Research methods – all three papers, particularly paper two
Research Methods Summary
2
Features of Science
Research should be:
Replicable - If repeated the results should be the same/similar
Variables should be operationalised – IVs, DVs, variables, generating a testable
hypothesis
The judgements should be as fair and unbiased as possible – possibly made by
independent researchers
It should be objective (fact based) rather than subjective (opinion based)
Scientific knowledge should lead to generating and refining theories
Example:
A teacher has worked in the same primary school for two years. While chatting to the
children, she is concerned to find that the majority of them come to school without having
eaten a healthy breakfast. In her opinion, children who eat 'a decent breakfast' learn to read
more quickly and are better behaved than children who do not.
She now wants to set up a pre-school breakfast club for the children so that they can all
have this beneficial start to the day. The local authority is not willing to spend money on this
project purely on the basis of the teacher's opinion and insists on having scientific evidence
for the claimed benefits of eating a healthy breakfast.
Why isn’t it scientific?
The teacherhas onlyexperiencedone school inaparticularcatchment areaso she has only
observedaverylimitednumberof 5 year-olds(issuesof samplingand replicability).
She has foundoutthat childrendonoteat anythingnourishingsimplybychattingwith the
children.She hasnocorroborative evidence from e.g. parents(issuesof objectivity).
She usesvague phrasessuchas 'decentbreakfast'withoutbeingclearwhatthis means(no
operationalisation).
She has generatedatheoryandmade predictionsbasedonflimsyevidence.
She has not usedanyscientificmethod toleadtoher conclusions e.g.acarefully controlled
experiment,surveyorobservation.
She has drawnconclusionsaboutthe effectsof breakfastwithoutconsideringother
variableswhichmightaffectreadingskillsandbehaviour.
Research methods – all three papers, particularly paper two
Peer Review
Peer review is very important throughout the whole of the scientific community and it has two
major functions:
1. Researchersgettoread otherpeople’sstudiesandkeepintouchwithnew waysof thinking
and scientificdevelopments.Theymayalsobe workingonthe same topicor thinkthey
couldimprove uponorevendisprove someone else’stheory.Knowledge growsthroughthe
sharingof information.
2. Studiessubmittedforpublicationare subjectedtocritical appraisal,whichactsasa check to
ensure poorqualityresearchdoesnotenterthe publicdomain.
Most researchersaimto publishtheirfindingsinprestigiousscientificjournalsandthere isan
agreedformatfor the wayin whichworkispresented.
Havingyour studypublishedmeans thatthe scientificcommunityhasgivenitaseal of approval;
thisisveryimportantfor universitydepartmentsastheyare assessedforfuture government
fundingonthe qualityof theirpublishedresearch.
Unlike textbooks,journalsare publishedperiodicallyandbuildintoyearlyvolumesthatserve asa
permanentrecordof research.Universitieskeeptheseintheirlibrariesandalsosubscribe toonline
publications.Somejournalsacceptresearchfromvariousareas,e.g., NatureorScience. In
psychology,the British Journalof Psychology publishesstudiesfrommanydifferentfieldswhereas
some journalsare extremelyspecialised,e.g., Personality and IndividualDifferences.
The systemof peerreview isheldinhighesteemandbeginswhena
research papersubmittedtoa journal isconsideredtobe worthyof
publication.The editorsendsthistootherexperts(whoare generally
unpaid) inthe fieldwhocriticallyappraise all aspectsof the study
thenreturnit withtheirrecommendationsastowhether the workis
of acceptable quality.If not,researchersrevise theirworkandre-
submittheirpaper.Thisensuresthathighstandardsare maintained.
However,peerreview isnotinfallible andthe systemsometimes
breaksdown.Provencasesof fraudare a rarity,but include plagiarism, falsificationof dataand
fabricationof data (UKParliamentaryOfficeof Science andTechnology,2002). Researchmaybe
supressedbecause itdisagreeswithaprevailingtheoryorbecause there isrivalrybetweentwo
researchgroups.There isa bias towardspublishingpaperswithpositive (significant) results.There
may notbe an expertforsome topics, particularlywhentheyare new areasof research. Peerreview
doesnotalwaysspot the mistakes.
Example:
AndrewWakefieldpublishedastudysayingthatthe MMR vaccinationwaslinkedwiththe
developmentof autism. Itturnedoutthat he was wrong – hismethodswere faulty,and the
conclusions were basedonunscientificwork.There wasanundeclared conflictof interest ashe was
partlyfundedbyan anti-vaccinationlawsuit.
He has beenstruckoff the medical council, andcanno longerbe a doctor. Howeverthe rumours
aboutMMR persist,andmanyparentschose notto vaccinate theirchildren.The rate of measles
infectionshasshotup,andit can be fatal inyoungchildren. Poorscientificpractice canbe
dangerous.
4
Psychology and the Economy
The most obvious application is in mental health. The cost of mental treatment in the UK is
approximately £22 billion annually.
A current focus of research is in Dementia - as the population rises there will be more and more
cases. How should they best be treated? Can anything be done to avoid it?
Research weighs up the benefits and costs of therapies – for example drugs are cheap, but is
psychological therapy better even though it is much more expensive?
Other Links:
Daniel Kahnemanhasledresearchonirrational thinking,how people behave in their financial lives
and why.
Behavioural economicsisprimarilyconcernedwiththe rationalityof decisionsrelatingtoeconomics.
AVAILABILITY HEURISTIC: the rule that the likelihood of selecting something is linked to its
availability.
For example: we tend to overestimate the chances of being involved in a disaster of some sort – a
plane crash.Actuallythe chancesare tiny,butbecause we all hear about them when they happen it
makes it seem more likely than it actually is. If you asked the public what the chances were they
would overestimate the probability.
Framing
The way information is presented has an effect on whether it is successful or not. This is known as
‘framing’, and it can be positive or negative.
Understandingthe biasescausedbyirrational thinking has changed business practice. Layard (2014)
said that Kahnemen’s work had revolutionised business because it has been applied to so many
areas:
Decision making in juries
Treatment of mental health problems
Financial advice
Government programmes such as healthy eating campaigns
Research methods – all three papers, particularly paper two
MEMORY
How could research into Eye Witness testimony have implications for the economy?
Improving EWT can lead to more accurate EW’s and therefore money saved on police resources,
because they will be looking for the correct person/people.
What about the development of the cognitive interview?
If the cognitive interview is more effective than the standard police interview then ti me and
therefore money could be saved by doing more effective interviews.
SOCIAL INFLUENCE
How could understanding minority influence lead to implications for the economy?
Minoritiescanadvocate change that will benefitthe economy. Byunderstandinghow aminoritycan
bring about change minorities can be encouraged to make changes that will save money. E.g.
EnvironmentalistsandRecyclingsavedmoneybyrecyclinggoodsratherthanbuyingeverything new
(It also saves finite resources in the Earth)
PSYCHOPATHOLOGY
If treatments for mental illness are found to be more effective than other treatments. How could
more effective treatments impact on the economy?
If the treatmentismore effective thenitcouldleadto money being saved by the NHS not having to
pay out for treatments that aren’t as effective. So the person could be treated quicker, saving
money on further treatments, and more people can be treated successfully in the long term.
Why would it be beneficial to the economy if treatments are so effective people could return to
work?
If people canreturnto work because treatmentshave beeneffective thenitmeansthat they will be
contributingtothe economy(e.g.payingtaxes/national health) rather than being unable to work &
claiming from the government.
Eventhoughdrug treatments might be as effective as CBT, why might it be more beneficial for the
economy if people have CBT instead?
Because the drug treatmentsare sometimes accused of being a quick and short term fix. By having
CBT, it targetsthe root cause rather thanjust masking symptoms. Therefore it could mean that the
treatment might have a more long term effect – saving money on not needing further drug
treatments.
6
ATTACHMENT
Bowlby’sview that children need their mothers led to women feeling like they needed to “stay at
home”. What effect could this have on the economy?
If mothers are staying at home then they aren’t working and contributing to taxes etc.
Later researchshowedthatactuallychildrenare generallyfine aslongastheyhave a good substitute
carer (e.g good day care or their father). What could this mean for the economy?
It couldmeanthat mothersare able to workand therefore theyare adding to the economy through
taxes. Also if they are working it aids the organisations/businesses through having the workforce.
How could research into the harmful effects of deprivation/privation have implications for the
economy?
Childrenthathave secure childhoodattachments go on to become productive members of society,
so holddownjobsandcontribute to the economy. Whereas children with deprivation or privation
may find it difficult in later life to be a productive member of society. They may also cost more to
the government to help and support in later life.
BIOPSYCHOLOGY
How could research into “smart machines” – i.e machines that think like humans - affect the
economy?
Because it could lead to saving money for businesses and organisations by having machines doing
jobs to save money – e.g recognising faces on CCTV etc.
Psychology and the economy may also come up on the paper as an application question – the
company want to sell something, how could they frame it positively?
How could they persuade patients to take their medication?
Woulditsave the Police time andmoneyoninvestigationsif Cognitive Interviews were much more
accurate?
Have some examples ready to go – it’s possible that this could be a long question (8 marks for
example)
Research methods – all three papers, particularly paper two
Methods for Conducting Studies
What are the Advantages/disadvantagesof a Lab study?
The most likelyadvantagesof the laboratorysettinginclude:
Control overextraneousvariables.The labsettingmeantthatextraneous variablescouldbe
minimised.
Replicability.
It can be verycontrolled,andaffectthe behaviourof the participants.
It can lack ecological validity
A common mistake isin the exam isgiving general criticismsand not specificto the
example inthe question.Youmust relate it to the scenariothey have givenyou.
What are the advantages/disadvantagesof a fieldstudy?
The environmentismore natural whichshouldproduce more realisticbehaviour
There isless control of extraneousvariables
The PPs maynot knowtheyare beingwatchedinwhichcase they cannotgive informedconsent.
If theydon’tknowthat theyare beingstudiedthenitmayreduce experimentereffectsanddemand
characteristics.
What are the advantages/disadvantagesof Natural and Quasi Studies?
The experimenterdoesnotmanipulate the IV,theyobserve anexisting one –natural means
somethinghappenedoutof the ordinary,butwasnothingtodo withthe researchere.g.accident.
Quasi – somethingthatcan’tbe controlledisthe IV, mostcommonisage.
Positive:Behaviourshouldbe natural
Negative:Causal conclusionscannotbe drawn
No randomallocation
But may be the onlywayto ethicallystudysomething.
8
Othermethods
Self-ReportMethods:
Questionnaires
Structured, semi-structured,unstructuredinterviews
Can be openor closedquestions
The main problemishonesty –social desirabilitybiasmeansthatparticipantsmaytryto
portray themselvesinthe bestpossiblelight.
Observational Studies:
Participantsare observedtonote behaviour– can be overt(whentheyare aware of being
watched) orcovert (notaware of beingwatched).
Will use eithertime sampling –e.g.whatare theydoingevery15 seconds
Or eventsampling –recordingeverytime theydosomethinginparticular
Case Study:
A detailedstudyof one individual oraunique situation
Usuallylongitudinal
Providesrichdetail andmassesof data
But can’t be generalisedtoothersbecause of the verysmall sample orunusual situation
Correlational Studies:
The relationshipbetweentwovariables
Easilyshowsa link,andcan be replicated
But does notshowa cause and effect relationship asthere maybe an interveningvariable
Research methods – all three papers, particularly paper two
Reliability
Reliabilityreferstothe consistencyof a measurement
How can Reliabilitybe checked?
Example:Reliabilitycouldhave beencheckedbyadministeringavalidandreliable questionnaire to
the participantsaswell asinterviewingthemandthencomparingthe scoresonthe twomeasures.If
the interview score was reliable,there wouldbe strongpositivecorrelationbetweenthe scores.
Inter-Raterreliability– comparingtwo or more researcherswhoare watching/scoring the same
thing– there shouldbe astrong positive correlation –0.8 or similar.
Example:The interviewscouldhave beenfilmedandgiventoanothertrainedtherapisttoassess.A
strongcorrelationbetweenthe scoresgivenbyeachtherapist woulddemonstratereliability.
Split–Half – on a mathstest(or similar) the questionsare divided,andthe answerscompared.E.g.
One participantwill have theiranswerscomparedforthe oddnumbersagainstthe evennumbers,
and the resultshouldbe the same/similar.
Test-Retest– the experimentisrunagainand the resultscompared.Theyshouldgetsimilarresults
inthe replications.Theycaneithertestthe same people againafterasuitable interval of time,test
differentpeopleusingthe same test,ortestthe same people withmore thanone test,eachof
whichhas the same standardiseddifficulty.
Why isReplicabilityimportant?
Replicabilityisanimportantpartof the scientificprocess.Scientificmethodinvolvesdefininga
problemandformulatingahypothesiswhichistestedwithempirical research.
Researchfindingsare animportantpartof thisprocess.If we wishto draw conclusionsfrom
researchstudies,the proceduresand findingsshouldbe repeatable.
Unrepeatable resultsmayimplyflawsor alack of control withinthe methodusedandare of limited
use intheoryconstruction.
10
Validity
Validityis about whethera testmeasureswhat it is meant to measure.
Internal validityiswhat happenswithinthe study –isthe researchtestingwhatitis meantto?
External validity- can the resultsbe generalisedtootherpeopleandothersettings?
The ideais to try andmake sure thatextraneousvariablesdon’taffectthe result.The change inthe
DV shouldbe asa resultof the changesinthe IV,andnothingelse.The control availableinthe
laboratoryishelpful,butcanmake somethingartificial.Fieldstudiesare usuallybetter,butnot
alwaysso,as theycan alsobe verycontrivedandartificial.
Validityislinkedtoreliability,because if astudyisnotreliable thenitisn’tvalid.Considerthe
followingexamples:
If a participanttakes anintelligencetestonseveral occasionsandthe resultschange eachtime then
the testis not reliableandthe resultsare meaningless,solackvalidity.If apersontakesanIQ test
several timesandgetsthe same score itis a reliable test,butmaytestwhatthe personhas learntat
school rather thantheirintelligence,inwhichcase itlacksvalidityasitisn’ttestingwhatitismeant
to.
Temporal validity– the abilitytogeneralise aresearcheffectbeyondaparticulartime period.
Ecological validity– the abilitytogeneralise aresearcheffectbeyond the particulartime periodof
the study.
Validitycanbe affectedbymanydifferentthings,dependingonthe study.Commonissuesinclude:
Sample – all one type of person,usuallystudents,is notgood
All one genderalsoa problem
If the testusesindividual groupsthere maybe differencesbetweenthe groups
Lab experimentscanbe lowinexternal validity,butthisisnotautomaticallytrue – it
dependsonthe task
Mundane realismcanbe a problemif the taskisveryartificial,andthisreducesthe
generalizabilityof findings
Observationscanhave problemswithobserverbias,affectinginternal validity
How can validitybe checked?
Face validity– doesitlooklike itistestingwhatitis meantto onthe surface?
Concurrent validity– checkingitagainst a pre-existingandpreviouslyvalidated
measure
Research methods – all three papers, particularly paper two
Ethics
• Good psychological researchispossible onlyif there ismutual respectandconfidence
betweeninvestigatorsandparticipants.
• Ethical guidelinesare necessarytoclarifythe conditionsunderwhichpsychological research
isacceptable.The BritishPsychological Societypublishesethicsguideswhichare updated
regularly.
• A fundamental ethical questioninpsychological researchis `Doesthe end justifythe
means?`.A balance mustbe struckbetweenthe interestsof the participantsandthe value
of research.
• Ethicsis that whichisdeemedacceptableinhumanbehaviourinpursuitof certaingoalsor
aims.It isnot simplyaquestion of right,butof balance betweenthe interestsof the
participantandthe scientificvalue of the research.
• Basicallyethicsisall aboutwhatyoucan andcan’t do withparticipants.
CONSENT
• Participantsshouldgive informedconsent.
• In studies involvingchildren,informedparental consentshouldbe obtained.
• Paymentshouldnotbe usedtoinduce risktakingbehaviour.
• Special safeguardingproceduresare necessarywithsubjectswithlimitationsin
communicationorunderstandingorwhere unable to give realconsent.
• DECEPTION
• Intentional deceptionoverthe purpose of the investigationshouldbe avoidedwhere
possible.
• There mustbe strongmedical orscientificjustificationforanydeception.
• Appropriate consultationwithdisinterestedcolleagues orethicscommitteesmustprecede
the investigationif itinvolvesdeception.
12
CONFIDENTIALITY
• The sourceof all information should remain confidential.
• Legislation,includingtheData Protection Act, should be adhered to.
• If confidentiality cannotbe guaranteed, the subjectshould be duly warned.
DEBRIEFING
• Participantsshould befully debriefed.
• Participantsexperienceshould be discussed to assessany negativeeffects.
• Debriefing should be in the form of active intervention before leavingthe research setting.
• Participantsshould betold what happened in other conditions
• They should have the opportunity to ask questions
WITHDRAWAL FROM THE INVESTIGATION
• Subjects should be aware of the right to withdraw at any time.
• This may be done retrospectively by refusingpermission for their data to be used.
PROTECTION OF PARTICIPANTS
• Participantsshould beprotected from physical or mental harm.
• Participantsshould beasked of any factors which may create risk;i.e. medical conditions.
• Any risk should beno more than could be expected in the courseof normal daily lifestyle.
• There should be the opportunity to contact the investigator ata later date if distress does occur.
OBSERVATIONAL RESEARCH
• Observation should beonly carried outin places where people could expect to be observed by
strangers,unless specificpermission isobtained beforehand.
• Care should be taken about the invasion of privacy.
GIVING ADVICE.
• If an investigator comes across a problemthat the subjectmay be unaware of, he/she has a duty to
informthe subject,if they believe failureto do so will causeharmor distress.
• The investigator should refrain fromgivingadviceif they are not competent to do so and should
recommend suitablecourses of action.
• Example – if an fMRI reveals a medical problem the participantshould beadvised to take further
action by seeing a specialist.
Research methods – all three papers, particularly paper two
Hypotheses
Studieshave twohypotheses.
The alternative/experimentalhypothesisisthe one thatsayssomethingwill happen
The null hypothesissaysthatthe IV will have noeffectonthe DV – nothingwill happen
A directional hypothesisisknownas one-tailed
A non-directional hypothesisisknownas two-tailed
The experimental/alternativehypothesisisastatementof the relationshipbetweentwovariables,
whereasthe null hypothesisisastatementof norelationshipbetweentwovariables.
For example,the hypothesismightsay‘There isadifference betweenthe scoresona memorytest
fromGroup A whouseda mnemonicandGroup B whodidnot’
whereasthe null hypothesiswouldbe ‘There is nodifference betweenthe scoresfromamemory
testfrom GroupA whouseda mnemonicandGroupB who didnot’.
The hypothesismustcontainanumberof thingsto gainfull marks.
Hypotheses have tobe operationalised.Thismeans thathow the DV will be measured
shouldbe part of the hypothesis.
It shouldcontainthe rightword – correlation,association,difference
Say whatwe are measuringandhow we are measuringit,aswell aswhat we are tryingto
find
Commonmistakesincludegivingthe wrongtype of hypothesis –null insteadof non-directional for
example. The othercommonmistake isusingcorrelationwhenit’satestof difference orvice versa.
If it is a correlational testandyouwrite ahypothesisthatsaysdifference, youwill endupwithno
marks.
Examplesof hypotheses:
There isan associationbetweenbirthorderandchoice of career
There is a correlationbetweenpupils’scoresonatest of mathematical abilityandpupils’scoresona
testof musical ability.
Thissays whatwe are measuringand how we are measuringit,aswell as what we are tryingto find
– e.g. correlation.
14
Statistical tests – Basics
In everyexperimentthere are twohypotheses,andatthe endof a piece of researchwe mustaccept
one of them. There isan alternative/experimental hypothesisthatsayssomethingwill happen,anda
null hypothesisthatsaysnothingwill happen - there will be norelationshipbetweenthe IV andthe
DV .
Statistical testsare usedtodecide whetherthe experimental/alternativehypothesisisacceptedor
not. It mightlooklike itssignificant,butitneedstobe proven. Beingsignificantmeansthatthe
difference betweenthe two(ormore) conditionsisenoughtobe sure thatit didn’t justoccurby
chance.
The testsusedcalculate the probabilityof somethingoccurringbychance – inorder to acceptthe
hypothesiswe needtobe sure that the change in the DV was a resultof a change inthe IV.
We alwaysuse a significancelevel of 0.05 – thismeansthat we are 95% sure that the resultsdidnot
occur by chance. There isstill a 5% chance the resultsoccurrednaturally,andwere notrelatedto
the change in the IV.
This(0.05) isnot the value thatwe are comparinganythingto,nor isit the result,orthe
critical value,orthe calculated value. Itisthe level of significance. Youshouldnotbe
comparinganythingto0.05. It tellsyouwhere tolookon a table of critical values.
The calculated/observedvalue isthe endresultof the study – once all the maths is done we are left
withone figure – the observedvalue,whichwe thencompare tothe table of critical values.
The critical value isthe value inthe table that we are comparingresultsto.
To selectthe rightcritical value inthe table youneedtoknow the following:
Degreesof freedom –inmost casesthisisthe numberof participants(chi-square isan
exception)
Whetheryouhave a one-tailed(directional)ortwo-tailed(non-directional) test–thisis
basedon the hypothesis
The significance level–usually 0.05
Whetherthe calculated/observedvalue needstobe higherorlowerthanthe critical value
(lookunderthe table of critical values)
Whetheritisparametric– parametrictestsuse interval data
Type one error: a false positive, or error of optimists – most likely when the level of
significance level is too lenient (10%), and leads to the results being judged as significant
when they aren’t, and the experimental hypothesis is accepted incorrectly.
Type two error: a false negative, most likely when the level of significance is too stringent
(1%), leads to the experimental hypothesis being rejected, when it should be accepted.
Research methods – all three papers, particularly paper two
Levels of measurement anddata types
Primary data – collectedbythe researcherforthe purposesof theirstudy.Thissuitsthatparticular
studybut istime consumingandexpensive.
Secondary data – usingdata collectedforapreviouspiece of work,e.g.government statistics.This
savestime andmoneybutmay not fitthe currentstudyperfectly.
Nominal data: If you are placingdata ina tallychart,thiswouldbe nominal data.If you are counting
the numberof timessomethinghashappened,thisisnominal datae.g.numberof males/females
stoppingfora pedestrianata zebracrossing
Ordinal data: Use of self-reportquestionnairewitharatingscale wouldbe ordinal data.The likert
scale isthe mostcommonlyusedscale,where Ppsare givenastatementandasked tostate level of
agreement,form‘agree verymuch’to‘disagree verymuch’,orratingsomethingoutof ten.
Interval data: This iswhere datais inthe form of equal unitse.g.score onan objective test.Insucha
case,if one P has scored20 on a test and anotherhasonlyscored10, we wouldbe correct insaying
the firstpersonhas performedtwice aswell asthe second. Italsoincludesdatawithafixedscale,
e.g.centimetres,seconds,correctanswers.
**************************************************************************************************************************
You also need to know what experimental design is:
Independent groups – each participant only does one condition
Avoids practice effects, you could use the same set of stimuli BUT you need more participants and there
can be differences between the groups
Repeated measures – each participant does all the conditions – must be counterbalanced
Avoids individual differences e.g. in reaction times, less participants needed BUT more chance of demand
characteristics and practice effects
Matched pairs – the participants are matched on key characteristics and their results compared. Counts as
repeated/related data for the purposes of statistical tests.
Avoids the practice effects with repeated measures and the individual differences of independent groups
BUT takes ages to do, needs a large pool of potential participants and may not account for variables that
affect the results
16
Statistical tests: Spearman’s Rho
Hypothesis states a correlation between two co-variables
The two sets of data are related
Data is ordinal or interval
1. State the hypothesis: the digit ratio between index finger and ring finger is positively
correlated to numeracy skills
Null: there is no correlation between digit ratio and numeracy skills
2. Record the data, rank each co-variable, and calculate the difference
3. Find the calculated value – for spearman it is called the correlation co-efficient or rho
4. Find the critical value to compare the observed value to
In this case the hypothesis is directional, therefore you are looking for a one-tailed test
Use the number of participants down the side of the table, and look for 0.05 and state the
conclusion
Always go with 0.05
Number of
participants
To be significant the
observed value must be
higher than the critical
value for this test
Research methods – all three papers, particularly paper two
Statistical tests: Pearson’s r (parametric)
Hypothesisstatesacorrelationbetweentwosetsof data
The two setsof data are related
The data are interval
The requirementsforparametrictestsare satisfied(the populationsare assumedtohave a
normal distributionandthe variancesof the sample are assumedtobe the same – don’t
worryabout thisbit)
1. State the hypothesis
2. Place the data ina table
3. Findthe calculatedvalue of r – the formulaiscomplex andwon’tbe inthe exam, sotheywill
give youthe calculatedvalue of r.
4. Findthe critical value bylookinginthe table of critical values.Rememberthat the degrees
of freedomis participants -2.
5. Is the resultsignificantandisinthe rightdirection?If the predictionwasapositive correlationdid
youget a positive?If the hypothesisisnondirectional thenitdoesn’tmatterwhichwaythe
correlationgoes.
6. Report the conclusion.
As the calculated value is…puttheactualfigure……andthe critical value is …. Putthe actualfigure
We mustaccept/rejectthe… experimental/alternative/nullhypothesis
Therefore,we canconclude thatthe resultis significant/notsignificant wheredf =…………………………..
(degreesof freedom) andp<0.05 for a one-tailed/two-tailed test
Research methods – all three papers, particularly paper two
Statistical Tests: Chi-Square
Deals with nominal data
Used when we’ve counted how many things there are in each category (frequencies)
Used when hypothesis predicts an association or difference between co-variables
Sets of data must be independent, so no individual should have a score in more than one cell
Step 1: Hypotheses
Alternative hypothesis: there is an association between gender and the amount of hours slept
Null hypothesis: there is no association between boys and girls and the amount of hours slept
Step 2: Contingency Table – draw a table to record the frequencies in each group
Step3: Findthe critical value,sothatyou can compare it to the calculated value
Calculate the degreesof freedom(df) by:
Multiplying(rows-1) x (columns-1)
The calculated value (X2) of chi-squareis 1.984 – this isthe value of the data from the experiment
Look at the critical value fromthe Table of critical values- we alwayslookat the column p≤0.05
Reminder:THIS(0.05) ISNOT THE CRITICALVALUE, OR THE CALCULATED VALUE, IT IS THE LEVEL OF
SIGNIFICANCE.
Step4: State yourconclusion
Nondirectional hypothesis,sotwo-tailedtest.
As the calculated value is…puttheactualfigure……andthe critical value is …. Putthe actualfigure
We mustaccept/rejectthe… alternative/nullhypothesis
Therefore,we canconclude that the resultis significant/notsignificant wheredf =……………………
(degreesof freedom) andp<0.05 for a one-tailed/two-tailed test
20
Statistical Tests: Sign test
Data goesintoa table - happinessratingsbefore andafteraholiday:
participant
Happiness
score
before
Happiness
score
after
Difference
between
the two
sign
1 6 7 1 +
2 3 4 1 +
3 4 6 2 +
4 8 6 -2 -
5 5 7 2 +
6 7 5 -2 -
7 5 7 2 +
8 5 8 3 +
9 4 7 3 +
10 8 5 -3 -
11 4 4 0
12 8 9 1 +
13 6 7 1 +
14 5 6 1 +
You will have tofindthe calculatedvalue yourself.
It isthe sumof the lessfrequentsign –inthiscase there are more plusesthanminuses,sothe total
of minussignswill be the calculatedvalue: s=3.
N=13 – there are 14 participantsbutany withnodifference (like participant11) are notcounted,
meaningthatwhenyoucome to lookfor a critical value youare lookingfor13 and not14.
The critical value for13 is3, equal to the calculatedvalue,sothe resultinthiscase wouldbe
significant.
Conclusion:
The calculatedvalue of S (3) is equal tothe critical value of S (3) where N=13, p<0.05 and a one-
tailedtestisusedsothe resultsare significant. The null hypothesiscanbe rejected andthe
experimental hypothesis accepted.
Where there is no difference
the result is excluded, reducing
the N value
Research methods – all three papers, particularly paper two
Statistical Tests: Mann-Whitney U
The hypothesisstatesadifference betweentwosetsof data
The two setsof data are from separate groups – independentgroups
The data are ordinal or interval
1. State the hypotheses:
Alternative:Male participants interviewed onahighbridge give higherratingsof the attractiveness
of a female interviewerthanthose interviewedonalow bridge
Null:there isnodifference inthe ratingsof attractivenessgivenbythose interviewedonahighor
lowbridge.
2. Record the data ina table andallocate points
Each score iscomparedto all the othersin the othergroup,and givenpoints
3. Findthe calculated value of U
It’sthe lowerpointsvalue wheneachcolumn isaddedtogether
4. Compare it to the critical value
Numberof participantsineachgroup
showsyouwhat the critical value
It’sa directional hypothesis,soa
one-tailedtest,andhasto be
lowerthanthe critical values
Step4: State yourconclusion
As the calculated value is…puttheactualfigure……andthe critical value is …. Putthe actualfigure
We mustaccept/rejectthe… alternative/nullhypothesis
Therefore,we canconclude that the resultis significant/notsignificant whereN=….. (numberof
participants) andp<0.05 for a one-tailed/two-tailed test
5 wouldbe the critical value for
6 participants in one group, six
in the other
22
Statistical Tests: Wilcoxon T
Hypothesisstatesadifference betweentwosetsof data
The two setsof data are from the same person – repeated measures
The data are ordinal
1. State the hypotheses:
Alternative:participantswillrate the more frequentlyseenface asmore likeable thanthe less
frequentlyseenface.
Null:there isnodifference inthe likeabilityscore forfacesseenmore orlessoften.
2. Record the data, calculate the difference betweenscoresandrank
For eachscore the differenceisworkedout,andthenall the scoresare rankedin order,andthe
difference betweenthe tworanksworkedout
3. Findthe observedvalueof T
The sums of the lessfrequentsignsare addedtogether
4. Compare T to the critical value,usingthe numberof participantsforthe row and0.05
For thistestthe observedvalue Thas to be lowerthanthe critical value.
5. State the conclusion
As the observedvalue is…puttheactualfigure……andthe critical value is …. Put the actualfigure
We mustaccept/rejectthe… alternative/nullhypothesis
Therefore,we canconclude that the resultissignificant/notsignificantwhereN=….. (numberof
participants) andp<0.05 fora one-tailed/two-tailed test
24
Statistical tests: T test, related
(parametric)
1. State the hypothesis –itshouldsaythat there will be adifference
2. Place the rawdata in a table – two for eachparticipantforrelatedttests
3. Findthe difference betweeneachpairof scoresandput that figure ina new table
4. Calculate the total of the differences –thisfigure isknownas Ƹd
5. Add anothercolumn forthe squares of the differences –d2
6. Calculate the total of the squaresof the differences
7. Findthe calculatedvalue –itis likelythattheywillgive youthe calculatedvalue
8. Compare the calculatedvalue of t to the table of critical values
9. Report the conclusion:
Example:Asthe calculatedvalue (t=1.8) islessthanthe critical value (t=1.812) it isnot significant
where p<0.05, df = 10, usinga one-tailedtest. Sowe mustacceptthe null hypothesisandconclude
there isno difference between……….
Rememberto check that the difference goesthe right way if it is a directional hypothesis
Research methods – all three papers, particularly paper two
Statistical tests: T test, unrelated, (parametric)
1. State the hypothesis –itshouldsaythat there will be adifference
2. Place the rawdata in a table – one foreach participantforunrelatedttests
3. Findthe difference betweeneachpairof scoresandput that figure ina new table
4. Calculate the total of the differences –thisfigure isknownas Ƹd
5. Add anothercolumn forthe squares of the differences –d2
6. Calculate the total of the squaresof the differences
7. Findthe calculatedvalue –itis likelythattheywillgive youthe calculatedvalue
8. Compare the calculatedvalue of t to the table of critical values
9. Report the conclusion:
Example:Asthe calculatedvalue (t=1.921) is greaterthanthe critical value (t=1.812) the resultis
significantwherep<0.05,df = 10, usinga one-tailedtest.Sowe mustacceptthe alternative
hypothesisandconclude there isadifference between……….
Remember to check that the difference goes the right way if it is a directional hypothesis
26
QualitativeAnalysis
Non-numerical datacannotbe analysedwithstatistics andsoothermethodsneedto be used.This
will involve some sortof analysisof writtenwords(interview transcriptsforexample),pictures,TV
programsor films –anythingthatdoesnot lenditself tobeingmeasuredinnumbers.
Contentanalysisisaway of analysingnon-numericaldatainwhichresponses/dataare grouped
togetherincategories.Itispossible toturnitintoquantitative databycountingthe occurrencesin
each group.
Stepsforcontentanalysis:
Decide on a sample,for example advertsabout cleaningproducts
Code the data usingcategories – genderof main character, action they are observeddoing
Watch the adverts recordingeach occurrence ofa particular behaviour
Thematicanalysisisaboutlookingforthemesindataandkeepsthe resultsdescriptiveratherthan
countingoccurrences.
Stepsforthematicanalysis:
Watch and re-watch the video/readand re-read the transcripts
Break the data down into meaningful units
Assigna label or code to each unit
Combine codesinto larger themes
Ensure all the data is representedbythe larger themes – there shouldn’tbe responsesthat
don’t fit
Example:
Qs 20 and 21 are based on a stem about eating behaviour, and the impact of having
restricted food and water and set meal times. Participants were asked to draw pictures
food.
20 What ismeantby the term contentanalysis?
(1 mark)
Anin-depthanalysisofqualitativedata,conductedbylookingforthemesorcategories.
21 Explainhowapsychologistmighthave usedcontentanalysistoanalyse these drawings.
(3 marks)
Sheshouldlookforcharacteristicsthatshewouldexpecttoseeinindividualswhowerebeingaffected
byfoodrestrictions.Eachpicturecouldbeassessedbylookingforcommonitems,forexampleportion
Research methods – all three papers, particularly paper two
sizes, whether it is fat or carbohydrate based and the colours used. If there are changes to the
categoriesorthemesoverthecourseofthestudythenthesecanberecognised.
28
Graphs
Quantitative data– bar graph or histogram
Correlations –scattergrams
Frequency polygons –same dataas histogramsbutbetterfor comparingtwoor more setsof data
Label the axis properly
and give the graph a
title
Research methods – all three papers, particularly paper two
Distribution
In large data setsthe distributionof scoresshouldresemble acurve:
Sometimesthe distributionwill be skewedbyafew extremescores
Positivelyskewedmeansthata few extremehighscoresaffectthe distribution
Negativelyskewedmeansthatafew extreme low scoresaffectthe distribution
The mean will be in the middle
Most scores will be less than two standard
deviation points away from the mean
30
Writing about a Psychological Study
Abstract
The purpose of thissectionisto provide abrief andcomprehensivesummaryof the study.Itisvery
importantbecause itisall that manypeople will read.Itshouldinclude abrief descriptionof the
problembeinginvestigated,the methodsused,the results,andtheirimplications.
It shouldbe accurate (do notinclude informationhere thatisnotinthe bodyof the manuscript),
self-contained(spell outabbreviations),concise (120wordmaximum),andspecific.Beginthis
sectionwiththe mostimportant informationandlimitittothe four or five mostimportantconcepts,
findings,orimplicationsof the study.
Introduction
The main purpose of thissectionistotell the readerwhyyouperformedthe study.Inotherwords,
youhave to informthe readerof the researchquestionandindicate whyitisimportant,andhow it
isunique whencomparedtopreviousstudies.
It starts outbroad and becomesmore andmore specific.Forexample,youmight beginbydefining
any relevantterms.Thengoonto review the relevantliterature.Thengoonto make clearthe
connectionbetweenpreviousresearchandthe presentwork.
Include astatementof the hypothesesandthe rationale forthem.
The final paragraphusuallycontainsastatementwhichclearlyandexplicitly stateswhythe study
was performed,suchasThe purpose of thisstudywas to...or The presentstudywasdesignedto
investigatethe...
Methods
The purpose of thissectionisto describe indetail how youperformedthe study.Someone shouldbe
able to replicate yourstudybasedonthe informationyouprovideinthissection.
Make itsoundprofessional,donotmake itsoundlike a classproject.Assume youare writingfor
submissiontoascientificjournal.
For an experiment,thissectionistypicallydividedintofoursubsections:subjects,apparatus,design,
and procedure.The orderof designfollowedbyprocedureisarbitrary.Inotherwords,youcould
have the procedure come before the design.Sometimesresearcherscombine the designand
procedure sections,however,inanexperimental psychologyorresearchmethodsclass,aseparate
designsectionistypicallyrequired.
Research methods – all three papers, particularly paper two
For a surveystudy(i.e.,one inwhichthe participantsare simplyaskedasetof questions),the design
sectionisnotnecessary(andthe surveyitself maybe includedasanappendix).
Subjects/Participants
Thissectionislabelledassubjectsorparticipantsdependingonwhetheranimalsorhumansare used
inthe study.If animalsare used,use the termsubjects.If humansare used, use the term
participants.
Indicate whoparticipatedinthe study,how many,andhow were theyselected.Withhuman
subjects,be sure toaddressthe issue of informedconsent.
Include anydetailswhichare relevanttothe study.For animals,includethe gender,age,strain,
weight.Forhumans,include the gender,age,race/ethnicity,and,whenappropriate,the
socioeconomicstatus,disabilitystatus,sexualorientation,etc.If the subjectswere human,what
type of rewardor motivationwasusedtoencourage themtoparticipate?
Design
Describe the designandclearlyspelloutthe independentanddependentvariables.Indicate what
the levelsof the independentvariableswere,andwhetherthe factor(s) wererepeated,matched,or
independent.Describe how the subjectswere assignedtogroups.Describe anycontrol procedures
used.
Procedure - the mostlikely sectionto write inan exam
Carefullysummarize each stepin the executionofthe study.Indicate what a typical test,trial, or
sessioninvolved.Describe anyphases that the study had or any instructionsthat the subjects
received.Whenreferringtogroups,tryto use descriptive labels.Forexample,insteadof saying
Group 1 or the experimental group,youmightsaythe druggedgroup.Anothertechniqueinthis
regardis to use abbreviationsthatemphasize meaning.Forexample,there were three groups,
includingthe control groupwhichreceived0mg/kgof morphine (M0),a low dose groupreceiving1
mg/kgof morphine (M1),anda highdose group receiving4mg/kgof morphine (M4).
Results
Look carefullyatthe results.Thatis,take a good hard lookat all those numbersyoucollect.Thinkof
differentwaystosummarize them,aswell astomake sense of them. The resultssectionwill include
descriptive andinferential statisticsinmanycases. There will usuallybe tablesof data(lesslikelyin
qualitative research) andgraphsaswell asstatementsaboutthe resultsof statistical tests –whether
theywere significant.
32
Designing Your Own Study
• An – Aim– what youare aimingtofindout /investigate
• Everso – Experimental andnull hypotheses(mustbe operationalised)
• Violent– Variables –IV/DV/Extraneous variablesandtheircontrol
• Person– Participants,who, where from, numberandsamplingmethod –how exactly.
• Doesn’t– Design– experimental andresearchmethodsdesign
• Do - Procedures – whatwill the participantsdo? Exactrecipe of whatyou will do. Include
howdata will be collectedandwhatdatawill be.
• Enough – Ethical issuesandhow youwill deal withthem
• Martial – Materialse.g.itemsandtestsandconsentforms
• Arts – Analysis –whichtestwouldyouuse and why?
Research methods – all three papers, particularly paper two
Reminder:Year One Research Methods Glossary
Attrition The lossof participantsfromastudyover time whichislikelytoleave abiasedsample ora
sample whichistoosmall.
Behaviour checklistA listof the behaviourstobe recordedduringanobservational study.
Behavioural categories Dividingatargetbehaviour(suchasattachmentor sociability) intoasubset
of behaviours.Thiscanbe done usingabehaviourchecklistoracodingsystem.
Case study A researchmethodthatinvolvesadetailedstudyof asingle individual,institutionor
event.Case studiesprovide arichrecordof humanexperience butare hardto generalise from.
Closedquestions Questionsthathave arange of answersfromwhichrespondentsselectone;
producesquantitativedata.Answersare easier toanalyse thanthose foropenquestions.
Codingsystem A systematicmethodforrecordingobservationsinwhichindividualbehavioursare
givena code forease of recording.
Cohort effects One groupof participants(cohort) mayhave unique characteristics becauseof time-
specificexperiencesduringtheirdevelopmentsuchasbeinga childduringthe SecondWorldWar.
Thiscan affectbothcross-sectionalstudies(because one groupisnotcomparable withanother) or
inlongitudinal studies(becausethe group studiedisnottypical).
Confederate Anindividual inastudywhoisnot a real participantand hasbeeninstructedhow to
behave bythe investigator/experimenter.
ConfidentialityA participant’srighttohave personal informationprotected.
Contentanalysis A kindof observational studyinwhichbehaviourisobservedindirectlyinwrittenor
verbal material suchasinterviews,conversations,books,diariesorTV programmes.Behaviouris
categorised(qualitative analysis) andmaybe counted(quantitative analysis).
Control conditionIn a repeatedmeasuresexperiment,the conditionthatprovidesabaseline
measure of behaviourwithoutthe experimentaltreatment(IV),sothatthe effectof the
experimental treatmentmaybe assessed.
Control group Inan independentgroupsexperiment,agroupof participantswhoreceive no
treatment.Theirbehaviouractsas a baseline againstwhichthe effectof the IV maybe measured.
Control referstothe extenttowhichany variable isheldconstantorregulatedbya researcher.
Do I really need to know
all these? It’s long . . . .
YES!
34
ControlledobservationA form of investigationinwhichbehaviourisobservedbutundercontrolled
conditions,incontrastwithanaturalisticobservation.
Correlation/ correlational analysis Determiningthe extentof arelationshipbetweentwovariables
CorrelationcoefficientA numberbetween –1and +1 that tellsushow closelythe co-variablesina
correlational analysisare related.
CounterbalancingAnexperimental technique usedtoovercome ordereffects.Counterbalancing
ensuresthateachconditionistestedfirstorsecondinequal amounts.
Covert observationsObservingpeople withouttheirknowledge,e.g.usingone-waymirrors;
knowingthatyourbehaviourisbeingobservedislikelytoalteryourbehaviour.
Cross-cultural study A kindof natural experimentinwhich the IV isdifferentcultural practicesand
the DV isa behavioursuchasattachment.Thisenablesresearcherstoinvestigate the effectsof
culture/socialisation.
Cross-sectional studyOne group of participantsof a youngage are comparedwithanother,older
groupof participants,withaviewtofindingoutthe influence of age onthe behaviourinquestion.
DeceptionWhere a participantisnottoldthe true aimsof a study(e.g.whatparticipationwill
involve) andthuscannot give trulyinformedconsent.
Demand characteristics A cue that makesparticipantsaware of whatthe researcherexpectstofind
or howparticipantsare expectedtobehave.Demandcharacteristicscanchange the outcome of a
studybecause participantswilloftenchange theirbehaviourtoconformtothe expectations.
Dependentvariable the resultof changesinthe IV (hopefully!),the thingthatismeasured
Difference studies Studiesinwhichtwogroupsof participantsare comparedintermsof a DV (such
as malesversusfemales,orextravertsversusintroverts).Thisisnota true experimentbecause the
apparentIV (genderorpersonality) hasnotbeenmanipulated.
Directional hypothesis statesthe directionof the predicteddifferencebetweentwoconditionsor
twogroups of participants.
Double blindNeitherthe participantnorthe experimenterare aware of the researchaimsand other
importantdetailsandthushave noexpectations.Ina single blindstudyonlythe participantis
unaware.
Ecological validity The degree towhicha researchfindingcanbe generalisedtoothersettings.
Effectsize A measure of the strengthof the relationshipbetweentwovariables.
Ethical committee (alsocalledinstitutional reviewboard,IRB) A groupof people within aresearch
institutionthatmustapprove astudybefore itbegins.
Ethical guidelinesare concrete,quasi-legal documentsthathelptoguide conductwithinpsychology
by establishingprinciplesforstandardpractice andcompetence.
Research methods – all three papers, particularly paper two
Event samplingAn observationaltechniqueinwhichacount iskeptof the numberof timesa certain
behaviour(event) occurs.
Experimental conditionIna repeatedmeasuresdesign,the conditioncontainingthe independent
variable.
Experimental designA setof proceduresusedtocontrol the influenceof factorssuchas participant
variablesinanexperiment.
Experimental group Inan independentgroupsdesign,agroupof participantswhoreceive the
experimental treatment(theIV).
Experimental realismThe extenttowhichparticipantsbecome involvedinanexperimentand
become lessinfluencedbycuesabouthow to behave.
External validityThe degree towhicha researchfindingcanbe generalisedto,forexample,other
settings(ecological validity),othergroupsof people (populationvalidity) andovertime (historical
validity).
FieldexperimentA controlledexperimentthatisconductedoutsidealaboratory.The IV isstill
manipulatedbythe experimenter,andtherefore causal relationshipscanbe demonstrated.Field
experimentstendtohave lowerinternal validity(more difficulttocontrol EVs) andhigherexternal
validity(greatermundane realism).Participantsare usuallyunawarethattheyare participatinginan
experiment,thusreducingparticipanteffects.
Hawthorne effectThe tendencyforparticipantstoaltertheirbehaviourmerelyasaresultof
knowingthattheyare beingobserved.
Historical validity The degree towhicha researchfindingcanbe generalisedtoovertime.
HypothesisA precise andtestable statement,whichincludeshow the variableswill be measured
Imposedetic A technique ortheoryisdevelopedinone culture andthenusedtostudythe
behaviourof people inadifferentculture whichhasdifferentnorms,values,experiencesetc.
Independentgroups Participantsare allocatedtotwo(ormore) groupsrepresentingdifferent
experimental conditions.Allocationisusuallydone usingrandomtechniques.
Independentvariable The thingthatis manipulatedbythe researcher,or changesnaturallyina
natural study
InformedconsentParticipantsmustbe givencomprehensiveinformationconcerningthe nature and
purpose of the researchand theirrole init,inorder that theycan make an informeddecisionabout
whethertoparticipate.
Inter-interviewerreliabilityThe extenttowhichtwointerviewersproduce the same outcome from
an interview.
36
Internal validityWhetherthe studyhastestedwhatitset outto test.;the degree towhichthe
observedeffectwasdue tothe experimental manipulationratherthanotherfactorssuchas
extraneousvariables.
Inter-observerreliabilityThe extenttowhichthere isagreementbetweentwoormore observers
involvedinobservationsof abehaviour.Thisismeasuredbycorrelatingthe observations of twoor
more observers.A general rule isthatif (total numberof agreements)/(total numberof
observations)>0.80, the data have highinter-observerreliability.
Interval Data are measuredusingunitsof equal intervals,suchaswhencountingcorrectanswersor
usingany‘public’unitof measurement.Manypsychological studiesuse plasticintervalscales where
the intervalsare arbitrarilydeterminedandtherefore we can’tactuallyknow forcertainthatthere
are equal intervalsbetweenthe numbers.However,forthe purposesof analysis,suchdatamaybe
acceptedas interval.
Interveningvariable A variable thatcomesbetweentwoothervariableswhichisusedtoexplainthe
relationshipbetweenthose twovariables.Forexample if apositive correlationisfoundbetweenice
cream salesandviolence thismaybe explainedbyaninterveningvariable –heat– whichcasesthe
increase inice creamsalesand the increase inviolence.
InterviewA researchmethodor technique thatinvolvesaface-to-face,‘real-time’interactionwith
anotherindividual andresultsinthe collectionof data.
InterviewerbiasThe effectof aninterviewer’sexpectations,communicatedunconsciously,ona
respondent’sbehaviour.
InvestigatoreffectAnythingthatthe investigator/experimenterdoeswhichhasaneffectona
participant’sperformance inastudyotherthanwhat was intended.Thisincludesdirecteffects(asa
consequence of the investigator/experimenterinteractingwiththe participant)andindirecteffects
(as a consequence of the investigatordesigningthe study).
Investigator/experimenterbias The effectthatan investigator/experimenter’sexpectationshason
the participantsandthus onthe resultsof a researchstudy.
Laboratory experimentAnexperimentcarriedoutina controlledsetting.Labexperimentstendto
highinternal validityandlowexternalvalidity thoughthisisn’talwaystrue.
Longitudinal study Observationof the same itemsoveralongperiodof time.Suchstudiesusually
aimto compare the same individualsatdifferentages,inwhichcase the IV isage.A longitudinal
studymightalsoobserve aschool or otherinstitutionoveralongperiodof time.
Matched pairs Pairsof participantsare matchedintermsof keyvariablessuchas age and IQ. One
memberof eachpair isplacedinthe experimentalgroup andthe othermemberinthe controlgroup.
Mean The arithmeticaverage of a groupof scores.Takesthe valuesof all the data intoaccount.
Measuresof central tendency A descriptive statisticthatprovidesinformationabouta‘typical’
response fora data set.
Research methods – all three papers, particularly paper two
Measuresof dispersionA descriptive statisticthatprovidesinformationabouthow spreadoutaset
of scoresare.
MedianThe middle valueinasetof scoreswhentheyare placedinrank order.
Meta-analysisA researcherlooksatthe findingsfromanumberof differentstudiesinordertoreach
a general conclusionaboutaparticularhypothesis.
Mode The mostfrequentlyoccurringscore inadata set.
Mundane realismRefersto howa studymirrorsthe real word.The simulatedtaskenvironmentis
realistictothe degree towhichexperiencesencounteredinthe environmentwilloccurinthe real
world.
Natural experimentA researchmethodinwhichthe experimentercannot manipulate the
independentvariable directly,butwhere itvariesnaturallyandthe effectcanbe observedona
dependentvariable.Strictlyspeaking,anexperimentinvolvesthe deliberate manipulationof anIV
by the experimenter,socausal conclusionscannotbe drawnfroma natural experiment.Inaddition,
participantsare not randomly allocated toconditionsinanatural experiment,whichmayreduce
validity.
Naturalistic observationA researchmethodcarriedoutina naturalisticsetting,inwhichthe
investigatordoesnotinterfereinanywaybut merelyobservesthe behaviour(s) inquestionthough
thisislikelytoinvolve the use of structuredobservations.
Negative correlationIn a correlation asone co-variable increases,the otherdecreases.
Nominal.The data are inseparate categories,suchasgroupingpeople accordingtotheirfavourite
football team(e.g.Liverpool,InvernessCaledonianThistle,etc.).
Non-directional hypothesis predictssimplythatthere will be adifference betweentwoconditions
or two groupsof participants,withoutstatingthe directionof the difference.
Observationtechniques The applicationof systematicmethodsof observationinanobservational
study,experimentorotherstudy.
Observerbias In observationalstudies,there isthe dangerthatobserversmight‘see’whatthey
expecttosee.Thisreducesthe validityof the observations.
OpenquestionsQuestionsthatinvite the respondentstoprovide theirownanswersratherthan
selectone of those provided.Tendtoproduce qualitativedata.
Operationalisingmakingsure thatthe resultsare measurable.
Opportunity sample A sample of participantsproducedbyselectingpeoplewhoare mosteasily
available atthe time of the study.
Order effectIna repeatedmeasuresdesign,anextraneousvariablearisingfromthe orderinwhich
conditionsare presented,e.g.apractice effectorfatigue effect.
38
Ordinal Data are orderedinsome way,forexample askingpeopletoputa listof football teamsin
orderof liking.Liverpool mightbe 1st
,followedbyInvernessandsoon.The ‘difference’between
each itemisnotthe same,i.e.the individualmaylike the 1st
itema lotmore than the 2nd
one but
there mightonlybe a small difference betweenthe itemsrankedas2nd
and 3rd
.
Participant effects A general termusedto acknowledgethe factthatparticipantsreactto cues inan
experimental situationandthatthis may affectthe validityof anyconclusionsdrawnfromthe
investigation.
Participant variables Characteristicsof individual participants(suchasage,intelligence,etc.) that
mightinfluencethe outcome of astudy.
Pilotstudy A small-scale trial of astudyrun to testanyaspectsof the design,withaview tomaking
improvements.
Populationvalidity The degree towhicha researchfindingcanbe generalisedtoothergroupsof
people.
Positive correlationIn a correlationco-variablesbothincreasetogether,
Presumptive consentA methodof dealingwithlackof informedconsentordeception,byaskinga
groupof people whoare similartothe participantswhethertheywouldagree totake partin a
study.If thisgroup of people consenttothe proceduresinthe proposedstudy,itispresumedthat
the real participantswouldagree aswell.
Privacy A person’srighttocontrol the flow of informationaboutthemselves.
Protectionfrom harm Duringa researchstudy,participantsshould notexperience negativephysical
or psychological effects,suchasphysical injury,loweredself-esteemorembarrassment.
Qualitative Data that expressesthe ‘quality’of things –descriptions,words,meanings,pictures,
textsandso on. Qualitativedatacannotbe countedor quantifiedbutitcanbe turnedinto
quantitative databyplacingthemincategories.
Quantitative data analysis Anymeansof representingtrendsfromnumerical data,suchasmeasures
of central tendency.
Quantitative Data that representhowmuchorhow long,or how many,etc.there are of something;
i.e.a behaviourismeasuredinnumbersorquantities.
Quasi-experimentsStudiesthatare ‘almost’experimentsbutlackone or more featuresof a true
experiment,suchasfull experimenter control overthe IV andrandomallocationof participantsto
conditions.Thismeansthattheycannotclaimto demonstrate causal relationships.
Questionnaire Data iscollectedthroughthe use of writtenquestions.
Random allocation Allocatingparticipants toexperimental groupsorconditionsusingrandom
techniques.
Research methods – all three papers, particularly paper two
Random sample A sample of participantsproducedbyusinga randomtechnique suchthatevery
memberof the targetpopulationbeingtestedhasanequal chance of beingselected.
Random techniquesAnytechnique inwhichthere isnosystematicattempttoinfluence the
selectionordistributionof the items.
Range The difference betweenthe highestandlowestscore ina data set.
ReliabilityA measure of consistencybothwithinasetof scoresor items(internal reliability) andalso
overtime suchthat it ispossible toobtainthe same resultsonsubsequentoccasionswhenthe
measure isused(externalreliability).The reliabilityof anexperimentcanbe determinedthrough
replication.
Repeatedmeasures Each participanttakespart ineveryconditionundertest.
Right to withdraw Participantsshouldhave the righttowithdraw fromparticipatinginastudyif
theyare uncomfortable inanyway,andshouldalsohave the rightto refuse permissionforthe
researchertouse any data theyproduced.
Role play A controlledobservationinwhichparticipantsare askedtoimagine how theywould
behave incertainsituations,andactoutthe part. Thismethodhasthe advantage of permittingone
to studycertainbehavioursthatmightbe unethical ordifficulttofindinthe real world.
SamplingThe processof takinga sample whichisintendedtoarepresentative selectionof atarget
population.
Scattergram A graphical representationof the relationship(i.e.the correlation) betweentwosetsof
scores.
Significance A statistical termindicatingthatthe researchfindingsare sufficientlystrongforusto
accept the researchhypothesisundertest.
Single blindA type of researchdesigninwhichthe participantisnotaware of the researchaimsor
of whichconditionof the experimenttheyare receiving.
Situational variables Factorsin the environmentthatcouldaffectthe dependentvariable suchas
noise,time of dayandalso the behaviourof aninvestigator.
Social desirabilitybias A tendencyforrespondentstoanswerquestionsinsuchaway that presents
themina betterlight.
Standard deviationshowsthe amountof variationina data set.It assessesthe spreadof data
aroundthe mean.
Stratifiedsample Groups of participantsare selectedaccordingtotheirfrequencyinthe population
inorder to obtaina representative sample.Groupsorstrata are selectedfromthe targetpopulation
that needtobe representedandthenindividualssampledfromthe strata.If thisfinal sample isdone
usinga randomtechnique itisa stratifiedsampleotherwiseitisa quotasample.
40
Structured (systematic) observations Anobserverusesvarious‘systems’toorganise observations,
such as behavioural categories andsamplingprocedures.
Structured interviewAnyinterviewinwhichthe questionsare decidedinadvance.
Systematic sample A methodof obtainingarepresentativesample byselectingevery5thor10t
th
person.Thiscan be a randomsample if the firstpersonisselectedusingarandommethod;thenyou
selectevery5th
or 10th
personafterthis.
Target population The group of people thatthe researcherisinterestedin.The groupof people
fromwhoma sample isdrawn.The groupof people aboutwhomgeneralisationscanbe made.
Time samplingAnobservational techniqueinwhichthe observerrecordsbehavioursinagiventime
frame,e.g.notingwhata target individual isdoingevery30seconds.You mayselectone or more
categoriesfroma checklist.
UnstructuredinterviewThe interview startsoutwithsome general aimsandpossiblysome
questions,andletsthe interviewee’sanswersguide subsequentquestions.
Unstructuredobservations An observerrecordsall relevantbehaviourbuthasno system.This
technique maybe chosenbecause the behaviourtobe studiedislargelyunpredictable.
VolunteerbiasA formof samplingbiasbecause volunteerparticipantsare usuallymore highly
motivatedthanrandomlyselectedparticipants.
Volunteersample A sample of participantsproducedbya samplingtechnique thatreliessolelyon
volunteerstomake upthe sample. Oftenrecruitedbyanadvertisementinarelevantplace.
Zero correlationIn a correlationthe co-variablesare notlinkedatall.
Research methods – all three papers, particularly paper two
You may have to designsomething –informedconsent,debrief,examplesfromaquestionnaire
22 The psychologist needed to be able to make sure that his participants understood the
nature of the study so that they could give fully informed consent. Write a consent form for
this study. Make sure there is sufficient information about the study for the participants to
make an informed decision. (5 marks)
During this study, the aim is to investigate the effect of a restricted diet upon
psychological functioning,particularly foodperception.
Ifyouagree topartakeinthestudy,youwillspendfourweeksataresearchunitwith19
otherparticipants.Youwillbe keptincomfort,allowedtosocialise,watchtvetc.butyouwill
onlybeallowedtoeatatmealtimes.Youwillbegivenonlywatertodrinkandplainfood.
Over the course of the study, we will carry out tests of emotional and cognitive
functioning.Atthe endofeachweekyouwillbe askedtodrawapicture offoodwhichwillbe
usedtoassessyourperceptionoffood.
All information we acquire about you over the course of the research will remain
confidentialandwewillnotuseyourrealnameinanypublication.
It is NOT compulsory for you to participate. If you don’t wish to, simply return this
form. You can also withdraw at any point during the study by notifying one of the research
assistantsthatyouwishtoleave,and there doesnothavetobeareasongiven.Ifyouconsentto
participate,pleasesignbelow.Doyouhaveanyquestions?
Always include a check of
understanding – it’s worth a
mark
42
Other things worth a mark:
Titles on graphs – easily forgotten and will drop a mark
If you are writing the same thing twice one of them is wrong
RTFQ
Some sample answers:
Qs 23 and 24 are based on the following stem:
23 State whetherthe hypothesisforthisstudywasdirectionalornon- directional (1mark)
Directional
Research methods – all three papers, particularly paper two
24 UsingTable 1, state whetherthe psychologist’sresultwassignificant.Explainyouranswer. (3
marks)
Itwassignificant.Becausehehad20PPs, andasignificancelevelof5%, thecriticalvalueis60.HisT-
valuemustbelessthanorequalto60and,at53,itislesssoitissignificant.Thismeansthatwecanreject
thenullhypothesisandacceptthealternativehypothesis.
26 Usingthe data inTable,explainwhythe psychologistisconcernedaboutinter-raterreliability.
(4 marks)
Therearealotofdifferencesbetweentheratingsgivenbythetwoobserversforthesametimeslots.
Inter-raterreliabilityisthedegreetowhicharatingsystemproducesreplicableresultswhenusedby
differentratersforthesamesubject.Theobserversmusthavehaddifferentideasofwhatconstituted
‘verbalaggression’astheydisagreeineveryslot.Forexample
27 Identifyanappropriate statistical testtocheckthe inter-raterreliabilityof these twoobservers.
Explainwhythisisa suitable test. (3 marks)
TheyshouldusePearson’sr.Thedataisintervalandatestofcorrelationisrequired.
28 If the psychologistdoesfindlowreliability,whatcouldshe doto improve inter-raterreliability
before proceedingwith the observational research? (4 marks)
She could give the two observers training to try and make them aware of what counts as verbally
aggressive behaviour. She could also clarify the scoring system to help observers determine if a
behaviour should be counted or not. Having a behaviour checklist with well-defined behaviours
44
would help. She could run a pilot study using the checklist to ensure that both observers have had
practisewiththechecklist,andtocheckthattheyhaveareliabilitylevelof0.8orhigherbeforerunning
themainstudy.