-different results obtained by 2 different observers-have different views
observers have quite lengthy training=costly
other methods of recording data in observations
note taking-notes taken away-observer tries to identify patterns in behaviour
audio/video recording-not always practical
event sampling
observer focuses on specific pre-selected behaviour-their interested in + record every time it occurs
strength-useful when event happen infrequently
questionaries
standardised questions-handed out to Ps-supposed to be filled by Ps
closed questions
set of pre-determined answers
open questions
Ps express their ideas + opinions
Likert scales
indicates strength of agreement
rating scale
indicates strength of feeling
fixed choice option
Ps just tick from range of options
questionaries-strength
easily disturbed to Ps
obtain large sample of Ps
generate lot of data
questionnaires-weakness
socially desirable answers-appear in favourable light to researcher
lead to leading questions-urge Ps to give certain response
open question-strength
Ps can fully express themselves-in depth + meaning
generate lots of qualitative data
fuller understanding of behaviour observed
open question-weakness
very time consuming to analyse + draw conclusions from
pilot study
small scale trial run of any method eg.observation, lab experiment
done to ensure that Ps understand all Qs, material + instructions
help iron out any difficulties before main study
closed question-strength
easy to draw + analyse conclusions from
easy to statistically show data
closed question-weakness
lack deep meaning + data
no full understanding of behaviour researched
structured interviews
interviewer verbally asks questions
questions=pre determined
structured interviews-strength
contain standardised questions
easy to replicate
check for reliability
interviewer can explain Qs Ps don’t understand
structured interviews-weakness
socially desirable answers-appear in favourable light to researcher
interviewer effect-where age, personality, gender, ethnicity of interviewer affect responses
unstructured interview
conversation between Ps + interviewer
no standardised questions
unstructured interview-weakness
not standardised-diffucult to replicate
findings-not consistent + unreliable
difficult to analyse + draw conclusions from
unstructured interview-strength
rich, detailed, qualitative data
researcher can steer interview-in any direction-researcher can probe + ask Ps to expand on it
correlations
relationship between 2 variables-not cause + effect
2 variables= co-variables
positive correlation
high score on 1st variable associated w/high score on 2nd variable
negative correlation
1 variable increases= other variable decreases
no correlation
no relationship between data score
correlation co-efficient
number between 1 + -1 which shows strength or relationships-closer to 0 weaker= relationship between 2 variables
sign (+ or -) shows whether relationship is strong or weak
correlation-strength
allows relationship of 2 variables to be examined-when controlled experiment not possible due to ethical issues
good starting point for further research-produces quantitative data
correlation-weakness
can be misused-as finding correlation between 2 variables tell us very little other than relationship just exists
operationalisation
making sure variable can be easily measured
aim
general statement of intended purpose of study
investigate theories that have been developed
contain variables being investigated
aim-what researcher wants to find out
hypothesis
prediction about what will happen in study-precise + testable statement
can be directional OR non-directional
directional hypothesis
AWARE of any past research-results have similar outcome
makes clear sort of difference that is anticipated
predict why way results will go
non-directional hypothesis
UNAWARE of any past research OR findings unclear or contradictory
safer to use non-directional hypothesis in case findings go in either direction
states there is difference-but doesn’t predict which way results go
null hypothesis
statement which predicts that IV will NOT affect DV-so NO significant difference
directional correlation
specific type of relationship between 2 co-variables-eg. positive OR negative
non-directional correlation
states that there will be relationship between 2 co-variables-but doesn’t state whether positive OR negative
null correlation
NO correlation between 2 co-variables
ethical issues
potential for Ps to be harmed in some way during research-role of BPS encourages researchers to follow BPS guidelines
ethical issues: informed consent
researcher must attempt to get real content from Ps-only possible when Ps fully understand what they are agreeing to do-consent for children should be from parents
ethical issues: deception
Ps should be given all info + not lied to before the study-but in order to collect valid data Ps may not be told entire truth-minimal degree of deception should be used
ethical issues: confidentiality
data from Ps should be protected under Data Protection Act- Ps should be aware of what their data is used for-their confidentiality= respected
ethical issues: right to withdraw
Ps must be allowed to stop participating in study OR stop study altogether in order for research to follow ethical guidelines
ethical issues: protection from harm
Ps should be protected from harm by researcher + study should not be designed to deliberately cause harm-harm is both physical + emotional distress
ways of dealing with ethical issues: debriefing
used for deception OR psychological harm
fully inform Ps about nature of research + Ps allowed to discuss any issues they have
Ps have experienced any harm-debriefing offering counselling + advice
ethical issues: privacy
names of Ps should not be recorded so they cant be identified
repeated measure- strength
P variables are minimised bc same Ps take part in both conditions of experiment
strength bc it increases internal validity of research
half number of Ps is needed compared to other 2 designs bc Ps take part in both conditions of experiment
strength bc it means researcher can save time money by using same Ps for both conditions
repeated measures- weakness
order effects may occur means that order in which Ps complete conditions may affect their performance eg. boredom, practise or fatigue effects can occur as Ps are taking part in more than 1 condition
Demand characteristics can also be more likely- limitation bc if order effects do occur internal validity of research is lowered
repeated measures
1 group of Ps who take part in both conditions of study
independant groups
Different Ps are used in each condition of study
each P only experiences 1 condition of IV
independant groups-strengths
no order effects bc Ps are only taking part in 1 condition of experiment-reduces possibility of boredom, practice + fatigue effects occurring + reduces chance of Ps guessing aims of experiment + changing their behaviour
independant groups-weakness
Ppt variables can occur bc there may be individual differences between 2 groups of Ps that could affect results
matched pairs
different Ps used in each condition but they are matched on variable that could affect results if left unchecked
matched pairs-strengths
P variables are minimised bc Ps are matched on important variables therefore individual differences between groups are unlikely + so there is higher internal validity
not affected by order effects as Ps only take part in 1 condition P performance cannot be affected by boredom, practise, or fatigue as they do not take part in 2 conditions as in repeated measures design strength bc it increases internal validity of research
matched pairs-weakness
time consuming bc Ps are often pre-tested to match them up eg. to match Ps on intelligence all Ps must take an IQ test
if 1 partner of a pair drops out researcher risks losing both members- makes it less economic than other designs
random sampling
every member of target population has an equal chance of being selected
opportunity sampling
selecting people who are willing + available to take part at time of research
systematic sampling
every nth member of target population is selected
volunteer sampling
people put themselves forward to participate
Volunteers usually respond to newspaper or university noticeboard adverts that are placed by researchers
stratified sampling
involves researcher dividing population into subpopulations
Researchers then ensure each subgroup is represented in their sample
researcher 1st identifies different strata that make up population proportions needed for sample to be representative are worked out
random sampling-strength
unbiased bc researcher does not have any influence over who will be selected for sample means that sample will be free from researcher bias
equal opportunity of being selected increasing representativeness of sample
random sampling-weakness
Ps selected may not be available/ refuse take part-researcher will have small sample size so time consuming
random sample could just contain only males OR females Ps which makes sample bias
opportunity sampling-strength
Quick + convenient method bc researcher just makes use of people who are available at time most popular
opportunity sampling-weakness
likely to be bias- bc researcher influences who is asked to take part
Ps may support their hypothesis-unrepresentative + lack population validity
volunteer sampling-strength
Ps more motivated to take part-volunteer w/interest
Ps more likely to take it more seriously
volunteer sampling-weakness
biased sample bc often Ps who volunteer share certain characteristics or traits eg. are keen + helpful
problems when attempting to generalise findings from such biased sample
systematic sampling-strengths
researchers selection of Ps is not biased
stratified sampling-weakness
very complex + time consuming
systematic sampling-weakness
time consuming + not everyone in target population has an equal chance of being selected
Ps may refuse to take part
stratified sampling-strengths
representative of target population since characteristics of target population are represented proportionally
more likely that findings from this sample can be generalised
ways of dealing with ethical issues: confidentiality
personal details are held these must be protected
more usual to simply record no personal details eg. maintain anonymity
Researchers usually refer to Ps using numbers or initials when writing up investigation
purpose of conducting pilot study
aims to find out if aspects of design do or don’t work
eg. if Ps understand instructions if timings for tasks are appropriate or if parts of design make aims of research obvious.
conducting pilot study on small group of people it is possible for researcher to see what needs to be adjusted before investing time + money in larger scale research study
case studies
detailed study of an individual, group, or situation
often involve an analysis of unusual individuals or events such as person w/rare disorder
Case studies tend to take place over long period of time
When constructing case study researchers often include case history of individual concerned, using interviews, observations + questionnaires
case studies-strengths
Case studies provide rich + valuable insights on very unusual + atypical forms of behaviour allows researcher to investigate topic in far more detail than might be possible if they were trying to deal w/many research Ps as in an experiment findings are based on real life problems + issues increasing ecological validity of research
Case studies provide great deal of qualitative data that often generates ideas for future research
case studies- weakness
possible to generalise findings from single individual or small sample to wider population limitation bc it means findings are only representative of person whom study is focused lacking population validity
case studies are criticised due to their subjective nature bc researcher must decide which information to include in final report + must interpret vast quantities of qualitative data which is produced personal accounts from P + their family + friends may be prone to inaccuracy + memory decay especially if childhood experiences are being relayed lowers the validity of evidence from case studies
event sampling-strengths
relevant behaviours are not missed
event sampling-weakness
Observations based on event sampling may not take in account broader contextual factors that influence child’s behaviour
time sampling-strengths
manage observations more rather than being overwhelmed by every single behaviour that occurs
time sampling-weakness
behaviours sampled may be unrepresentative bc relevant behaviours displayed outside time frame are missed
Inter-observer reliability
data recording more objective + unbiased observations should be carried out by at least 2 researchers
improve reliability observers should familiarise themselves w/ behaviour categories being used
After observing same behaviour at same time observers should compare + discuss any differences in interpretation
qualitative data-strengths
rich in detail so you are more likely to find out more about topic being studied
more holistic understanding of phenomena under study
qualitative data-weakness
conducted w/small sample sizes
difficult to draw conclusions from
time consuming
quantitative data-strengths
objective
easy to draw conclusion from
not time consuming
quantitative data-weakness
less detailed data
open to misrepresentation
primary data
data collected by researcher specifically for purposes of their study
data comes first hand from the Ps
Data gathered using an experiment, questionnaire, interview, or observation would be classed as primary data
primary data- strengths
more accurate + reliable bc it comes from direct source
faster + easier to collect primary data than secondary data which can take weeks or even months to collect
primary data-limitation
Requires considerable planning, preparation + resources on behalf of researcher
secondary data
data that is collected by someone other than primary user
secondary data- strengths
allows researchers to investigate phenomena that cannot be tested now
Inexpensive requiring minimal effort and easy to access
secondary data- weakness
may be missing data that researcher is interested in investigating-limits utility
meta-analysis
research method that uses secondary data
where data from lots of studies already carried out is combined to provide an overall view on subject
meta-analysis may produce qualitative data eg. review of conclusions from research or quantitative data
meta-analysis: strengths
Results can be generalised across much larger populations
meta-analysis: weakness
difficult process to undertake bc it requires use of sophisticated tools
measure of dispersion: SD
Measures dispersion of scores around mean
higher standard deviation greater spread of scores from mean
low standard deviation number indicates that scores are close to mean
mean-strength
mean uses every value in data + hence is good representative of data
mean-weakness
unrepresentative if there are extreme values
median-strength
Not affected by extreme values
median-weakness
time consuming w/lot of data as it has to be put in order
mode-strength
Not affected by extreme values
mode-weakness
can be more than 1 mode + all values can be modal which means mode is not always representative of data
range-strength
Easy to calculate
SD-strength
Shows whether or not data is clustered around mean
Not affected by extreme values or outliers
range-weakness
Doesn’t take into account distribution spread of all numbers
SD-weakness
Difficult to calculate
Does not show full range of data
theory construction
develop theories all time to explain things we observe in our everyday life Scientific theories are constructed by gathering evidence
eg. may develop theory regarding capacity of short-term memory after series of experiments reveals that memory span is around 7
should be possible to make clear + precise predictions based on scientific theory an essential component of theory is that it can be scientifically tested
falsifiability
always be possible to prove a theory wrong eg. must have testable hypothesis
Popper says that rather than finding evidence to support theory scientists should actively try to find evidence to show that it is false
Freud’s psychodynamic theory is regarded as unscientific such as unconscious mind are impossible test + therefore cannot be proven wrong
Popper drew clear line between good science in which theories are constantly challenged + what he called ‘pseudoscience’ which could not be falsified
paradigm
Thomas Kuhn suggested that what distinguishes scientific disciplines from non-scientific disciplines is shared set of assumption + methods- paradigm
Some psychologists argue psychology is science bc it has paradigm but other say Psychology has too much internal disagreement + too many conflicting approaches to qualify as science
progress w/in scientific discipline occurs when there is a scientific revolution- handful of researchers begin to question accepted paradigm this critique begins to gather popularity + pace + eventually
paradigm shift occurs when there is too much contradictory evidence to ignore
peer review
assessment of scientific work by others who are experts in field prior to publication
aim of peer reviews
allocate research funding: Independent peer evaluation takes place to decide whether to award funding for proposed research project
All elements of research are assessed for quality + accuracy: formulation of hypothesis methodology chosen statistical tests used + conclusions drawn
Reviewers may suggest minor revisions of work + thereby improve report or extreme circumstances they may conclude that work is inappropriate for publication + should be withdrawn
peer review- weakness
Reviewers may use their anonymity as way of criticising rival researchers especially if findings contradict their own beliefs or research
slows down publication process especially when research findings are new + ground-breaking
not always possible to find experts in new area it can result in such work being judged by researchers who do not fully understand research
Publication bias is tendency for editors of journals to publish ‘headline grabbing’ findings to increase their credibility + sales tend to publish positive results- could mean that research which does not meet these criteria is ignored
peer review-strengths
acts as control mechanism to help prevent flawed or fraudulent research from being published
ensures that research published is academically rigorous + therefore can be trusted in comparison to opinion + speculation
encourages sharing of ideas between experts + collaboration on improvement of research
process is anonymous it is likely to produce an honest appraisal
format of scientific report
Title
Abstract
Introduction
Method
Results
Discussion
References
Appendices
characteristics of a normal distribution
mean, median + mode are all at exact same mid-point
data is symmetrical
consistent spread of scores on either side of mid-point eg. approximately 68% of data lies w/in 1 standard deviation of mean
Approximately 95% of data lies w/in 2 standard deviations of mean + 99.7% w/in 3 standard deviations of mean-‘empirical rule’
positive skew
most of scores are distributed left of graph
eg. very difficult test in which most people got low marks w/only handful at higher end
would produce positive skew positive skew, mean, mode + median are no longer in same mid- position
mode remains at highest point of peak mean is dragged to right towards tail
bc extreme scores affect mean very high scoring candidates in test have had effect of pulling mean to right
normal distribution
If you measure certain variables eg. height or IQ frequency of these measurements should form bell-shaped curve symmetrical spread of data is called normal distribution
W/normal distribution most people are in middle area of curve w/very few at extreme ends
mean, median + mode are in same midpoint of curve
negative skew
very easy test would produce distribution where bulk of scores are concentrated to right of graph
mean is pulled to left towards tail due to a few low scoring candidates
mode is not affected by extreme scores + is therefore at highest peak
both cases median lies between mode + mean
improve test retests
If correlation between 2 tests is lower than 0.8 researcher would need to review measures + then carry out another test-retest on new test
skewed distribution
spread of frequency data that is not symmetrical-data cluster to 1 end
ways of testing reliability-test retest
Ps are given questionnaire to complete + are then given same task on later occasion eg. 1 week later
Ps responses are then correlated to identify if they have given similar responses on both occasions
If correlation of 0.8 is established between tasks it is considered reliable measure
improving reliability-questionnaires
Rewrite confusing, leading or complicated questions
Avoid open questions as they could be misinterpreted
Internal validity
refers to whether research is measuring what it intended to measure
affected by presence of extraneous/confounding variables
improving reliability-interviews
Use same interviewer each time
Ensure interviewers are properly trained
Use structured interview
improving reliability-experiments
Use standardised procedure
Reword any confusing instructions
Use single or double-blind procedure
improving reliability-observations
Operationalise behavioural categories
Ensure categories do not overlap
Ensure observers are familiar w/ categories
external validity
refers to whether research findings can be generalised to other people, places + times
external validity-ecological validity
Generalising findings to real life settings
external validity-population validity
Generalising findings to other people in target population
external validity-temporal validity
Generalising findings to present day/modern
ways of assessing validity- face validity
extent to which test looks like it will measure what it is supposed to be measuring
could ask someone who has knowledge of area being investigated about if they think measure looks like it is measuring that topic area
ways of assessing validity- concurrent validity
extent to which test produces same results as another established measure
would compare score on new test w/score on test that has been proven to be valid
If valid 2 scores should be similar You can measure degree of similarity by correlating 2 sets of scores
correlation coefficient of above 0.8 would tell you that new measure/score is similar to valid measure/score + therefore you can assume new measure is valid
improving validity-questionnaires
Lie scales to assess consistency of responses
Anonymity to reduce social desirability bias
improving validity-observations
Covert observations
Ensure behavioural categories are not too broad
improving validity-experiments
Control groups
Standardised procedures
Single + double-blind procedures
content analysis
type of observational research in which people are studied INDIRECTLY
eg. instead of observing what people do in certain situation communications they produce are studied
aim of content analysis is to summarise this qualitative data + convert it to quantitative data
procedure for content analysis
data is collected
researcher reads through/ examines data-making themselves familiar w/it
researcher identifies coding units
data analyse by applying coding units
tally made of no. of times that a coding unit appears
coding
first stage of content analysis
Some data to be analysed may be extremely large + so there is need to categorise this information into meaningful units
may involve simply counting number of times particular word or phrase appears to produce quantitative data
coding units used will depend on data eg. newspaper reports may be analysed for number of times derogatory terms for mentally ill are used such as ‘crazy’ or ‘mad
eg. would be number of positive or negative words used by mother to describe her child’s behaviour or number of swear words in film
thematic anaylsis
analysing qualitative data by identifying patterns w/in material
material to be analysed might be diary, TV advertisements or interview transcripts
main process involves identification of themes theme refers to recurrent idea which keeps ‘cropping up’ in communication being studied
eg. mentally ill may be represented in newspapers as ‘drain on resources of NHS’
Such themes may then be developed into broader categories eg. ‘control’ or ‘stereotyping’ of mentally ill
in their final report researcher will use direct quotes from data to illustrate each theme
data here is NOT converted into quantitative data but stays in its written form
Content Analysis + Thematic Analysis-strengths
Content + thematic analysis can get around many of ethical issues usually found in psychological research
many resources eg. books + TV programmes already exist there is no issue w/getting permission to use it
resources used often have high external validity as they were designed for real life purposes
1) Familiarisation w/data – involves intensely reading the data + becoming immersed in its content
2) Coding – involves generating codes that identify interesting features of the data- Questions to consider whilst coding may include:
What are people doing?
What are they trying to accomplish?
3) Generating themes – involves combining codes to potential themes in order to identify meaningful patterns in data researcher then reviews themes to see if they work in relation to data
4) Defining themes - researcher then defines what each theme is + what is interesting about theme
5) Write up – researcher will write up final report typically using quotes from data to illustrate each theme
Content Analysis + Thematic Analysis-weakness
info is often studied outside of context in which it occurred + therefore researcher may attribute opinions or motivations that did not exist
research can lack objectivity as resources could be chosen which can reflect researchers aims
nominal data
‘discrete’ in that 1 item can only appear in 1 of categories
It is least precise level of measurement
eg. placing Ps into categories based on their gender + grade obtained in Year 1 Psychology
Counting no. of people who support Man United or Man City
ordinal data
does not have equal intervals between each unit
it would not make sense to say that someone who rated psychology as 8 out of 10 enjoys it twice as much as someone who gave it 4
Ordinal data also lacks precision because it is based on subjective opinion rather than objective measures
eg. what constitutes ‘4’ or an ‘8’ for people doing rating may be quite different
interval data
most precise level of measurement + consists of data that is measured on fixed, numerical scale w/equal distances between points on scale
Interval data is measured using equipment such as stopwatches, thermometers, weighing scales, which produce data based on accepted units of measures
eg. distance in centimetres, time in seconds
Levels of measurement + descriptive statistics
norminal=mode=n/a
ordinal=median=range
interval=mean=SD
sign test
research is looking for difference
data is nominal
research has used either repeated measures or matched pairs design
how to do the sign test
Once you have worked out sign for each pair of data you will need to find out calculated value represented as S for sign test
It is calculated by adding up number of plus signs in your table adding up number of minus signs in your table + selecting smaller value
Now you will need to find out if this result is significant or not
decide if it is significant you will need to compare your calculated value w/critical table value
critical table values are already worked out but you will need to select the correct value
decide what the critical table value is you will need to know:
1) total number of scores
This is your N value
2) Whether your hypothesis was directional or non-directional
3) level of significance to be used always 0.05 unless you are specifically asked to use different 1
paramedic tests
related t-test, unrelated t-test + Pearson’s r are collectively known as parametric tests
Parametric tests are more powerful + robust than other tests -3 criteria that must be met to use parametric test:
- Data must be interval level – actual scores, rather than ranked data is used
- data should be normally distributed
Variables that would produce skewed distribution are not appropriate for parametric tests - should be homogeneity of variance - set of scores in each condition should have similar dispersion or spread 1 way of determining variance is by comparing standard deviation in each condition if they are similar parametric test may be used
significance levels
significant result is 1 that is unlikely to be due to chance factors statistical tests use significant level – point at which researcher can reject null hypothesis + accept alternative hypothesis
significance level measures amount of chance factors that are permitted in research
It is chosen BEFORE research is carried out + is expressed as decimal
Psychologists have concluded that for most purposes in psychology 5% level of significance is appropriate
written as: P< 0.05. The 0.05 significance level means that probability of results of study occurring by chance is less than 5% We can therefore be 95% confident that IV caused change in DV
95%
occasions when it is necessary to use very strict level of significance eg. when testing new drug
researchers will occasionally allow for more chance factors by choosing less stringent level of significance
choosing a statistical test
test of difference
unrelated design related design
O=Mann-whitney wilcoxon
I=unrelated t-test related t-test
N=chi-squared sign test
test of association/correlation
N=chi-squared
O= spearman rho
I=pearsons r
Type I + Type II Errors
Sometimes researchers choose wrong hypothesis-mistakes are known as type 1 or type 2 errors
errors are more likely to occur when 0.1 or 0.01 significance level is used 0.05 is preferred significance level in psychology
Using level of significance which is too lenient eg. p<0.1 may lead to type 1 error where null hypothesis is rejected when it should in fact be retained as results are due to chance
Likewise using level of significance which is too strict eg. p<0.01 may lead to type 2 error where null hypothesis is retained when it should have been rejected
main reason for using 5% level in psychology is that it is neither too strict nor too lenient preventing type 1 + type 2 errors
using statistical data
statistical test has been calculated result is no.- calculated value check for statistical significance calculated value must be compared w/critical value– numerical cut-off point that tells us whether we can reject null hypothesis + accept alternative hypothesis
Critical value – no. created by statisticians
Calculated value – no. obtained from results of stats test
3 criterias:
- no. of Ps- usually appears as N value on table some tests use degrees of freedom instead
- One or two-tailed test? -one-tailed test if your hypothesis was directional + two-tailed test
for non-directional hypothesis - Significance level
discussed above 0.05 level is standard level used in psychology
calculated value is greater than critical table value in Chi squared test, related t-test, unrelated t-test, Pearson’s or Spearman’s Rho tests then null hypothesis can be rejected
calculated value is less than critical table value in Mann Whitney U Test, Wilcoxon Signed Ranks test or sign test then null hypothesis can be rejected
