Nigel Holt and Rob Lewis - Crown House Publishing.

Sample chapter:Psychological Research and Scientific MethodsCrown House Publishing Limitedwww.crownhouse.co.ukA2L E V E LPsychologythe student’stextbookNigel Holt and Rob Lewis

2 | A2 PSYCHOLOGY: THE STUDENT’S TEXTBOOKPsychologicalResearch andScientific MethodsNigel Holtwhat you need to knowThis section builds on the knowledge and skills of research methods developed at AS level.You are expected to be able to understand the application of scientific method in psychology,design investigations, understand how to analyse and interpret data arising from suchinvestigations, and report on practical investigations.Methods and techniques●●The major features of science●●The scientific process, including theory construction, hypothesis testing, use of empiricalmethods, generation of laws/principles●●Validating new knowledge and the role of peer reviewInvestigation design●●Selection and application of appropriate research methods●●Implications of sampling strategies●●Issues of reliability, including types of reliability, assessment of reliability, improvingreliability●●Assessing and improving validity (internal and external)●●Ethical considerations in design and conduct of psychological researchData analysis and presentation●●Appropriate selection of graphical representations●●Probability and significance, including the interpretation of significance and Type 1/Type 2errors●●Factors affecting choice of statistical test, including levels of measurement●●The use of inferential analysis, including Spearman’s Rho, Mann-Whitney, Wilcoxon,chi-square●●Analysis and interpretation of qualitative data●●Conventions of reporting on psychological investigationsExtract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

the application of scientific mETHOD in PSYCHOLOGY | 3THE APPLICATION OF SCIENTIFICMETHOD IN PSYCHOLOGYResearch methods is not just mathsResearch methods is one of the areas in psychology that some students find extremely difficult, and that moststudents have trouble with from time to time. Many people doing psychology confuse ‘statistics’ with ‘researchmethods’. It’s true that you do have to use some mathematical skills, but really methods are the vital skillsthat allow you to carry out research in psychology, and to look carefully and critically at the research that hasalready been done. A decent knowledge of research methods will enable you to look at the work of psychologistsand to criticise their methods and findings. Evaluating research in this way allows us to think clearly aboutwhether the work is reliable and valid, two terms that are extremely important in psychology, and which willfeature later in this chapter.Another problem people have is that they think that ‘research methods’ is just another separate area of psychology.That’s partly the fault of textbook writers like us, and partly the fault of generations of psychologists.It just so happens that research methods tends to feature as a separate chapter and is therefore regardedas something you read about once, and forget about for the rest of the book. We would urge you to regardresearch methods as central to your study of psychology, with all the other bits (developmental, cognitive,social, etc.) as separate important extras. Methods will provide the backbone and the other areas will providethe fabulous and interesting covering. Without one you cannot have the other.If you are using this book as part of an A Level course you will already be familiar with a good deal ofthe information you need to know about the scientific method, since it has been covered in the AScourse. Much of it will be summarised as we go through this chapter. Methods can be tricky in places:there’s no getting around the fact, but a sensible understanding and knowledge of where to findrefreshment for an overworked memory can do wonders for confidence. There is nothing at all to say thatyou cannot flick back to your AS books, or even get a study guide to help you summarise what you need to know.Remember, research methods at A2 extends the knowledge and understanding of this area that you gained at AS.The major features of scienceBy now you should be quite accustomed to theidea that psychology is a science. Many peopleare not at all familiar or comfortable with oursubject being described in this way. They feelthat science is something that people in whitecoats do while locked up in laboratories housingdangerous chemicals or machines. They feel thatbecause psychology often deals with opinions andemotions it is not a science at all. They are wrong.Psychology follows the principles of all good sciences.These are the application of the scientificmethod, with careful consideration of replicabilityand objectivity.There is no great secret to identifying the scientificmethod. Your previous studies have already introducedyou to the concept. The scientific methodrefers to ways of thinking about evidence and ofcarrying out research that allow us to develop whatwe already know. It is because of the scientificmethod that we now have the technology we do,and it is largely because of the application of thescientific method that we now know as much wedo about psychology.Hypothesis TestingThe scientific method progresses by hypothesistesting. You know by now that a hypothesis isa statement that can be tested in research. Forinstance, we might be interested in investigatingwhether people really do feel happier in the summerwhen there is more sunshine. Our hypothesismight be:‘People feel happier in the summer.’We must now do our best to test this hypothesis.We must try everything we can to disprove it. Ourefforts must be focused on finding out whetherwe can prove that people do not feel happier inthe summer. If we can do that then we can rejectour hypothesis. The process of trying to disprove ahypothesis is absolutely central to the progressionof all sciences, and psychology is no different inthat respect from physics, chemistry or biology.Extract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

the application of scientific mETHOD in PSYCHOLOGY | 53a. Do NOTsupport theHypothesis1. TheHypothesis2. TestHypothesis3b. Support theHypothesis4. TheTheory1. The HypothesisThe scientific process begins with the hypothesis:this is the statement, based on the aims of ourresearch, that will be tested.2. Test HypothesisAppropriate methods of investigation are chosenthat allow us to test the hypothesis.3a. Do NOT support the HypothesisIf the results of the tests are inconsistent withthe hypothesis and therefore do not allow us tosupport it, we go back to the drawing board, andrethink the hypothesis itself. The procedure beginsagain, flowing through points 1, 2 and 3.3b. Support the HypothesisIf the results of the tests chosen in point 2 areconsistent with the hypothesis then we can acceptit and move on to our final goal.4. The TheoryThe findings of carefully constructed research thatsupport the hypothesis allow us to begin to form atheory or modify one that already exists.Theory ConstructionTheories are constructed by using the route ofhypothesis testing, development and retesting thatwe have identified here as the scientific process.In the majority of cases a theory does not dependon one piece of research. It is the end result of arange of work, usually from a number of differentsources and researchers. Once the theory is developedit continues to be tested – and this bringsus to a very important point. The most importantaspect of a scientific theory is that it must betestable and ultimately falsifiable. This may soundodd, but a good theory must always be falsifiable.By this we mean that researchers must be able todevelop ideas that can ultimately end in the rejectionor development of the theory. Some of themost famous theories ever developed in scienceare just theories and are still being tested. One daythey too may be shown to be incorrect in someway, making room for new theories. For instance,Einstein developed the theory of relativity in aseries of scientific works between 1905 and 1915.While a huge amount of research in science hastaken place on the assumption that this famousidea is entirely correct, the theory may yet beshown to be wrong. If this is the case then it maybe developed or rejected altogether. This is thehallmark of a good, scientific theory.The problem of unfalsifiability is one of the criticismsof the work of Sigmund Freud. His theoryof the subconscious is that it is made up of threeinteracting components – the id, ego and superego.This idea is unfalsifiable. It cannot be testedand as such is not regarded as truly scientific. Itseems that some parts of psychology are morescientific than others!The generation of laws/principlesA scientific law, or principle, is a statement thatdescribes the behaviour of things in the real world.For instance, Newton described a series of lawsthat tell us how items behave in the universe.Newton’s first law, for instance, says that everyobject in a uniform state of motion tends toremain in that state of motion until an externalforce is applied to it. This means that if somethingis moving along it will continue to move along inthe same way unless it is influenced by somethingelse. This is a law. It always happens. A scientificlaw or principle can be thought of as a developmentof a scientific theory. The theory is testedand retested and, once there is enough evidencein its favour, it can be developed into laws andprinciples.Inductive reasoningLet’s return for a moment to Newton’s laws. Howcan we really accept Newton’s ideas as laws whenit is impossible to have tested every single incidenceof an object bumping into another object.We cannot possibly know whether Newton’s ideashold true all the time so how can they be regardedas laws? The answer to this lies in the principle ofinductive reasoning. If something usually works andif every observed case of something provides thesame result, we can induce from this that the samething will happen for all cases. In other words, weExtract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

6 | A2 PSYCHOLOGY: THE STUDENT’S TEXTBOOKcan generalise our law and apply it in many situationsto explain other things. Here’s an example.What colour are crows? You have not seen everycrow in the world, but no doubt you have seenquite a few. The ones you will have seen are allblack. You therefore induce from your observationsthat all crows are black. This example also demonstratesthe value of inductive reasoning in science.Conclusions based on inductive reasoning are onlyever generalisations based on limited evidence (forexample, the evidence of your observations thatall the crows you have ever seen are black). Wecan make inductive arguments more probable byadding evidence in their favour, or we can questionthem with contradictory evidence (for example,we go out and look for a crow that is not black).Either way, we are increasing our understandingthrough science.ParsimonyThere is often more than one explanation for ahypothesis. One of them might be quite complicated,requiring complex arguments and furtheruntested assumptions. On the other hand, anotherexplanation might be much more straightforward,with its assumption based firmly on what wealready know. Despite the fact that both accountsexplain the hypothesis, the principle of parsimonystates that we should select the least complex one.In this way, parsimony ensures that our explanationdoes not go beyond the available empiricalevidence. We can see the principle of parsimonyin the logical principle of Occam’s razor. This isa principle identified by a monk called Williamof Ockham in the 14th century which basicallystates that, given two or more explanations for anGreat thinkers: Karl PopperKarl Popper (1902–1994) provided a vital way of thinking about science that has led to today’s understandingof hypothesis testing and theory construction. It was Popper who said that a theory must be falsifiable in orderto be considered as scientific. Any amount of evidence can be presented to support a theory, but only a singlepiece of evidence that shows the theory to be incorrect is needed for it to be falsified. Popper went on to usethis central idea of science to attack the study of psychoanalysis and the work of Sigmund Freud as unfalsifiableand therefore not scientific.Great thinkers: Thomas KuhnThomas Kuhn (1922–1996) wrote a very important book called The Structure of Scientific Revolutions. In it hesays that science progresses quite happily most of the time but occasionally it undergoes a sort of shake up,which is described as a paradigm shift. A paradigm is a way of thinking about something. If, perhaps because ofthe importance of a small group of researchers, people begin to think about their science differently, then thedirection taken by science will change. Psychodynamic theory, behaviourism and biological ways of thinkingabout psychology, for example, can all be described as paradigms.Kuhn says that each paradigm, or way of thinking, has its own framework of assumptions. Because of this,someone working in one paradigm cannot provide evidence in support of, or falsifying, a theory in anotherparadigm, because the principles upon which each paradigm is based are different. In psychology we can thinkof it as being illustrated by the example that research into the ideas of Freud under the paradigm of psychodynamicismcannot falsify research provided by medical psychologists in the neuropsychological paradigm.This is because the psychodynamic paradigm assumes the existence of the subconscious, and it is partly uponthis assumption that the research is built. The neuropsychological paradigm is concerned with the physicalactivity of the brain and does not accept that the subconscious exists. It follows that information from the oneparadigm cannot be used to falsify theories from the other.The implications of this idea are huge for those interested in driving science forward and testing and buildingtheories. Earlier in this chapter we discussed one of the principles of science, formalised by Popper, thata theory must be falsifiable. If the work of a researcher shows that a theory is incorrect, that theory willbe rejected or changed to accommodate the new findings. If Kuhn’s thinking is to be accepted, a paradigmshift will change this idea completely. If the discipline undergoes a change in direction, brought about by aparadigm shift, then the findings of a researcher, working under assumptions of an old paradigm, may be seennot as ‘important, and falsifying the theory’. Instead they may be seen as a mistake made by the researcher,and the new paradigm theory continues, unchanged, simply because an old paradigm was used to derive theresults that would otherwise have falsified the theory.Extract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

the application of scientific mETHOD in PSYCHOLOGY | 7observed event, we should select the simplest oneas the better. It’s as straightforward as that.Validating new knowledgeJust because a researcher carries out some workthat he or she is convinced will change the world,this does not mean that it will be accepted bythe rest of the scientific community. Having workpublished really gives it a stamp of approval. Ifsomething appears in print then other scientistscan have access to it, and thus have the chanceto read and challenge it. Traditionally, research ispublished in scientific journals. These are magazine-likepublications which appear one or moretimes a year. Getting research published howeveris not as easy as just sending it to the publisher. Inscience the principle of peer review is applied.Peer reviewWhen a psychologist completes a piece of researchit is written in a generally agreed format and thensent to a psychological journal for consideration.If the journal thinks the work is good enough theywill publish it for all to see. The process of assessingwhether it is good enough for publication isnormally peer review. The work is sent to one ormore established scientists with similar expertise(who usually remain anonymous) for them to readand perhaps criticise its method and thinking. Itis then sent back to the researchers with recommendationsfor revision. If the work does not meetthe standards of the peer reviewers it will not bepublished. This system ensures that standards ofquality are maintained in research and that unsubstantiatedclaims are not made.The peer review system is certainly very highlyregarded but it is not without its critics. Forinstance, just because something is not thoughtby a reviewer to be appropriate for publishing, thisdoes not mean that others would feel the sameway. It could be that the reviewer is providing afair assessment, in which case the work should berejected and looked at by the researchers again. Itcould be, however, that the reviewer is biased insome way. It may be that the paper under considerationis expressing an opinion that the reviewerdisagrees with. If this is the case the reviewer maybe approaching the review from a position of bias.In situations like these, research may not be widelyseen by the community, just because of the opinionsof a single reviewer.Some peer reviewed academic journals are morehighly thought of than others, and researchers trytheir best to have their work published in them.For instance, the journals Nature and Science areregarded as extremely prestigious. Universitiesencourage their researchers to publish in suchjournals and not in less prestigious titles – and forthis reason, some research, if rejected by highlyrespected journals, may never see the light of daybecause researchers feel “If it’s not good enoughfor Nature, it’s just not good enough.”Academic journals are extraordinarily expensiveto buy. Many university libraries can only afforda relatively small number each year, and mostprovide the journals online through very expensivelicensing agreements, with annual costs oftenrunning into hundreds of thousands of pounds.This means that the general public, or less wealthycolleges and universities, may not have accessto the published research and so it remains relativelyunread. Its use in advancing psychology istherefore limited. For reasons like these, excitingalternatives to traditional publishing have appeared– the latest of which is open access.With open access, new knowledge can be viewedby the scientific community as well as the generalpublic, allowing everyone to make up their ownminds about the research. Open access can becombined with a form of peer review, as with theonline journal Philica.com. On sites like these newideas are rated by academics. Those rated as goodcontain ideas that might be trusted; those rated asbad are less likely to be trusted.Despite numerous obstacles and sometimes longtime scales involved in getting research published,the publication of psychological research is importantfor many reasons. Journals are international,which means that there will be widespread disseminationof the new research among peers. Thework will therefore be more likely to be discussedand debated and the ideas possibly built uponExtract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

8 | A2 PSYCHOLOGY: THE STUDENT’S TEXTBOOKThe Publishing ProcessResearcher submits an article to a journal. The choice of journal may be determined by:●●The journal’s audience: is it the appropriate audience for the article?●●The journal’s prestige: is it well known? Is it often cited?The author is unpaid, but may pay ‘page charges’ (see terminology).The journal selects two or more appropriate experts to peer review the article, without payment.The peer reviewers assess:●●The quality of the research and the way it is reported●●The relevance of the article to the journal’s readership●●Its novelty and interest●●Its content, structure and language.Feedback from the reviewers determines whether or not the article is accepted. Acceptance rates vary fromjournal to journal.The rejected article is returned to the author or the accepted article is passed to the editors employed by thepublishers, either in-house or freelance.The editors ensure that:●●The language of the article (particularly if it has been produced by a non-native speaker) is clear andunambiguous●●The standard style of nomenclature is adhered to●●Illustrative material is of a sufficiently high standard.Editors also establish live links in the electronic version to the references cited and to other material such asdata sets.Where a journal exists in both paper and digital formats, the article is sent to the printer to produce a papercopy; and will be formatted for the online version of the journal.Reproduced from: Select Committee on Science and Technology: Tenth Reportwith further research. The whole community willbenefit from news of new developments, andbecause peer review ensures that only the bestresearch gets published, the scientific communitycan be assured of the quality of the research.Publication in peer reviewed journals can do agreat deal to enhance the reputation of researchers.Not only are they more esteemed but, becauseof a good research and publication record, theyalso stand a better chance of obtaining fundingfor further research. Published research alsoprovides enormous benefits for the institutionsthe researchers work for. Most research takes placein universities, which are funded by governmentgrants. The amount of money a university getsis partly based on periodic audits of the university’sresearch output, including the amount ofpublished work they generate (the process iscalled the Research Assessment Exercise, or RAE).The publication of journal articles is therefore ofcrucial importance to both the researcher and theirrespective institutions.Extract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

10 | A2 PSYCHOLOGY: THE STUDENT’S TEXTBOOKone or two methods because their research is veryfocused on particular areas, but for new psychologistsa good overall knowledge is required.It is useful to make sure that you know the differencebetween an experimental design and anon-experimental design. In a true laboratoryexperiment, it is the researcher who makeschanges to variables and measures the result ofthose changes. For instance, in an experimentto measure the influence of chocolate eating onfeelings of happiness it will be the experimenterwho will give the participants one, two or threepieces of chocolate to eat, and who will measure,after each is consumed, how happy the personis. It is the experimenter who is in charge of theindependent variable, in this case the chocolate.In a non-experimental design a variable is notmanipulated by the researcher.Much of this material will be familiar toyou from AS level. However, becauseit is vital that you understand all theresearch methods you cover on the ALevel course, we will now provide a usefuloverview. Remember – you are expected to knowthis stuff! Be prepared to revisit your AS work toexpand on what we present here. For example,make sure you are aware of the advantages andweaknesses of the various methods.Experimental designsWhen we carry out an experiment, we changesomething and see what happens to somethingelse. As we’ve already said, the something wechange is called the independent variable, and thething that changes as a result is the dependentvariable. Once you’ve decided that you are runningan experiment there are three designs open toyou. Each has its advantages and weaknesses.Think carefully about the three differentdesign choices. If you are asked in theexam to evaluate the methodologyused in a piece of experimental research,displaying a knowledge of the differenttypes of designs and the advantages and weaknessesof each would be an ideal way to gain marks.To make things simpler, we will discuss three differentdesigns for the same experiment. Imagineyou were interested in investigating whether it waspossible to stand on one leg longer in silence thanwhen listening to music. You never know, youmight be interested in something like that. Yourindependent variable would be whether there wasmusic, or whether there was silence: it has twolevels, music and silence. Your dependent variable(the thing you measured) would be the time theperson managed to spend standing on one legwithout falling over. The three possible designsfor your experiment are identified and describedbelow.1. Repeated measures designA repeated measures design is one where each participanttakes part in each level of the independentvariable. In our imagined experiment, this meansthat each participant does the test in silence andthen each of them does the test with music.AdvantagesThe same person takes part in each level of theindependent variable and so there are no problemsof individual differences, such as fitness andage. This means you do not need to use a hugenumber of participants, so the research can becompleted sooner and, if you are paying yourparticipants, for less money.WeaknessesThere are possible order effects. This means thatthe time spent standing on one leg in silence mayhave influenced the time spent on one leg whenlistening to music. The person may have becometired and fallen over more easily on the secondoccasion. This is called a fatigue effect. Similarly,the person may have become more used to standingon one leg, lasting longer on the second occasion.This is called a practice effect.2. Independent groupsAn independent groups design uses two groupsof participants. One group stands on one leg insilence and one group stands on one leg whilelistening to music.AdvantagesThe major advantage of an independent groupsdesign is that it completely eliminates order effectssuch as practice and fatigue.WeaknessesThe difficulty with independent groups designsis that they are open to problems associated withindividual differences. For instance, imagine theUK ‘standing on one leg’ Olympic team was inone group, and the old ladies’ lawn bowls teamwas in the other group. That would hardly makea fair comparison! The way to avoid this is to haveExtract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

DESIGNING PSYCHOLOGICAL INVESTIGATIONS | 11two huge groups. If you do that, the range of individualdifferences in one group would be similar tothe range in the other group.3. Matched pairsThe matched pairs design is like an independentgroups design, but each person in one group ismatched with a person in the other group. Forinstance, if you had an 82-year-old pensionerin one group, you should match them with an82-year-old pensioner in the other group. Eachperson in each group has a matched person in theother group.AdvantagesThe great thing about matching is that it controlsfor individual differences, and since it is a variationof an independent groups design, there are noproblems with order effects either.WeaknessesMatching is extremely difficult, or even impossible.Even if you do have an 82-year-old pensioner ineach group you are not to know whether one hashad a huge amount of ‘standing on one leg’ experiencewhile the other may be an absolute novice. Infact, even if you had identical twins, and allocatedone to each group, you could not be sure whethertheir experiences and skills were identical. Theseindividual differences have not been matched,and may influence your result. It is impossible tomatch people in different groups perfectly.The laboratory experimentIn general, the laboratory experiment is run incarefully controlled conditions, and uses a standardisedprocedure. Variables are manipulated (theindependent variable), are controlled (extraneousor confounding variables) and measurements arecarefully taken (the dependent variable).A good laboratory experiment has the followingcharacteristics. It is…Replicable – it can be repeated by otherresearchersGeneralisable – the results can refer to peopleoutside the sample that you testedReliable – a repeat of the experiment will yield thesame resultsValid – It measures what it says it measures.The field experimentThis is like a laboratory experiment, but the variablesare manipulated by the researcher out in amore real-world setting. The advantages of this arethat the real-world setting offers ecological validityand the people acting as participants in the experimentoften need not know they are part of thestudy, eliminating any demand characteristics. Theproblem however is that these studies are oftenextremely expensive and time consuming to carryout.The natural experimentHere the researcher makes use of a naturallyoccurring independent variable. For instance,the researcher may have the idea that towns nearthe seaside are perceived as more fun than townsinland. The researcher cannot change the positionof the sea and so the independent variable, in thiscase proximity to the sea, is naturally occurring.The advantages of the natural experiment are thoseof the field experiment: it is high in ecologicalvalidity and there are few or no demand characteristics.The disadvantages are that the researcherhas no control at all over the independent variableand so the chance of unwanted things influencingthe experimental findings (confounding variables)is quite high.The quasi experimentThis looks rather like a lab experiment but it’s not.Imagine you wanted to see whether Americans orEnglish people were better at mathematics. Youwould locate a group of Americans and a group ofEnglish people and give each group a maths test.The allocation of participants to a group is doneon the basis of their nationality, your independentvariable. However, nationality cannot be manipulated,and so it’s not a true experiment.The case studySome research offers a careful and systematicinvestigation of the case of an individual. Examples●●ReplicabilityAdvantagesLABORATORY EXPERIMENTWeaknesses●●Cannot always use a lab experiment●●Easy to control variables●●Controllability●●More control means less natural●●May be demand characteristics and experimentereffects, both of which may influence the resultsExtract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

12 | A2 PSYCHOLOGY: THE STUDENT’S TEXTBOOKmight include Freud’s investigation of Anna O, orthe case of Phineas Gage.The advantages of a case study are that it can beextremely detailed and may provide wonderfullyrich and informative data about the person. Thedisadvantage of course is that generalisabilityis extremely low, as the case of that particularindividual may well be unusual and unrelated tothe rest of the population. Also, as case studiesgather data from the past, the information may notbe very reliable. Memory has an annoying habitof fading over time, and the responses a personmay give regarding their past may not be entirelyaccurate.The observational methodThis is extremely obvious at first glance, anddeceptively simple. It clearly means a methodwhere researchers observe the behaviour of others.You will not be surprised to hear that it is not quitethat simple. There are two main divisions of observationalmethod. Participant observation is wherethe person doing the observing is part of the groupthey are observing. For instance, if observing thebehaviour of a football team, the observer maybe part of that team. Non-participant observation iswhere the observer is not part of the group theyare observing. For instance, an observer may beinvestigating incidences of aggressive behaviour inthe playgrounds of secondary schools. They themselvesare not part of the group of students beingobserved. There are three things to watch out forin the observational method:1. Awareness of the observer: If those beingobserved are aware of the person doing the observingthen their behaviour may change and will notbe natural.2. Behavioural categories: The observer mustdecide what exactly they are looking for in theobservation. These categories can form a checklistfor use during the observation.3. Bias: If the observer is not objective the resultswill be biased. Using multiple observers andcomparing their observations after each sessioncan overcome this problem.Finally, the observational method can be carriedout in two different ways. Having chosen thebehaviour to be observed – for example, aggression– the observer may count how many incidencesof the behaviour occur during a period oftime (time sampling), or may record all incidencesof the behaviour (event sampling).When you think about observationalmethods, consider possible ethicalimplications of these methods. Is it ethicallyacceptable to observe someonewho doesn’t know they are being observed?Is it ethically acceptable to observe groups such aschildren, who may be regarded as ‘vulnerable’?Observational methods vs.observational techniquesObservation can be a method and observation canbe a technique. The observational method is a typeof research in its own right. The observationaltechnique, however, is something that you can usein any type of research. It’s best to think of theobservational technique as a tool that researchersuse and the observational method as a style ofresearch that they undertake. For instance, youmight use the observational technique to count uphow many people on the high street are carryingsales bags after an advertising campaign by a localshop. You are using the technique of observing tomeasure the impact of the campaign in a naturalexperiment.Content analysisContent analysis is a form of observationalmethod, but instead of watching people behave,the researcher analyses the content of a text orfilm. For instance, they may be investigating theuse in newspapers of terms that might be regardedas racist, and how this might have changed overthe years. They would decide on the racist termsSAMPLING OF OBSERVATIONSAdvantagesWEAKNESSES●●Natural settings, so natural behaviours are observed.●●Allows otherwise impossible or hard-to-examplebehaviours (e.g. aggression in children) to beinvestigated.●●Few demand characteristics because the observedare not put into a false situation and may not knowthey are being observed at all.●●Observer bias if researchers don’t remain objective.●●Confounding variables may interfere, making it hardto determine the cause for something happening.●●Small groups are usually observed, making theresults hard to generalise.Extract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

DESIGNING PSYCHOLOGICAL INVESTIGATIONS | 13to watch out for and then look through newspapersmarking down how often the terms wereused.Self-report techniquesThese include questionnaires and interviews.Here a participant records or expresses their ownopinions and feelings about something. Bothtechniques have individual strengths and weaknesses,but both share the weakness that they areself-report and so the truthfulness of the answersmay be in question as participants may respond insuch a way as to make themselves appear sociallydesirable.A questionnaire is a list of prewritten questions.These may be ‘closed’, where alternative responsesare offered; or ‘open’, where free responses areoffered; or possibly rank-order, where you arerequired to place things in some kind of order. Theadvantages of questionnaires are that they are acheap way of collecting lots of data while retainingparticipant anonymity. The disadvantages are thatthose who agree to complete the questionnairesmay be doing so because they like to help out,or the topic covered is of interest to them. Thesample may, in this way, be biased. Also, somepeople chosen to complete the questionnaire at‘random’ may decide not to do so, which meansyour random sample is no longer random, as youhave to choose someone else to take part.An interview is like a questionnaire, in that itprovides self-report data drawn directly from aparticipant. Interviews may be ‘structured’, wherethe questions are very organised and rigid; theymay be ‘semi-structured’, where the control andorganisation is less rigid and the interviewer isoffered more flexibility; or they may be ‘unstructured’,where the interview is completely free anduncontrolled. The advantages of the interview arethat it can provide very rich and insightful dataand can be very simple and fast to carry out, withgeneralisable findings if the sample is big enough.The disadvantages are those of the questionnaire.In addition to this, a person may provide quitedifferent responses to the same questions in aninterview on different occasions, thus threateningthe validity of the results.Correlation methodCorrelational analysis lets us see how two variablesare related. For instance, height and shoe size; orability at maths and ability at chess. Just as withobservation, the correlational method is a researchdesign option in its own right, but correlationanalysis can be used with data from any othermethod where appropriate. The analysis providesa statistic called the correlation coefficient, whichvaries from –1 to +1.A correlation coefficient of near to –1 indicates astrong negative correlation, which means as onevariable increases in size the other decreases. Forinstance, the more money you spend, the lessmoney you have to spend. This may not be aperfect negative correlation because we have notaccounted for the amount of money you may beearning, so occasionally spending some moneymay not, in fact, result in a reduction in theamount that remains, if some wages have beenpaid to you.A correlation coefficient of 0 means that there isno relationship at all between the two variables.For instance, there is no relationship between theamount of carrots grown in Israel and the numberof cartoons on the TV between 3 and 4p.m. on aTuesday afternoon.NEGATIVE (LESS THAN 0)A ‘downward slope’ is seen.ZERO CORRELATIONNo slope is seen.POSITIVE (MORE THAN 0) An‘upward slope’ is seen.amount of money left to spend20151050STRONG NEGATIVE CORRELATION++++ + ++ +0 3.75 7.50 11.25 15.00amount of money spentnumber of cartoons on TV20151050ZERO CORRELATION+ ++++ + + +0 3.75 7.50 11.25 15.00amount of carrots grownwetness rating20151050STRONG POSITIVE CORRELATION+ + + ++ + + +0 3.75 7.50 11.25 15.00time spent rainingThe more perfect the line, thecloser to –1 the correlation is.A strong negative correlation isone close to a perfectly straightdownward-sloping line.There seems to be no patternhere.The more perfect the line, thecloser to +1 the correlation is.A strong positive correlation isone close to a perfectly straightupward-sloping line.Extract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

14 | A2 PSYCHOLOGY: THE STUDENT’S TEXTBOOKCORRELATIONSadvantagesweaknesses●●Naturally occurring variables can be measured and ●●Cannot imply cause or effect, and variables area relationship between the two identified. This may almost impossible to control. Even if the analysislead to future research.indicates that there is no relationship we cannot becertain of that. It could be that something else thatwe have not measured or considered is happeningthat the analysis is not sensitive to.A correlation coefficient of near to +1 means thatthere is a strong positive correlation between thetwo variables. As one goes up, so does the other.For instance, the longer it rains, the wetter you willget if standing outside. This may not be a perfectpositive correlation because it may be that at somepoint your clothing becomes completely saturatedand, however much more rain falls on you, youcannot possibly get any wetter.Correlations are represented on scattergraphs (alsoknown as scattergrams). On page 13 we have threescatterplots identifying the relationships describedabove.Do not think that correlation means cause. It doesnot. Just because two variables are positively correlateddoes not mean that one causes the other.For instance, it is extremely likely that were you tocollect the data you would find that the numberof cars on the roads of the United Kingdom hasrisen steadily over the last 10 years. You would alsoprobably find that the number of people takingholidays abroad has gone up. The two variables(car sales and holidays abroad) are likely to bepositively correlated. We cannot conclude, however,that increased car sales have caused peopleto take more holidays. There may be a third factorthat we have not considered, such as ‘amount ofdisposable income’ that is related to both of thesevariables, but that we have not measured.Watch Out! Demandcharacteristics andinvestigator effectsThere are a few things to keep in mind whendesigning research. Demand characteristics, investigatoreffects, reliability and validity can all causeproblems for a researcher.Demand characteristics – Something in aresearch design may change a participant’sbehaviour. If this happens then your data will notgive you a true picture of what’s going on in yourstudy. For instance, participants who know somethingabout the procedure may alter their behaviourto provide the researcher with the responsesthey think the researcher would like. Researchersoften design ‘single blind’ procedures to eliminate,or at least attempt to eliminate, demand characteristics.These are procedures where participantsare kept naive about the aims of the research. Itis the job of the researcher to plan the study so asto minimise demand characteristics. Eliminatingthem is often very difficult indeed.Investigator effects – These are a little likedemand characteristics, but you should try not tomix them up! Research suffers with investigatoreffects when the researchers themselves influencethe behaviour of the participant in some way. Theymay not do so on purpose, but they may give awaytheir opinion in some way which may encourageInvestigator EffectsDemand CharacteristicsExtract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

DESIGNING PSYCHOLOGICAL INVESTIGATIONS | 15the participant to respond differently for the restof the session. Researchers may introduce a doubleblindprocedure to eliminate these effects. Here theresearcher and the participant do not know of theaims of the research, and so neither can influencethe results by inadvertently responding in certainways.Demand characteristics, investigatoreffects, reliability and validity areextremely useful things to considerwhen you comment on the researchyou read about. Have these in mind whenthinking critically about the methodology used. Youcan often pick up valuable marks by making one ortwo effective comments based on these issues whenevaluating research.Sampling strategiesFinding participants is something that all researchershave to do. A group of participants is referredto as a sample. A researcher cannot test everyonein the world, and would not want to. The idea isto select a sample of people who can be describedas representative of the wider population. If thesample really is representative then we can say thatwe are able to generalise our results to the largergroup. There are three sampling strategies, eachwith strengths and weaknesses that have implicationsfor your research.Random samplingThis is where all of the people in a populationhave the same chance of being selected.Generalisability and biasThe vulnerability of random samples to drop-outsmeans that the sample is easily biased, as theresearcher must either go with a reduced sampleor choose another random participant, who by thevery fact that they were not originally ‘randomlychosen’ is not very random at all.Similarly, it’s very difficult to carry out a trulyrandom sample if you have a very large populationas it is unlikely that you will have a complete list ofnames from which to choose. This means that noteveryone is equally likely to be chosen and so thesample is not truly random, but rather, it is biasedtowards ‘people on your list’.You must ask yourself how the names got onto thelist in the first place. For instance, researchers mayhave used a telephone directory. If this was thecase then the sample would be limited to thosethat own a phone and have their details listed.Some researchers may use the electoral register. Ifthey do this then the sample is limited to thosepeople who are 18 or over who have indicated thatthey wish to be able to vote in elections; and furtherlimited because people whose names are onthe full register are allowed to opt out of inclusionon the shorter version, to which researchers andothers have access. In both these cases the sampleis not generalisable to those in the population whodo not meet these criteria.Opportunity samplingThis is where anyone handy is asked to participate.The most typical opportunity sample in psychologyis one where researchers simply ask peoplein their department or college to take part. Forinstance, a researcher may put up a notice askingfor people to help out in their research, or maysimply approach people in the cafe and ask themfor help.Generalisability and biasTypically, researchers choose people to take partfrom their own social group. They may, if notdeliberately, choose people they would like to takepart, whom they find attractive in some way. Orthey may choose people who have taken part inprevious research, and whom they trust. It mayeven be that the researchers only ask those whothey think are least likely to reject their requests.Since choices are made, and the samples aredrawn, from such a narrow population, the extentto which the findings can be generalised to thewider population is very limited.Similarly, those who take part may have their ownmotives for doing so. At university, students maytake part in their lecturer’s research because theymay feel that helping out will gain them moremarks. This may leave the research open to bias inthat the participant may attempt to provide datathat will please the researcher in some way.The other problem with an opportunity sample isthat only those who want to take part do so. Thismeans that you may find that the sample has alarge proportion of ‘helpful’ or ‘interested’ participants,who may provide biased data. Obviously,this reduces the generalisability of the findings, asthe sample is not reflective of the wider, generalpopulation. This means it has low ecological validity– we’ll cover this more in a moment.Volunteer samplingThis is also known as ‘self-selected’ sampling. Herepeople who want to take part do so. For example,Extract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

16 | A2 PSYCHOLOGY: THE STUDENT’S TEXTBOOKa notice board or newspaper advertisement may beused to ask for volunteers to respond.Generalisability and biasThe findings of the research may lack generalisabilitybecause a volunteer sample may consistof mostly helpful, interested, more motivated orobedient people and as such it may not be like thegeneral population. In addition to this a volunteersample may be open to bias as those who take partmay do so in order to seem helpful and so pleasethe researcher – who is often their psychology lecturer,and who therefore has the power to providegood marks on their work!Issues of reliabilityReliability refers to how well the research can bereplicated at another time, or if two researcherscarried out observations how well their opinions ofwhat was observed agree. In short, reliability refersto the consistency of the research. As reliabilityincreases so does our confidence in the results.Observer reliabilityA widely used method and technique in psychologicalresearch is observation. Here the researchersobserve behaviour, marking down what theysee. Often they are looking for particular types orcategories of behaviour. For example, researchersmay be interested in investigating whether malechildren behave more aggressively at play whensurrounded by other males or when in a mixedmale/female environment. In this case researcherswould observe the behaviour of males in a maleonlyplay environment, recording the number oftimes they saw aggressive behaviour. They wouldthen compare these results with similar observationsmade in a mixed male/female environment.The results of research like this will not be ofmuch use if the observations are not done in thesame way – i.e. if they are not consistent. Weneed to ensure that another person observingthe behaviour would come up with the same, orat least a very similar, result. For this reason twoobservers make observations at each session. Theirresults are compared for similarity. If the two setsof observations are similar then we can say thatthere is good observer reliability.Assessing observer reliabilityComparing the observations of two observers isbest done using correlation. This relatively simplestatistical technique allows us to see how similartwo sets of values are. A good positive correlationshows that the two observers provided similarresults. If you can show a good positive correlationto two observations of the same event then youcan say confidently that you had good observerreliability.Improving observer reliability1. TrainingIf the two observers are able to easily identify thetype of behaviours they are watching then theywill be able to record those behaviours efficiently.Practice certainly makes perfect as observation is agreat skill. Until you try it, you’ll not fully believehow fast the different behaviours come and howvaried the behaviours that might be recorded arein different situations. For this reason, carefullytraining the observers before the research beginsis advisable as a means of improving observerreliability.2. OperationalisationBoth observers have to be clear exactly what theyare looking for. For instance, if they are observing‘helping’ behaviours then they must agree whatthis means. For instance, how ‘helpful’ does abehaviour need to be before they record it as‘helpful’? If the definitions of the behaviours tobe observed are clear and carefully laid out thenobserver reliability will be better than if categoriesof behaviour are badly defined.3. View of the behavioursEach observer must have the same ability to seethe behaviours being observed. The best way todo this is to have them both base their recordingsof behaviour on exactly the same video filmof the behaviours being recorded. This ensuresthat each observer sees exactly the same things.If the observations are recorded ‘live’, perhaps bywatching behaviours in a school playground, thendifferent sight lines may mean that incidences ofa behaviour may be missed by one observer butrecorded by the other. This would weaken thecorrelation between their results, and so weakenobserver reliability.Test reliabilityVarious tests are employed in psychologicalresearch to measure behaviour. If a researcher isinterested in assessing someone’s personality thenthey can use a test carefully designed to do thejob. These tests are often in the form of questionnairesand measurement scales that require theparticipants to indicate opinions of things, andhow they might respond or behave in differentsituations.Extract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

DESIGNING PSYCHOLOGICAL INVESTIGATIONS | 17Assessing test reliabilityAssessing test reliability is very similar to assessingobserver reliability. The method used here istest–retest assessment, or test–retest correlation. Herethe test is given to the person again, on a differentoccasion. The person’s responses to the test oneach occasion are compared and a correlationcarried out. If there is a high correlation betweenthe responses on each presentation of the test thenwe can say that there is a high level of test–retestreliability. This means that the test measures whatit says it does consistently.Improving test reliabilityAltering the test to improve correlationIf the test–retest correlation is low, then the reliabilityof the test is in question. This means thatthe researchers must alter the test to improve thisreliability. They can do this by looking carefully atthe test itself, and identifying the parts of it thatdid not correlate well on the two occasions it wasgiven. They can then remove those componentsof the test that are weakening its reliability andreplace them with alternative questions and tasks.This new test is then tested again, to see whetherreliability has improved, using the test–retest correlationassessment. This process is repeated untila high level of test–retest reliability is shown by astrongly correlated first and second running of thetest.Assessing and improvingvalidityValidity refers to how well the test or researchactually measures what it says it measures. A tapemeasure, for instance, is a very valid tool for measuringheight, but a stop-watch is not. Similarly,research that aims to measure the relationshipbetween mathematics ability and age is valid if itdoes just that, but not if it actually measures therelationship between age and general intelligence.Internal validityInternal validity is the extent to which we can saythat our findings are truly to do with what wethink they are to do with, rather than somethingthat we have no control over. For instance, we maybe interested in investigating whether memoryability is improved by eating chocolate. We mustbe sure that the person has not secretly drunklots of cups of tea before the experiment. If theresults of our research showed that their memorywas indeed improved by chocolate eating we cannotbe absolutely sure that the tea did not havesomething to do with it, and so the task has lowinternal validity.Improving internal validityAnything that influences the dependent variable(in this case the score on a memory task)rather than our independent variable (here it’schocolate eating) reduces the internal validity ofour research. In general these things are describedas extraneous variables. If an extraneous variable(such as fatigue or practice) influences our findings,then internal validity is reduced. One of themost important aspects of research design is tominimise the influences of extraneous variables.Improved internal validity comes from improvedresistance of your research to the influence ofextraneous variables. This is best achieved by verycareful research design and planning.External validityThe results of our research have external validityif they can be generalised from our sample to thegeneral population, in other settings beyond thosein which the research was carried out and at differenttimes.Ecological validityThis is a form of external validity. If the researchhas high ecological validity then we can say thatthe results relate to different situations from thosein which the research was carried out. Some workcarried out in a laboratory, for instance, may lackecological validity, because outside the carefullycontrolled environment of the laboratory peoplemay behave differently.Just because something is done in a labdoes not mean that it lacks ecologicalvalidity. Findings from laboratory basedresearch may transfer to other settingsvery well. Be careful of this in the exam.Don’t just say ‘It lacks ecological validity becauseit was done in a lab’. This won’t get you marks. Bemore specific. Explain what circumstances outsidethe lab might bring the ecological validity intoquestion.Population validityIf research has high population validity then thefindings can be said to relate to the general population.For instance, if a piece of research has usedan opportunity sample of 10,000 people chosenfrom 20 countries worldwide, its findings are morelikely to be relevant to the whole population of theExtract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

18 | A2 PSYCHOLOGY: THE STUDENT’S TEXTBOOKworld than if the sample had been three peopletaken from a cafe in Bristol. If we can say confidentlythat something has high population validity,then similar research carried out on a differentsample elsewhere should provide similar results.Test validityThe validity of a test can be measured in threeways.1. Content validityThe test being used is carefully scrutinised to seewhether it really does test what it says it tests. Thisis usually done by experts who assess the theorybehind the content to ensure that there are noomissions.2. Face validityThis is like content validity but far less rigorous.Face validity refers to how valid a test seems to be‘on the face of it’. To be more certain, a contentvalidity assessment should be made.3. Predictive validityThis assesses how well a score obtained on the testat one point in time might predict a score obtainedon the test at another time. For instance, if the testis carried out on a person when they are 20 yearsold, we might like to be able to predict their scoreon the test if carried out at 30 years old. A highpredictive validity would allow us to do this.Ethical considerations inpsychological researchIt is extremely important to design and carry outpsychological research ethically. You will havecovered ethical issues in research during your ASstudies. A summary of the major issues is providedin the table below.There is no reason why research should not becarried out ethically. In places where psychologicalresearch is conducted, work does not progressuntil a summary of the proposed investigationhas been considered by a committee drawn fromdifferent disciplines and walks of life who lookspecifically at any ethical issues. This ‘ethics committee’can agree to allow the research to go aheadas planned; it may opt to reject it until certain ethicalissues have been considered; or it may decideto reject the proposal entirely. The committee actsas a safeguard to stop questionable research fromtaking place.If ethically questionable research is conductedthe penalties for the researchers can be severe.The work is unlikely to find a publisher, and thepsychologists responsible may find themselveswithout access to further research funding: theymay even be expelled from the British PsychologicalSociety, the professional body for psychologistsworking in the UK. Other countries have theirown professional bodies who take a similar viewon unethical research.SUMMARY OF ETHICAL CONSIDERATIONS WHENDESIGNING A RESEARCH STUDYETHICAL ISSUEINFORMED CONSENTDECEPTIONDEBRIEFINGRIGHT TO WITHDRAWCONFIDENTIALITYPROTECTIONEXPLANATIONParticipants must be told what they will be doing andwhy they are doing it so they can provide ‘informed’consent.Participants should not be deceived unless absolutelynecessary. If deception is required, great care and carefulconsideration must be given to the project.After the experiment is complete, participants mustbe ‘debriefed’ and informed of the motivations for theexperiment. They must be given the chance to ask anyquestions they have.Participants should be free to leave the experiment atany time.Any information and data provided by the participantsmust be confidential.The safety and well-being of the participants must beprotected at all times.Extract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

DATA ANALYSIS AND REPORTING INVESTIGATIONS | 19DATA ANALYSISAND REPORTINGINVESTIGATIONSAppropriate selectionof graphsData can be presented as graphs, tables or statistics.Always remember that the point of presentingdata is to make your results easy to read for thoseinterested in your research.A graph should be simple and clear, and shouldbe the appropriate type of graph for the data youhave. A table should not be messy or complicated,or include too much information, and your statisticsshould be presented simply and as obviouslyas you can. Experience has shown us that somepeople, including professional researchers, do nottake nearly enough care when presenting their dataand it can really spoil what would otherwise havebeen an elegant and careful piece of work. Professionalresearchers will tell you that when lookingover the many hundreds of papers that are producedin their field each year, they first look at thetitle of the article, then the summary at the start,then the results section where the research data ispresented. If they find the presentation of the datainteresting they may spend time reading the wholearticle from start to finish. If they do not find thepresentation of the data clear then they may putthe article to one side, maybe not returning to it atall. It is for this reason that the importance of datapresentation should not be underestimated.There are four types of graph to know about. Wehave already discussed one of them – the scattergraph– when we talked about correlations.Each type of graph is appropriate for differenttypes of data. Do not draw all of them for all datasets, and do not make the mistake of just drawingyour favourite type of graph. It may well not beappropriate.It is important that you are familiarwith graphs. It is much more likelythat you will be required to read andinterpret the information contained ingraphs, rather than draw one. Make sure youpractice this skill.bar chartNumber of sightingsSIGHTINGS OF DIFFERENT GARDEN BIRDS IN ADOMESTIC GARDEN, SPRING 200735302520151050Sparrow Starling Blue-titCategory of garden birdThe bar chart is an extremely useful and commonway of presenting data. It is used to depict thenumber of incidences in a particular ‘discrete’category. This means that the categories identifieddo not overlap in any way. In our example we have‘types of bird’. You cannot have a starling that isalso a blue-tit. There is no overlap. You might alsouse this type of graph if you were depicting dataseparated by gender – for instance, the number ofboys enrolled on a football summer school and thenumber of girls. You cannot have a boy who is alsoa girl. The two categories do not overlap: they arediscrete. Bar charts have:»»Gaps between the bars»»Frequency (number) on the y-axis»»Category labels on the x-axis.HistogramNumber of push-upsNUMBER OF PUSH-UPS POSSIBLE AT DIFFERENT MASS35.031.528.024.521.017.514.010.57.03.5050 100 150 200Mass (kg)The histogram is similar in many ways to the barchart. The major difference is that in a histogramthe x-axis does not depict discrete categories,rather it shows a continuous scale. In our examplethat scale is mass (kg). If, on the other hand, yourdata were concerned with temperature and fatigue,you would measure ‘degrees Celsius’ on the x-axis.Histograms have:»»No gaps between the bars»»A continuous variable on the x-axis.Extract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

20 | A2 PSYCHOLOGY: THE STUDENT’S TEXTBOOKFrequency polygonNumber of push-upsNUMBER OF PUSH-UPS POSSIBLE AT DIFFERENT MASS35.031.528.024.521.017.514.010.57.03.5050 100 150 200Mass (kg)The frequency polygon is very similar to the histogram.Imagine a histogram with a point drawnin the centre of the top of each bar. Now, removeall the bars and join up the points. A frequencypolygon has:»»A continuous variable on the x-axis»»A continuous line, no bars.TablesThe whole point of a table is to present otherwisecomplicated information in as simple a way aspossible. A table should allow readers to find theinformation they need with as little effort as possible.Tables can be large, containing lots of data,or they may be small, containing data in the formof summary statistics, such as mean, mode andmedian. In the case below, we have a table showingthe mean scores of male and female studentsin maths, English and science examinations.MathsEnglishScienceMale564351Female446752When drawing a table make sure you follow themost important rule: be as clear and straightforwardas you can.Which is best?Each measure of central tendency and dispersion has its good and bad points. A strong descriptive statistic isgenerally one that takes in a good deal of the raw data in its calculation, so bear that in mind when thinkingabout which is most suitable. The strengths and weaknesses of each are described below.StrengthWeaknessMeanMost powerfulmeasure ofcentral tendencyas it usesall of the dataOne roguescore (largeor small) canheavily influenceit. Forinstance, themean of 3, 4and 8 is 5. Themean of 3, 4,8 and 1,005is 255. Theextreme valuehas seriouslyinfluenced themeanModeThe bestmeasure to useif you wantto know howoften thingshappenSometimes adata set doesnot have amost commonvalue andsometimesit has lots ofcommon valuesMedianNot heavilyinfluenced byrogue scoresNot good forusing withsmall data sets.For instance, ifyou only havethe numbers 1,17 and 2,000in your dataset the medianis 17. Not veryinformativeStandarddeviationUses everyvalue in thedata set,not heavilydistorted byextreme valuesand is the mostsensitiveThe mostlaborious ofthe measuresof centraltendency tocalculateRangeTakes extremescores intoconsiderationand is simple tocalculateIf either of thetwo scores areextreme, rangewill be distorted.It tellsus little abouthow spread outor clusteredtogether thedata areSemiinterquartilerangeLess distortedscores than therangeUses only 50%of the data inthe calculationand is quitelaborious tocalculateIt’s not really useful to ask which of the measures is best and which is worst. Each is appropriate in differentcircumstances, depending on the data you have and what you are looking for in your research.Extract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

DATA ANALYSIS AND REPORTING INVESTIGATIONS | 21Descriptive statisticsDescriptive statistics are ways of representingraw data simply. Measures of central tendencyand measures of dispersion are the two types ofdescriptive statistic with which you are familiarfrom your AS studies.A measure of central tendency is sometimesreferred to as an average. An average is the scorethat is typical of all the data you have in your set.Measures of dispersion tell us how spread out thedata are. The larger the measure of dispersion, themore spread out the data.Each measure of central tendency (mean, medianand mode) has an appropriate measure of dispersion:this is easily described in a table.“when using a ….Mean ….Median ….Mode ….use a ….”Standard deviationSemi-interquartile range,or rangeRangeCalculating these values is not very difficult: it justtakes a little time. This is how you make the calculations.It is worth refreshing your memory.Mean1. Add up the numbers in your data set (call thisnumber A).2. Count up the number of values in your dataset (call this number B).3. Divide A by B.Mode1. Identify the most common value in your dataset. This is the mode.2. If there are two equally common numbers youhave a ‘bi-modal’ set (2 modes).3. If you have three equally common numbersyou have a ‘tri-modal’ set (3 modes).Median1. Put your data in order (smallest to largest).2. The data value in the middle is the median.3. If you have an even number of values, take themean of the two in the middle.Standard deviation1. Calculate the mean of the values in your dataset.2. Subtract the mean from each value in turn andsquare the result in each case.3. Add up all of the squared values from step 2.Mathematical symbols – a handy guideThe problem with maths is that it sometimes uses symbols that make it look more like Ancient Greek thansomething we might possibly understand. It’s not that complicated actually. Here’s a handy guide to whateach symbol means.Symbol√∑Nxx2s=~Meaningsquare rootsigma. This just means ‘the sum of’the number of items in your data setthe value you are working withx-bar. This just means ‘the mean of your data set’multiply by itself. Also known as ‘squared’.lower case sigma. This means standard deviationequals. What’s on the left of the symbol equals what is on the rightless than. What’s on the left is less than what’s on the rightgreater than. What’s on the left is greater than what’s on the righta lot less than. What’s on the left is a great deal less than what’s on the righta lot greater than. What’s on the left is a lot greater than what’s on the rightapproximately. What’s on the left is approximately the same as what’s on the rightExtract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

22 | A2 PSYCHOLOGY: THE STUDENT’S TEXTBOOK4. Divide the result from step 3 by the number ofvalues in your data set.5. Take the square root of the figure you calculatedin step 4.Range1. Find the largest and smallest values in yourdata set.2. Take the smallest from the largest.Semi-interquartile range1. Put your values in order (smallest to largest).2. Count up how many values you have. Call this‘N’.3. Calculate N+1 and divide by 4. Call this ‘L’.4. In your ordered data set, find the number atposition L.5. Calculate N+1, multiply it by 3, then divide itall by 4. Call this ‘U’.6. In your ordered data set, find the number atposition U.7. Take the value in step 4 from the value in step6.8. Divide the result of step 7 by 2.When you take each step at a time, the mathematicsare not too difficult at all. It’s important to beorganised when making these calculations. Makesure your working-out paper does not becomemuddled, and that you keep track of the values. Ifyou do that then life will be much simpler.EquationsBefore writing this we asked as many students as we could what they would find useful, and what they wouldlike, when learning about research methods and statistics. 82% of them told us that they would like to avoidequations at all costs, and 87% said that learning how to use equations would be extremely useful. It is clearthen that most students hate equations and at the same time most of them think that it’s a good idea thatthey know how to use them. How right they are!∑ (x – x) 2NThis is the equation for working out standard deviation. The trick with these things is to understand whateach part means, and to do one thing at a time. Let’s split it up into its parts and then work out how to use it.Armed with your knowledge of the more common symbols used in mathematics, let’s take a look at the equationagain. Look at the line right in the middle for the moment:∑ (x – x) 21. Calculate the mean value (x) of your data set.2. Take your value (x) and subtract from it the mean of the data set (x).3. Multiply what you get in step 2 by itself.4. Do this for all of the values in your data set in turn, writing the result down each time. If you have 20values then you will end up with 20 numbers.5. Now add up (∑) all the numbers you calculated in step 4.You may get a negative number in step 2, but that doesn’t matter at all, because once you’ve squared it (multipliedit by itself) you always have a positive number, much easier to deal with. Tedious and some might evensay very boring, but not terribly complicated so far. The rest is even simpler.6. Take the total you get in step 5 and divide it by the number of values in your data set (N).7. Put this number into your calculator, find the square root button marked √ and push it. The result you getis s, the standard deviation for your data set.If you do one thing at a time, equations are not very complicated at all. The best thing to do is just practisea little, take your time and be organised. Get to know what each part of the equation means. It’s a bit likelearning another language, but there’s not too much to know. Each equation uses pretty much the sameinformation in a slightly different order. If you know the basics, any equation is child’s play. Oh, and if you feela little daunted by this type of thing, so do 82% of our students: and some of them are at university – you arenot alone.Extract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

DATA ANALYSIS AND REPORTING INVESTIGATIONS | 23Probability and significanceWe’ve already let you in on a few secrets aboutstatistics. Here’s another. All they are for is forshowing that your result didn’t occur by chance.That’s pretty much it. Don’t let anyone fool youinto thinking they are any more complicated thanthat. When we do some statistics all they give usafter we’ve finished adding, multiplying, dividingand things is a single important number, which werefer to as ‘p’. It stands for ‘probability’.The p value gives us an idea of how likely it isthat our results happened by chance or not. Thep value can be anywhere between 0 and 1. A p of1 means that something is definitely, absolutelygoing to happen. For instance, there is a probabilityof 1, an absolute certainty, that in the UKChristmas Day will fall on December the 25th. Ap of 0 means that something is never ever goingto happen. For instance, there is a probability of 0that this book will turn into a badger.The nearer to 1 your value of p is, the more likelyit is that the results happened by chance. Thesmaller the value of p, the more likely it is thatwe can confidently accept our hypothesis, andreject our null hypothesis. A p of 1 means that weare 100% certain that something will happen bychance. A p of 0.9 means that we are 90% certainthat something will happen by chance. A p of 0.8means we are 80% certain that something willhappen by chance and so on. The smaller the pthe less likely it is that something will happen bychance. Remember:Small p – GoodBig p – BadIf we can reject our null hypothesis we can say thatour results were ‘significant’. By this we mean thatour findings did not occur by chance. Statistics letus measure how ‘significant’ our results are. Statisticaltests are used to analyse the data gatheredin research to tell us the p value (we will describethese tests in more detail shortly). The smallerthe p value, the more significant the result. Forexample, the statistical analysis might tell us thatwe have a p value of 0.02. This is the significancelevel. When we report ‘p’ in research papers andbooks it is more often than not written like this:p

24 | A2 PSYCHOLOGY: THE STUDENT’S TEXTBOOKsuch a case. You might employ an alpha like this,for instance, if you had been investigating howwell an experimental drug influenced the memoryof elderly people with Alzheimer’s disease. Inorder to put your drug onto the market and allowpeople to start taking it, you need to be very surethat it does what it is supposed to.If the p value you find does not fall at or belowyour chosen alpha, then you are unable to acceptyour hypothesis. Instead you are unable to rejectthe null hypothesis and you cannot say that yourresults were significant. In these circumstances,you have to say that your results were non-significant.This is quite different from insignificant bythe way! Your findings were not at all insignificant.Even a result that does not support the hypothesisis interesting to scientists, but statistically speaking,it is a non-significant result.InvestigationHypothesis andnull hypothesisAlphaCalculatedp valueCorrect decisionIncorrectdecisionDoes eatingcrisps make youfeel sick?Hypothesis: Themore crisps you eatthe sicker you feel.Null hypothesis:You do not feelsicker by eatingmore crisps.0.05p≤0.02(p is less than orequal to 0.02)The p value is lessthan alpha.We can reject thenull hypothesisand say that theresult supports ourhypothesis.We can confidentlysat that we are95% certain thatour results did notoccur by chance.Our results aresignificant.A type 2 error:We cannot rejectthe null hypothesisand must reject thehypothesis.The implicationsof an error likethis may result inpeople eating loadsof crisps becausethey believe it willnot make them feelsick when in factit does.A type 1 error:We can rejectthe null hypothesisand accept thehypothesis.Does wearingperfume makefemales moreattractive tomales?Hypothesis: Wearingperfume makesfemales moreattractive to males.Null hypothesis:Wearing perfumedoes not makefemales moreattractive to males.0.05p≤0.09(p is less than orequal to 0.09)The p value is largerthan alpha.We cannot acceptour hypothesis andmust retain our nullhypothesis.We can onlysay with 91%confidence thatour results did notoccur by chance.Our results arenon-significant.The implicationsof this error maybe serious for theperfume industry.Sales of perfumemay drop becausewomen may feelthat wearing itmakes no differenceto their perceivedattractiveness.Should weprescribe apotentially dangerousexperimentaldrugto those withschizophrenia?Hypothesis:Treatment withthe experimentaldrug improves thequality of life ofthose suffering withschizophrenia.Null hypothesis:Treatment with theexperimental drugdoes not improvethe quality of life ofthose suffering withschizophrenia.0.01p≤0.017(p is less than orequal to 0.017)The p value is lessthan alpha.We can reject thenull hypothesisand say that theresult supports ourhypothesis.We can coincidentallysay that weare 99% certainthat our resultsdid not occur bychance.Our results aresignificant at the0.01 level.A type 2 error:We cannot rejectthe null hypothesisand must reject thehypothesis.The implicationsof this error arepotentially veryserious. Millionswho suffer fromschizophrenia willnot benefit fromthis wonderful newdrug.Extract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

DATA ANALYSIS AND REPORTING INVESTIGATIONS | 25Type 1 and Type 2 errorsAfter you have carried out some statistical tests,you must decide whether you accept or reject thenull hypothesis. There are two major errors thatcan be made when doing this. These are describedrather confusingly as type 1 and type 2 errors.Type 1 errorThis is also known as the ‘false positive’ error. It isthe mistake of rejecting the null hypothesis whenit is actually true. You have made the decision toaccept your hypothesis by mistake. A very strictalpha value makes this kind of mistake less likely.Type 2 errorThis is also known as the ‘false negative’ error.This is the mistake of accepting the null hypothesiswhen it is in fact false, and you should haverejected it in favour of your hypothesis. A verystrict alpha value makes this kind of mistake morelikely.Do not underestimate the importanceof learning about probability, significance,and type1/type 2 errors! Theyare a crucial aspect of data analysis andyou can bet that they will feature frequentlyin exam questions!Choosing the correctstatistical testThe choice of your test is influenced by the type ofresearch you have done, and the type of measurementsyou have made.Levels of measurementThere are three types of measurements you canmake in psychological research. These are nominal,ordinal and interval. The best way to describeeach is with an example.Nominal levelIf you have nominal data you have data that canbe classified in categories. By this, we mean thatif something is in one category it cannot be inanother category also. For instance, if you arecounting up the number of men and women ata rugby match, you cannot have someone whocounts as a man and also as a woman. Similarly, ifyou make a trip to a safari park to carry out a surveyon animals you may want to count the numberof monkeys you see and the number of hippos.You cannot have a monkey that is also a hippo –they exist as discrete categories. If your data is likethis, it is described as nominal level data.Ordinal levelThe clue for this one is in the name. Ordinalsuggests that there is an order. Horse racing is agood example. Horses are recorded as finishingfirst, second, third, fourth and so on. The order inwhich they finish is the important thing, not thedistance between them. If your data is like this, insome kind of order or rank, then it is described asordinal level data.Interval levelIf you are measuring something on a scale,perhaps the height of something or the time ittakes someone to do something, then you areusing an interval scale. Time, temperature, weightand height are all examples of interval levels ofmeasurement.A knowledge of your experimental design andthe level of measurement used will allow you toanswer three simple questions, which will lead youto the appropriate statistical test.Example 1Our research is concerned with the relationshipbetween a person’s level of sadness and theamount of chocolate they eat. Here we are measuringsadness ratings and volume of chocolateconsumed in 10 different people.Question 1: Do I have correlation data?The answer to this is YES. You are looking for arelationship between variables and so your choiceof test is the Spearman’s Rho.Example 2Our research is concerned with investigatingwhether males are better at science subjects thanfemales. We collect science test scores from 10males and 10 females and want to compare them.Question 1: Do I have correlation data?The answer here is NO. You are looking atdifferences.Question 2: Am I looking at numbers incategories?The answer is NO. You are looking at differences.Question 3: What type of design did I use?The answer to this is an independent samplesdesign. You have two groups, one male and onefemale. A comparison is made of the scores of differentparticipants in the different conditions. Thetest for you is the Mann-Whitney Test.Extract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

26 | A2 PSYCHOLOGY: THE STUDENT’S TEXTBOOK1. Do I have nominal level data?YesNoChi-square2. Is my research about the relationship between variables?YesNoSpearman’s Rho3. What type of design did I use?Repeated measuresIndependent samplesExample 3Our research is investigating whether memory isbetter after a high-protein meal than before. Eachparticipant carries out a memory task before a highprotein meal of fish and chicken, and again afterthe meal.Question 1: Do I have correlation data?The answer is NO. You are looking at differences.Question 2: Am I looking at numbers incategories?The answer is NO. You are looking at differences.Question 3: What type of design did I use?The answer to this is a repeated measures design.Each participant provided information before themeal and also after the meal. A comparison ismade of the same participant’s scores in the twoconditions. The test for you is a Wilcoxon test.Example 4Researchers are investigating whether psychologystudents and maths students revise differently andwhether this influences test scores. The two methodsused are cramming information or organisedrevision. Students can be either psychology ormaths students (not both) and they may be organisedlearners or crammers (not both).Question 1: Do I have correlation data?The answer is NO. You are looking at categories ofbehaviour.Question 2: Am I looking at numbers incategories?Yes you are. You are looking at whether test score isinfluenced by the type of learning and the subject.Each person cannot be in more than one category.They are either psychologists or maths students,and they are either crammers or organised learners.The test for you is a chi-square test.WilcoxonInferential analysisMann-WhitneyData analysis:teachers LOOK AWAY NOWLet’s be absolutely honest about this: most peopleare not terribly fond of mathematics. In psychology,we spend quite a lot of time using numbersand talking about statistics. Why? Because weneed to know whether our findings allow us toconclude anything about our hypotheses. Becausestatistics are seen as extremely complicated theytend to get more than their fair share of coveragein books and in lessons. This is because peoplefind them hard to understand and so authors andteachers spend a great deal of time explainingthem.It may come as a shock to you, but most psychologistsworking professionally are also not very fondof statistics. Most of them, however, have realisedthree very important secrets that we are about tolet you in on here. Don’t tell anyone though.1. Statistics are just a tool.It’s as simple as that really. Statistics are the endpoint of your research. Nearly all of your time as aresearcher is taken up deciding what you want todo, how you are going to do it and actually doingit. The statistics come right at the end. They areonly a tool for finding out which conclusions youcan make from your data. Just as a hammer is atool for making nails go into walls and an iron is atool for making clothes flat, statistics are tools tolet us weigh up our findings.The second secret you need to know about statisticsis a pretty controversial one. We include ithere because we have found it very useful in ourlearning over the years.2. Statistics ARE often hard to understand.There, we’ve said it. There are two reasons forfinding things hard to understand. The first isthat you think you are not very smart, and soExtract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

DATA ANALYSIS AND REPORTING INVESTIGATIONS | 27if something is tricky, then it’s your fault. Thesecond, just as valid, reason that something seemshard is that it IS hard. Anyone who tells you thatstatistics are always simple would be lying to you.They do require some thought. However, thereason most people find them hard is neatly sidesteppedwhen we let you in on secret number 3.3. You do not have to understand it all.At this point, teachers all over the country arefainting with the realisation that this, the mostcontroversial of their secrets, has been told totheir students. Here’s an example of what wemean: stick with us here, it’s worth it. Sometimespsychologists may use equations when workingsomething out. In these equations they may needto divide by things, multiply numbers or add alot of numbers up. Here’s where secret number 3comes in.The reason many students setting out to learnhow to use statistics in psychology find it hard isbecause they are naturally thoughtful and inquisitivepeople. They are used to asking questions like“WHY am I doing this?” and “WHY do I need todivide by this number?” The problem is that theanswers to these questions are often much morecomplicated that you might expect, and so the studentpsychologist can become bogged down andsometimes confused. They learn that statistics canbe hard and confusing and then everything thatinvolves them is to be hated and avoided.Now, rewind a little. What might have happened ifthe student psychologist had not asked the WHYquestions? If they had not needed to know whysomething needed to be divided by somethingelse? In fact, in many cases, most psychologistsdo not know why certain things are done whenapplying the tests, they just do them. Here’s themost controversial tip of all. Just for the moment,and just where following instructions on statisticsare concerned, switch off the bit of your brainthat makes you ask ‘Why?’ Just do it, in the firmknowledge that it works. Remember, statisticsare a tool, just like any other. You do not need tounderstand the physics that go along with the actof hammering a nail into a wall do you? You do notconsult textbooks to understand how the thermostatworks in an iron when smoothing your clothesdo you? Then don’t worry about why we divideby something, or why you are told to multiply twonumbers together.An often overused but really good example is cakemaking. Following a tried and tested recipe onestep at a time will result in a good cake. Changingthe recipe might improve the cake, but it maywell ruin it, so it’s safer to stick to the recipe as italways gives successful results. Cooking is basicallychemistry with food. When you mix thingstogether and apply heat you change the structureof the ingredients, turning them magically intowonderful cakey goodness. You do not need to bean expert in molecular chemistry to know howto make a cake. You do not need to know howmolecules are broken and re-formed during thecooking process and why these molecules areeasy to digest. You do not ask ‘why’ questions incake making, so there is no need to ask them instatistics.We have been as careful as we can to set out thestatistics you need to know in a form that is easyto follow and reproduce. Where appropriate we’lltell you what’s happening and why, but the ruleis, if you follow each step carefully you will beprovided with the information you need in yourresearch.Using inferential statisticsDescriptive statistics allow us to describe the datawe are using. They include presenting data in theform of graphs and tables or summarising themas measures of central tendency or dispersion. Inthis section we will describe how to carry out fourdifferent statistical tests that allow us to infer somethingabout our data. By that we mean that theyallow us to deduce or conclude something aboutthe research we have carried out, not just describethe data. Inferential statistics are often usedright at the end of the research process to allowresearchers to conclude things about whether theirresults have occurred by chance or not. Remember,the tests are only tools to allow you to arrive ata p value for your research. The various tests mayseem a little confusing at first, but stick with it,and they will become much easier to handle.You are required to know about the useof four inferential tests: Spearman’sRho, Wilcoxon, Mann-Whitney and chisquare.Don’t learn every step of eachtest – instead, develop an appreciation ofthe logic of inferential analysis. Understand why youwould use a particular test, how the score from eachtest leads to a p value and how this is interpreted forsignificance. A good understanding of these thingswill serve you well in the exam.Extract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

28 | A2 PSYCHOLOGY: THE STUDENT’S TEXTBOOKSpearman’s RhoThis test is used in correlation research. If you arelooking at the relationship between two variablesand have drawn a scattergraph, then Spearman’sRho is the test to use. We’ll start by summarisingsome research involving correlation and using thereal data in our calculation.Aim:To investigate the relationship between ability atmathematics and ability at playing chess.Hypothesis:The better you are at maths, the better you will beat chess.Null hypothesis:Being good at maths does not mean that you willbe good at chess.Operationalising the variables:Maths ability was operationalised as ‘score out of100 on a maths test’.Chess ability was operationalised as ‘number ofchess problems out of 100 successfully solved’.The raw data:12345678910Maths Score58347759469022788967Chess Score67437261509920739070The graph:This is correlational research investigating therelationship between one variable and another. Theappropriate graph to draw is a scattergram.The relationship between maths and chess ability1008060402000 20 40 60 80 100Chess scoreMaths scoreThe descriptive statistics:MeanStandarddeviationMaths Score6222.81Chess Score64.522.78It seems from these data that there is indeed astrong positive correlation between maths abilityand chess ability. This is shown most clearly inthe graph. As maths score increases so too doeschess score, typical of a positive correlation. To bemore certain, and to see whether the correlationis significant, we need to carry out the Spearman’sRho statistical test.Doing Spearman’s RhoStep 1: Rank the scores for each variableRanking will become extremely straightforwardas you have to do it for three of the tests you arerequired to understand. It is simple enough, buttake care to get it right. Take your original tableand add two more columns. Ranking meansputting the scores in order, smallest to largest, andgiving the smallest the rank of 1, the next smallestthe rank of 2 and so on. You can see how this isdone by looking at the ranks for maths scores inthe table below. Notice, the smallest maths scorewas 22. This gets the rank of 1. The next smallestwas 34. This gets the rank of 2 and so on, right upto the highest score of 90, which gets the rank of10.12345678910MathsScore58347759469022788967ChessScore67437261509920739070MathsRank42753101896ChessRank52743101896The ranking of our data is uncomplicated. In othercircumstances, though, you may find that twoparticipants score the same on the same task. Thiscommon problem makes the ranking slightly moredifficult. Take a look at the section on tied ranksbelow for more on this.Extract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

DATA ANALYSIS AND REPORTING INVESTIGATIONS | 29Step 2: Work out differences and square themIn our case we need to work out the differencebetween maths score and chess score for each ofour participants. To start you off:Participant number 1 was ranked at 4 for mathsand 5 for chess.4–5 = –1–12 = 1For each participant, the ‘difference’ number, andthen the figure which is the result of squaring thatnumber, are recorded in two new columns, to theright of those shown in the table above. Whenyou’ve done this for all your participants youshould end up with a table like this:12345678910MathsScore58347759469022788967ChessScore67437261509920739070MathsRankChessRankd(difference)d 2(d squared)4 5 -1 12 2 0 07 7 0 05 4 1 13 3 0 010 10 0 01 1 0 08 8 0 09 9 0 06 6 0 0Step 3: Add up the numbers in the d 2 column(∑d 2 – ‘sigma d 2 ’ in English)You need this number for working out the equation.In our case all we do here is add up thenumbers in the far right column:1+0+0+1+0+0+0+0+0+0 = 2Step 4: Identify N and N 2N refers to the number of pairs you have in yourdata. In this case we have 10. N 2 is simply thatnumber squared: in your case that’s (10 x 10)which is 100.Step 5: Get organisedGet all the numbers you need for your equation.r6(∑d 2 )s=1–N(N 2 –1)r s– this is the statistic we are after – Spearman’sRho.N – the number of people who provided data inyour correlation research.d – The difference between each rank of onevariable and the corresponding rank of the othervariable.Your equation requires ∑d 2 , N 2 and N. Make sureyou know what these are and write them downon the top of the paper on which you will do yourcalculations.∑d 2 = 2N = 10N 2 = 100Step 6: Insert your numbers into the equationWith your numbers it looks like this:6(2)r s=1–10(100–1)This looks a little less frightening and gets evensimpler. The first rule is to work out everythingwithin the brackets first. When you do that youget the following:6(2)r s=1–10(99)You’ll know this from maths, but it won’t hurt toremind you that you need to multiply the numberinside the brackets by the number on the outside:i.e. you need to calculate 6 x 2 and 10 x 99. Whenyou do that, your equation looks like this:12r s=1–990Next, work out the fraction. Remember, youdivide the top number by the bottom number,(in other words, 12÷990) to give you the statisticwe need which will indicate the strength of yourcorrelation.r s=1– 0.012= 0.988This is your correlation coefficient. You will recallthat anything positive means that as one variableincreases so too does the other (the absence of asign in front of the number means that it is positive).You will also know that the nearer to 1 yourcoefficient is, the stronger the correlation. It isclear that we have an extremely strong correlationhere, but is it significant? Might it have occurredby chance?Step 7: Finding the critical valueTo see whether the calculated r sis large enoughfor you to say that your results did not occur bychance you need to look in statistical tables. At theend of this book you will find an appendix withthe table you need for this test. To use it, you needto know:N The number of participants in your task: 10Extract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

30 | A2 PSYCHOLOGY: THE STUDENT’S TEXTBOOKr sThe value of Spearman’s Rho that you calculated:0.988The type of hypothesisWhether you chose a directional (one-tailed)or non-directional (two-tailed) hypothesis. Inthis case our hypothesis was directional, as wepredicted that better maths scores would go alongwith better chess scores.Your level of alpha (∝)The level of significance (alpha) you are using.Remember, this is usually 0.05, but may in othercases be much lower.What you need to do now is compare the calculatednumber (sometimes called the observedvalue) to a number in an appropriate statisticaltable (sometimes referred to as the critical value).You look up the relevant number in the table andsee if your observed value is equal to or larger thanthe critical value from the table.The full table can be found in the appendix, butfor simplicity, the section from the table that refersto your research is as follows:Levels of significance for a one-tailed testN 0.05 0.025 0.01 0.00510 0.564 0.648 0.745 0.794What we can see is that our calculated level of0.988 is much larger than the level required for asignificance (p) level of 0.05, or any other significancelevel for that matter! We can clearly be 95%certain that our results did not occur by chance.We can definitely reject our null hypothesis andsay that our results provided clear support forthe hypothesis that there is a strong relationshipbetween maths and chess ability.Remember, just follow the logic ofthese tests, don’t try to rememberhow to do them. You start researchwith a hypothesis and the inferentialanalysis allows you to see whether or not itis supported by the data you have gathered in yourresearch. If the analysis says that your findingsmight be due to chance then you have supportedthe null instead of your hypothesis!Dealing with tied ranksTake a look at the following data. The ranking hasalready been done for you.12345678910SpellingScore3087695341MemoryScore7634408124SpellingRank3.519871063.552MemoryRank98466110236Look at the first column where spelling scoresare indicated. Participants 1 and 8 both scored3 on this test. Now look at the ‘Spelling Rank’column. When ranking a column of scores likethis, each tied score gets an average of the rankpositions that they would normally be in. In thiscase, the scores would be in rank positions 3 and4. The mean rank value for these two positions is(3+4)/2 = 3.5.Look now at the ‘Memory Score’ column. Noticethat three participants scored 4 on this test. Thesescores would occupy rank positions 5, 6 and 7and so each receives the mean rank of (5+6+7)/3= 6.The rest of the test progresses in exactly the sameway as usual. The only slightly tricky thing here ismaking sure that you get the ranking right.The Wilcoxon TestThe Wilcoxon is used when you are looking fordifferences, rather than a relationship, with arepeated measures design. This means that thesame participant provides data for the differentconditions in your research. We can use a simpleexample of comparing memory in silence andmemory with music. In the experiment we couldsee how people perform on a memory test insilence and how they perform on the memory testwhile listening to music. The test can be as simpleas remembering a list of numbers.Aim:To investigate whether memory is influenced bymusic.Hypothesis:Memory for numbers will be worse while listeningto music than in silence.Null hypothesis:Memory for numbers will NOT be worse whilelistening to music than when in silenceExtract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

DATA ANALYSIS AND REPORTING INVESTIGATIONS | 31Here are some results.Mean memory score807468625650MeanStandarddeviationSilenceMemory scoresSilence72.5515.15MusicMusic64.5516.36It seems that silence does indeed provide betterconditions for memory than music. Unfortunatelywe cannot yet reject our null hypothesis andaccept our hypothesis. We cannot be sure that ourresults did not occur by chance and so we need totest the significance of our findings.Doing the Wilcoxon TestStep 1: Take the scores on one condition fromthe scores on the other conditionIn this case, we simply subtract the memory scorewhile listening to music from the memory scorein silence. In this test we must be careful to markdown whether the difference is positive (+) ornegative (–).Step 2: Rank the differencesHere we rank the differences, not the scoresthemselves. There are two rules to remember forranking in this test.1. Ignore all the ties when ranking2. Ignore the sign when rankingOnce you’ve done this you should end up with atable that looks like this:SilenceMusicd (difference)Ranks of dSign of therankNegativeranksPositive ranks17680-44–427667+910.5+10.535554+11+148978+1114+1457665+910.5+10.568781+67+776764+32+283427+78.5+8.596745+2219+19106551+1417.5+17.5115955+44+4129187+44+41399990(tie)..145651+56+6156855+1315.5+15.5168976+1315.5+15.5177864+1417.5+17.5187467+78.5+8.5196555+1012.5+12.5208070+1012.5+12.5Extract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

32 | A2 PSYCHOLOGY: THE STUDENT’S TEXTBOOKNotice that in our data there is one tied score, andonly one negative difference, ranked at positionnumber 4.Step 3: Add up the negative and positive ranks,and choose the smaller of the numbersIn our data, the total sum of the negative ranks is4. The total sum of the positive ranks is a lot more,and comes to 186. The number we want is thesmaller of these, 4. This is your Wilcoxon statistic,referred to as T. It is your observed value.Step 4: Finding the critical valueYou must now compare the observed value to thecritical value in the relevant table. It is included inan appendix at the back of this book. To do this,you need to know:N The number of participants in your task: 20T The observed value you calculated: in this caseit is 4.The type of hypothesisWhether you chose a directional (one- tailed) ornon-directional (two-tailed) hypothesis.Your level of alpha (∝)The level of significance (alpha) you are using.Remember, this is usually 0.05, but may in othercases be much lower.The first thing to do is look down the first columnfor your value of N. Next, read across to therelevant column that refers to your level of alpha.Here you will find the number you are looking for.For simplicity, only the relevant line in the table isreproduced here.Levels of significance for a one-tailed testN 0.05 0.025 0.01 0.00520 60 52 43 26If your value is smaller than the value in the table youcan say with confidence that your results did notoccur by chance. Here, the critical value from thetable is 60, for the significance level of 0.05. Yourvalue (4) is definitely smaller than this.This means that you can say that your results aresignificant. We report this as follows: ‘The differencebetween memory in silence and when listeningto music is significant (p

DATA ANALYSIS AND REPORTING INVESTIGATIONS | 33The descriptive statistics:Mean errorsStandarddeviationMen10.383.74Women6.93.07From the graph and the descriptive statistics itlooks very much as if women make fewer errorswhen doing more than one task at once than domen. We’ll need to do some statistics, in this casea Mann-Whitney, to assess the likelihood thatthese results have occurred by chance.Doing the Mann-Whitney TestStep 1: Rank the dataIn the Mann-Whitney, we rank all the datatogether. By this we mean that the men’s scoresand the women’s scores are all ranked at the sametime. We’ve done this for you here. Notice theusual problem of tied ranks. If you are not sureabout this, look back to the section above.Men(N = 8)1241711812109Rank16318147.51612.510Women(N = 10)Rank9 10584109453125.57.5312.51035.5Step 2: Add up the ranks in each group, callingthem Rank A (R A) and Rank B (R B)Sum of ranks for men (R A)16+3+18+14+7.5+16+12.5+10 = 97Sum of ranks for women (R B)10+5.5+7.5+3+12.5+10+3+5.5+1+16 =74Step 3: Use your values to calculate the resultsof two formulaeFor the Mann-Whitney, we need to calculate twovalues, referred to as U Aand U B. Here are theformulae. They are very similar indeed, so be116careful you are using the right numbers when youcalculate them!1.2.U A=N AN B+U B=N AN B+N A(N A+1)2N B(N B+1)2–R A–R BThese look a little frightening at first, but let’s takeone at a time and see what they look like when weinsert our numbers. As usual, begin by identifyingthe numbers you need and organising yourselfcarefully.U A– the number we are calculatingwith the formulaU B– the number we are calculatingwith the formulaN A= the number in Group A (men) = 8N B= the number in Group B (women) = 10R A= the sum of ranks in Group A (men) = 97R B= the sum of ranks in Group B (women) = 741.U A=N AN B+N A(N A+1)2–R AInserting our values, the formula now looks likethis:U8 (8+1)A=8x10+ –972A little basic mathematics can simplify it furtherto:U8 (9)A=80+ –972This is equal to:U72A=80+ –972which equals:so finally2.U B=N AN B+U A=80+36–97U A=19N B(N B+1)2–R BInserting our values, the formula now looks likethis:10 (10+1)U B=8X10+ –742Extract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

34 | A2 PSYCHOLOGY: THE STUDENT’S TEXTBOOKAgain, some straightforward mathematics cansimplify it further to:10 (11)U B=80+ –742This is equal to:U110B=80+ –742which equals:so finallyU B=80+55–74U B=61Step 4: Select the smallest value of U and consultthe tables for a critical valueIn our case, the smallest value of U calculated wasU A: this is our observed value at 19. We now needto consult the table of critical values for the Mann-Whitney. To do this we need to know the numberof participants in each group. We had eight malesand ten females, so we read across the top of thetable until we reach 8, and we then read down tothe N=10 line.A portion of the table is reproduced for you here,but make sure you refer to the real table in theappendix of this book to be sure you know how toread it.N17891011N27 8 9 10 114 6 7 9 106 7 9 11 137 9 11 13 169 11 13 16 1810 13 16 18 21You can see that the critical value of U for N of8 and N of 10 is 11. If your value of U is equalto or less than the critical value then you can sayconfidently, with 95% certainty, that your resultsdid not happen by chance.Our value of U was 19. This is not smaller than11, so we cannot reject our null hypothesis. Eventhough the graph and the descriptive statisticssuggested that it might be the case, we cannot saywith confidence that women are indeed better atdoing more than one task at once. This is a verygood example of why we should always do thestatistics! If we had not done so, we might havemade the error of rejecting our null hypothesisby mistake, and accepting our hypothesis eventhough the difference could well have happenedby chance.Chi-square (c 2 )Chi-square test (pronounced ‘ky square’) is oftenwritten as c 2 . We use this test when we have numbersof people in different categories. Here’s anexample from earlier, to help clarify what we mean:‘Researchers are investigating whether psychologystudents and maths students revise differently andwhether this influences test scores. The two methodsused are cramming information or organisedrevision. Students can be either psychology ormaths students (not both) and they may be organisedlearners or crammers (not both).’In this case we have a group of psychology studentsand a group of maths students. The rule isthat we cannot have a student studying both. Eachof our subject groups is split into two according tothe revision style of its members: those that cramfor exams (crammers) and those that revise in anorganised way (organised).To summarise, students can be in one of four completelyseparate categories. These are as follows.1. Psychology: Crammers2. Psychology: Organised3. Maths: Crammers4. Maths: OrganisedWhat’s important here is that no one can be inmore than one group. The Chi-square test looks atwhether the number of students in each categoryis different from what we might expect just bychance. In this way the test tells us whether thenull hypothesis (that something happened bychance) can be rejected or not, thus allowing us tomake a decision about our hypothesis.Aim:We are interested in whether the type of revision(cramming or organised) works differently for differentsubjects.Hypothesis:There will be a difference in the effectivenessof different types of revision depending on thesubject.Null hypothesis:There will not be a difference in the effectivenessof different types of revision depending on thesubject.Operationalising the variables:A self-report questionnaire of psychology andmaths students asks them to identify the type ofrevising they do, cramming or organised revision.Extract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

DATA ANALYSIS AND REPORTING INVESTIGATIONS | 35The raw data:CrammersOrganisedPsychology535MathsEach square relating to each of the four categoriesis referred to as a ‘cell’. Each ‘cell’ contains anumber which relates to the number of studentsin that category and is referred to as the observedfrequency.The graph:We have two different groups. The appropriategraph to draw here is a simple bar chart.The question is, are the numbers of students inthese categories different from what we mightexpect to see just by chance The way to find out isto carry out a Chi-square test.Frequency (number of students)604530150PsychologyCrammersRevising styles: Maths and PsychologyPsychologyOrganisedMathsCrammersDoing the chi-square testMathsOrganisedStep 1: Add up the values in the rows and thecolumns of the results table, and calculate anoverall grand totalThe easiest way to do this is to reproduce theresults table, adding the totals of the rows andcolumns and the grand total, like this:CrammersOrganisedTotalPsychology53540(5+35)Maths501060(50+10)5010Total55(5+50)45(35+10)Grand total100(5+50+35+10)to as the expected frequency. We use the followingvery basic formula:RCE=TThe expected frequency for the cell (E) equals therow total (R) times the column total (C) divided bythe grand total (T).The four sums are easy to calculate. Make sureyou work out the row total times the columntotal before dividing by the grand total though,and make sure you use the correct row totals andcolumn totals in each calculation.Observed frequency = 5Expected frequency = (5x40)/100= 200/100 = 2Observed frequency = 50Expected frequency = (55x40)/100= 2200/100 = 22Observed frequency = 35Expected frequency = (45x40)/100= 1800/100 = 18Observed frequency = 10Expected frequency = (60x45)/100= 2700/100 = 27To summarise, the expected frequencies for ourdata look like this:CrammersOrganisedPsychology218Maths2227Step 3: Use your observed and expected frequenciesto calculate chi-squareNow that we have observed and expected frequenciesfor each cell we can calculate chi-square usingthis equation:(0–E) 2x 2 =∑ETo calculate chi-square (c 2 ) we have to calculate(O–E) 2 /E for each cell, then add up (∑) the resultsof the sums.We’ve done this for you here. Each cell containsa single calculation, using the observed andexpected frequencies for that cell.Step 2: Work out what might be expected justby chanceTo work out the number in each cell that mightoccur by chance is easy, and the result is referredExtract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

36 | A2 PSYCHOLOGY: THE STUDENT’S TEXTBOOKCrammersOrganisedPsychologyObserved Observedfrequency = 5 frequency = 50Expected frequency Expected frequency= 2= 22(5–2) 2 /2=3 2 /2=9/2= 4.5Observed Observedfrequency = 35 frequency = 10Expected frequency Expected frequency= 18= 27(35–18) 2 /18=17 2 /18=289/18= 16.06 (2 decimal places)Maths(50–22) 2 /2=28 2 /22=784/22= 35.54 (2 decimal places)(10–27) 2 /27=-17 2 /27=289/27= 10.70 (2 decimal places)To complete the calculation of chi-square, add upthe results of the individual calculations:chi-square = 4.5+35.54+16.06+10.70= 66.8Step 4: Using the table, find the relevantcritical value of Chi-squareTo find the correct number you must knowwhether you chose a directional (one-tailed) ornon-directional (two-tailed) hypothesis. In ourstudy we simply stated that there would be a differencebetween the revising styles: we did not saywhether one would be larger than another, so wehave a non-directional hypothesis.Before you can use the table you must calculatea value called ‘degrees of freedom’, shortenedto ‘df’. This is done using the following simpleformula:df = (number of rows – 1)x (number of columns – 1)For our data it could not be simplerdf = (2–1)x(2–1)= 1x1= 1Finally, consult the relevant table. It’s printed atthe end of this book for you, but for convenience,the key portion of it is reproduced here. As usual,make sure you consult the table itself, so that youknow how it works.Levels of significance for a two-tailed test1230.20 0.10 0.05 0.02 0.011.64 2.71 3.84 5.41 6.643.22 4.60 5.99 7.82 9.214.64 6.25 7.82 9.84 11.34The value for chi-square that you calculated mustbe the same as, or larger than, the critical valueyou read from the table. Your calculated valuewas 66.8. The critical value for the significancelevel of 0.05 is 3.84. It seems that you can veryconfidently reject the null hypothesis and say thatyour data did not occur by chance. Your result issignificant (p

DATA ANALYSIS AND REPORTING INVESTIGATIONS | 37help present some of the data in such a way thatthe reader does not have to sort through a list towork out what has been said about the UK andwhat has been said about issues outside the UK.There may be many more comments, and thesemay be included in another table, but a good summarytable with samples of the comments will bevery useful to readers interested in seeing the kindof comments that have been made. The table maylook like this:CategoryCommentsdirectly relatingto UK issuesCommentsdirectly relatingto internationalissuesExamples‘The Scottish parliament can workhand in hand with the parliament inWestminster’‘House prices in the UK are, frankly,ridiculous’‘The floods in the West Country weredevastating. We must do all we can tostop this happening again’‘The US elections have huge implicationsfor the wars in the Middle East’‘The crisis in central Africa is worsening. Ifear for the lives of these people’‘Rebel fighters in parts of the RussianFederation have grown in strength overthe last two years’The table makes it much easier and faster to findthe relevant information. Tables like this might alsoinclude the total number of comments in eachcategory. If we have a total number we can draw agraph, like this:Number of comments604530150Number of comments made in each categoryUKInternationalIt is immediately clear from this graph that thespeeches of the politicians used in this researchhad many more comments that directly related tointernational matters than ones that dealt with UKissues. This is a perfect example of what a graphis for. Graphs allow readers to access a good summaryof the data in an immediate, visual way.Conventions of reportingon psychologicalinvestigationsOnce we have completed our research we need towrite it up before sending it for peer review and(we hope) publication. There are certain conventionsthat should be followed when reportingpsychological research. These are conventionsrather than ‘rules’ because some scientific journalsrequire a slightly different style, but generallyspeaking, psychological research should be presentedin a certain way. Imagine that we wantedto present our research into whether memory isworse with music playing than in silence. Ourreport should be organised into the sections weidentify here.1. TitleIt seems obvious to say it, but a strong and cleartitle is extremely important. Thousands of piecesof research are published every year. Researcherspour through the titles of this new research to seeif anything grabs their interest and is relevant totheir own work. A title that clearly expresses whatthe research is investigating is extremely helpful. Inour study of whether memory is worse in music orin silence we may choose the following title:‘Memory: An investigation into whymusic really makes it rubbish’However, this is not very informative, or formal. Abetter title might be:‘An investigation into how memory for numbers isinfluenced by music’This is more like it. From this title the reader isable to see that the research is into memory: itinvestigates memory for numbers, and music isused to disrupt it.An even better title might be:‘The irrelevant sound effect: The disruption ofmemory for serially presented digits in conditions ofmusic and silence’The phenomenon we are investigating here is actuallycalled ‘the irrelevant sound effect’. It has beenwidely studied and refers particularly to memoryin silence compared to memory with sound presentedat the same time. This title clearly indicatesto the reader what type of research is in the report.In this section of the book we have the permissionof the authors to use sections of a real paperwhich appeared in a publication called the Journalof Applied Psychology in 2007. The authors wereBeaman and Holt, and the copyright for the paperbelongs to the publishers, John Wiley and Sons.The title was:‘Reverberant auditory environments: The effects ofmultiple echoes on distraction by “irrelevant” speech’Extract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

38 | A2 PSYCHOLOGY: THE STUDENT’S TEXTBOOKIn their research, Beaman and Holt looked at howsounds with an echo (reverberation) influencedmemory differently than places without an echo.2. AbstractThe abstract is a short summary of the researchpaper placed at the very start. A good abstractshould provide a little background, some detailsof the method, a statement of results and a shortconclusion. All this in fewer than 150 words. Theabstract comes first, straight after the title, but youshould always write it last. The reason is that itincludes a very brief summary of the informationyou have included in every section of your reportand so it’s much easier to write when you havefinished the rest of the research report!Here’s an abstract that, unfortunately, is actuallybetter than some we have seen over the years! Wereally hope you agree that it is not terribly good.We did some research that was looking atwhether people were more rubbish at rememberingstuff when they had music comingthrough headphones. We gave them things toremember and they remembered them. Weworked out some maths and decided whethertheir remembering was better when theycouldn’t hear anything. It was. There wereabout 10 people we tested. At the end of itwe decided that it was useful to know this, forinstance, that’s why we think it should be quietin libraries mostly. Right at the end we thoughtabout some other research and we think wewill now investigate whether listening to oldfashionedmusic like your dad listens to willmake memory worse than listening to stuff youhave on your own iPod.There is very little detail in the abstract. Thebackground to the research is not clear at all. Wedo not know what the task itself was, what thehypothesis was, what the variables were or howthey were operationalised. We do not have a clearunderstanding of the statistical tests carried outor what exactly was found, and the suggestions atthe end for further research are vague to say theleast! Not very good at all. Now look at the versionbelow.The research detailed here is an investigationinto whether memory for digits presented inseries is influenced more by the presentation ofmusic during the task than when participantsproceeded in silence. Previous research hassuggested that any kind of sound interfereswith recall of numbers presented in series, suggestingthe hypothesis that memory would beworse when listening to music. An opportunitysample of 10 participants took part in anindependent samples design and provided datathat supported the hypothesis (p≤0.05). Howthe sound influences memory is discussed,with reference to the working memory model.Concluding comments included possibilitiesfor further research.This abstract is much better. Here we have somebackground, some detail of our design, the resultand even something of our discussion and conclusion– all in fewer that 150 words. Finally, hereis the real abstract from the Beaman and Holtresearch. It’s quite complicated, but it does giveyou an idea of the kind of information and detailyou find in professionally produced work.Two experiments examine the effect on animmediate recall test of simulating a reverberantauditory environment in which auditory distractersin the form of speech are played to theparticipants (the ‘irrelevant sound effect’). Anecho-intensive environment simulated by theaddition of reverberation to the speech reducedthe extent of ‘changes in state’ in the irrelevantspeech stream by smoothing the profile of thewaveform. In both experiments, the reverberantauditory environment produced significantlysmaller irrelevant sound distraction effects thanan echo-free environment. Results are interpretedin terms of changing-state hypothesis,which states that acoustic content of irrelevantsound, rather than phonology or semantics,determines the extent of the irrelevant soundeffect (ISE).3. The IntroductionThe introduction is a really important part ofthe write up. It’s here that you get the chance todescribe the background to your research andhow you really came up with your idea in the firstplace. This is where you place your work alongsideother research in the same field. All introductorysections differ slightly in length, depending onwhat it is that we need to cover and how muchbackground research there has been in the area.Introductions, however long, should neverthelessfollow the golden rule.Start WideEnd NarrowExtract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

DATA ANALYSIS AND REPORTING INVESTIGATIONS | 39Think of your introduction as a funneling system.You should start with all of your backgroundinformation and details of research that has alreadybeen done. You really need to provide a good generaloverview in the introduction. As you proceed,you should begin to focus your ideas on yourresearch. Towards the end of the introduction, talkabout the aims of your research, ending finallywith your hypothesis. Take a look at some excerptsfrom Beaman and Holt (2007).The use of virtual environments (VE) in psychologyis now well-established. Studies of spatialnavigation (Ruddle, Payne, & Jones, 1998),spatial memory (Ruddle, Payne, & Jones,1999)… and neuropsychological rehabilitation(Wann, Rushton, Smyth, & Jones, 1997)have all benefited from these VE techniques…… However, virtual auditory environments aremuch less common than virtual visual environments.When virtual auditory environmentshave been created, however, they have proveduseful to the study of the ‘sense of place’associated with particular buildings and historicsites… The purpose of the current study is toapply similar techniques to the study of auditorydistraction by so-called ‘irrelevant sound’known to disrupt working memory.The introduction begins very generally, andquickly proceeds to a statement of the broadaims of the research. The authors have really toldthe reader what to expect, the background andthe general ideas behind their study. Later in theintroduction Beaman and Holt begin to be morespecific.In a reverberant environment the direct signalis overlayed (sic) with multiple reflections,the intensity of each reflection and the delayimposed upon it being a function of the intensityof the original signal and the nature (e.g.the size, surfaces and contents) of the room…It follows that larger spaces with more reverberationmay intrinsically be less distracting environments(in acoustic terms) than small roomswith low levels of reverberation. This predictionfollows from the changing-state hypothesis ofJones and colleagues…we propose to take adifferent approach and ask whether extremevalues of reverberation, which we expect torepresent a powerful manipulation, will affectthe results obtained for a much smaller samplesize. The advantage of this approach is that itmakes positive, rather than null, predictionsand provides a direct test of the changing-statehypothesis.Here, the introduction is really setting the scene,explaining that certain hypotheses and theoriesdescribed in other research will be investigated.The authors finish by explaining that predictionswill be made to test a hypothesis. We can see thatthe introduction covers a great deal of ground.We’ve only included a few excerpts here: muchmore is described in the original article. We cansee from the sections we have reproduced that theintroduction begins very generally, then focuses,using the ‘start wide, end narrow’ rule, funnelingthe information down from a general discussion toa much more focused one.4. The MethodWhen writing a method section, you need to askyourself one simple question: ‘Would someoneelse be able to do exactly what I did from readingmy method?’If the answer to this is ‘No’, then your methodis not good enough. A good method has just theright amount of information for someone who hasnever watched you work before, and who maynever have carried out research in the area, tocarry out a ‘replication’ of your study. Replicationis a very important aspect of scientific research.If a study can be replicated then we can be morecertain that the findings are to be trusted. Replicationadds to the reliability of the research, withunreplicable research being regarded as unreliable.A good method provides other researchers withthe information they need to repeat your workclosely enough that they can hope to replicateyour findings. A method might include a numberof sections. Not all methods include all of thesesections, but we’ll include them here for reference.Design: A statement of the design used may beincluded. Here you might indicate the differentconditions in your research and if appropriateyou might say whether a repeated measures orindependent samples design was used, or eveninclude the independent and dependent varietiesIV and DV.Four conditions were presented, each with 15trials, giving a total number of 60 experimentaltrials per participant. Presentation order foreach candidate was chosen randomly from the24 possible condition orders… The four conditionswere … three levels of reverberation …and a ‘no sound’ control condition.Participants: In this section you would givedetails of your sample, including how many menExtract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk

40 | A2 PSYCHOLOGY: THE STUDENT’S TEXTBOOKand women there were and the age range. Youshould also indicate how the sample was identified,perhaps using an advertisement for a volunteersample, or the phone directory for a randomsample. It may also be appropriate to indicatewhere the research was conducted and from whatcountry the participants came. You must be verycareful to make sure that the participants remainanonymous as this is their right, and it is also oneof the principles of ethical research. This is fromBeaman and Holt (2007):Eleven students of … University volunteeredto participate in this study. All participantsreported normal hearing levels and normal orcorrected-to-normal vision. All had English astheir first language.Apparatus and stimuli: In this section researcherswould indicate the apparatus they used in theirresearch. If they used certain sounds, or visualstimuli, then they would give details of these, perhapsindicating how each was produced.Sounds were digitally recorded on MiniDiscand presented from a Sony MiniDisc player at acomfortable listening level to students throughSennheiser HD570 headphones. Listeninglevels were set to a comfortable level with atest recording before the experiment beganand kept at the same level for the durationof the experiment. Stimuli were presentedon a Pentium Class PC running Windows XPand MS PowerPoint software. …using CoolEdit sound editing suite (Syntrillium SoftwareCorporation).Procedure: This section of the report is particularlyimportant for those wishing to replicate theresearch. It is here that researchers show exactlywhat they did.Each participant was tested individually in aquiet room. The irrelevant sound was presentedcontinuously … throughout the presentation… directly from the digital recording. Participantswere instructed to ignore anything theymight hear and were reassured they would notbe tested on it in any way. In each trial … theseven numbers were presented in successionon a VDU. After the presentation the instruction‘recall now’ was presented. Participantsrecorded the numbers manually on responsesheets provided, writing from left to right.They were asked not to omit any responsesand, if they could not remember a particularitem, to write down their best guess and moveonto the next item. They were asked not to goback and amend previous answers if they feltthey had made an error. The visual instruction‘Push space for next trial’ cued participants toprogress through the experiment.5. The resultsAfter researchers have identified their method, theresults are presented. The presentation of resultsmay be in the form of descriptive statistics, tables,graphs and also inferential statistics. In the caseof Beaman and Holt (2007) a more complicatedstatistic called an ANOVA (Analysis of Variance)has been used. Excerpts from the results of thisresearch are shown below.… analysis of variance (ANOVA) … showed asignificant effect of … speech condition on themean number correctly recalled…, p

DATA ANALYSIS AND REPORTING INVESTIGATIONS | 41irrele-vant sound stream and that anythingwhich acts to reduce the number or extentof these changes will reduce the size of thedistraction effect.The words in the irrelevant stream in theseexperiments were not, so far as can be ascertainedby passive listening, rendered unintelligibleby the process of adding reverberation,although no formal tests were carried out onwhether the relative levels of intelligibility variedbetween the conditions and that remains alimitation of the current study.The present results are inconsistent with theview that it is always helpful, when aiming tooptimise cognitive efficiency, to reduce the levelof reverberation within the environment……These experiments are the first to demonstratethat intrinsic factors of particular environmentscan modify the extent of auditory distractionobjectively observed in such environments.Concluding remarks can be brief and clear. Theysummarise what the researchers feel can sensiblybe concluded from their results.In conclusion, we have established that reverberationcan reduce the auditory distractioneffect produced by irrelevant speech.At the end of the discussion you might decide toidentify other things that could now be investigatedthat relate to your research.Further research looking at the subjectiveexperience of such noise should, in addition,lead to a greater understanding of how auditoryenvironments can be made more conducive toefficient cognitive function both objectively andsubjectively.7. ReferencesThe final section is a reference list. Here theresearchers give full details of the other research inbooks or academic journals that they have referredto in their investigation. This is very useful forthose reading the work as they may be interestedto extend their reading to the articles and booksindicated by the researchers who wrote the paper.The reference section from Beaman and Holt(2007) includes the following:1. Ruddle, R. A., Payne, S. J., & Jones, D.M. (1998). Navigating large-scale ‘desk-top’virtual buildings: Effects of orientation aidsand familiarity. Presence: Teleoperators & VirtualEnvironments, 7, 179–192.This is a research paper by Ruddle, Payne andJones. It was published in 1998, in the seventhvolume of an academic journal called Presence:Teleoperators & Virtual Environments on pages 179to 192.2. Ruddle, R. A., Payne, S. J., & Jones, D. M.(1999). Spatial knowledge and virtual environments.In J. M. Noyes & M. Cook (Eds.),Interface Technology: The Leading Edge (pp.135–146). Baldock, England: Research Studies.This is a chapter of a book called InterfaceTheory: The Leading Edge. The book was editedby Noyes and Cook, and was published in 1999by a company called Baldock, based in England.The chapter was written by Ruddle, Payne andJones and is titled “Spatial knowledge and virtualenvironments”.3. Wann, J. P., Rushton, S. K., Smyth, M., &Jones, D. (1997). Rehabilitation environmentsfor attention and movement disorders. Communicationsof ACM, 40, 49–52.This is a short article, written by Wann, Rushton,Smyth and Jones in 1997, called “Rehabilitationenvironments for attention and movement disorders”.It appeared in volume 40 of an academicpublication called Communications of ACM, onpages 49–52. ACM stands for Association of ComputerMachinery.Extract from A2 Psychology: The Student’s Textbook © Nigel Holt and Rob Lewis ISBN: 9781845901004 www.crownhouse.co.uk