Research Methods: Literature & Lecture PDF

RESEARCH METHODS: LITERATURE & LECTURE CHAPTER 4: OBSERVATION AND DESCRIPTION: 2 kinds of research methods: 1.​ Experimental research ○​ Investigator intervenes to make something happen → manipulates a variable ○​ Allows to make causal claims → ‘stress aff...

RESEARCH METHODS: LITERATURE & LECTURE CHAPTER 4: OBSERVATION AND DESCRIPTION: 2 kinds of research methods: 1.​ Experimental research ○​ Investigator intervenes to make something happen → manipulates a variable ○​ Allows to make causal claims → ‘stress affects short term memory performance’ ​ First the cause, and then the effect ○​ Control over environment and observes effect of manipulation ○​ Manipulation of IV _causes_ variation in DV 2.​ Observational research ○​ Study natural variation ○​ Observes and describes it - trying not influence ​ Bc not investigating effect of intervention per se ​ Assumption that will have no effect on result, but only helps him find out something (her opinion on some issue, who vote for) of the subject ​ These assumptions may not be justified → observer effect ○​ Mere fact that something can occur is interesting → if happens, devote time to it → case study ○​ Less invasive ○​ Natural variation of IV _is associated with_ natural variation of DV ○​ Different claims: ​ Frequency → ‘what percentage of Dutch children have ADHD?’ ​ Association → ‘Do children with ADHD bully more?’ ​ Associated with each other ​ 2 variables  Disadvantages: Experimental Observational Impossible to manipulate Confounds → make causal conclusion on two random variables, need to eliminate all confounding variable Impractical - can be difficult No causality Unethical Unnatural Advantages: Experimental Observational Control Eariser to do, less invasive Casualty Natural behaviour CASE STUDIES: Where study a single case ​ Eg. Phineas Gage → rod penetrates phineas skull into orbito-frontal cortex ​ He survived ​ How brain trauma affected behaviour → link ○​ Constant personality changes ​ Bad planning ​ More cursing ​ Restless Strength & Limitations of Case Studies Strength Limitation Tell us what can happen - specific failure Not generalizable (can be generalized to linked to specific area of brian damage → rest population) → no tell us ab what important is that they can happen typically does happen -​ Eg. → sigmund freud depended on case studies and very criticised -​ From his conclusion, neurotic symptoms can expression unconscious urges -​ Went on to claim all behaviour has such sources -​ Freud's theory of mind → went too far beyond his data → few case  studies with atypical people (affluence viennese with symptoms) Generate new ideas/lines of study Are susceptible to 2 sources of error ​ Damage project ​ Dangerous bc close, released contact with observer-scient and subject 1.​ observer bias → see what she expects to see, select what fits their theory → attach important events to one that fit the theory and less importance to those that do not 2.​ Observer effect → may steer conversation in which they think it should go Add to evidence favouring/challenging a Might have alternative explanations principle - high degree of depth EXAMPLE: BROCA’S SPEECH AREA: NEUROPSYCHOLOGY ​ Paul Broca, physician ​ Meets patient called ‘Tan’ → bc only world that could say ○​ Suffered from expressive aphasia → inability produce speech ○​ Died 6 days after meeting broca ​ Autopsy revealed damage to ‘Broca’s Area’ on left frontal lobe ​ Patient → trouble producing speech, but can understand it ​ Later case studies → supported conclusion → patients with broca’s aphasia can understand, write, intelligence → as if cannot remember how use speech apparatus ​ Broca meets another patient, same as ‘Tan’ (6mon. later) → similar symptoms, developed & more damage to brain  ​ Broca concluded → found area brain specialized for speech production Fed into long-lasting controversy → many students doubted it: ​ The mind is one, not plural (many parts that function to make up the brain and function) ​ Seemed as if it was a working unit Broca brought 2 cases → associated parts of brain → damage just in Broca's area affects producing speech Other scientists found another example of localization of function ​ Controversy → eg. on how many parts has the brain, does the soul have many parts SURVEY: INTERVIEWS OR QUESTIONNAIRES: Survey research → more likely to ask ​ What is true of large/many/small groups of cases on which to obtain data? ​ Survey consider larger part of a population, not single cases ​ Used in political/opinion/market research ​ Face technical problem → sampling can be difficult & phrasing survey questions is an art ​ Used bc some behaviour is hard to observe directly Advantages Disadvantages Topics that are hard to study directly Possibility of subjectivity in data Quick and easy Questionnaires → no response Lot of information Questionnaires → not a lot of depth EXAMPLE: DEMOGRAPHIC SURVEY: COHABITATION AND DIVORCE ​ From 1940s-1960s → decrease USA, canada, sweden adults married and increase living together unmarried ​ Objective indicator → marked by change in american society attitude toward sexuality ​ We can say that cohabitation does not make divorce less likely, though it might have ​ Couples cohabitate before marriage more likely to divorce than who did not, 2:1 Limitations of study Not necessarily mean that prior cohabitation causes greater risk of divorce later on → Correlation not imply causality Maybe cohabitation make no difference but people with certain attitudes & values are more likely than others to cohabit before marriage and dissolve if becomes unsatisfying EXAMPLE: SOCIAL PSYCHOLOGY: BENNINGTON STUDIES Surveys often conducted on small scale Late 1930s → Bennington college in vermont: ​ Close knit community with reputation for social awareness and political liberalism ​ New student → + likely have conservative attitudes ​ Newcomb wondered effect values of school on new entries → change in attitude ​ Intensive interviews with students ​ Drift from liberal attitudes on political and social issues → attitudes operationalized in many ways ○​ Asked who preferred for upcoming president, juniors and seniors preferred Roosevelt 3:1 ​ Students with closed mid/ no change their ideas → be rather isolated, no admired, friends Problem: Peer Pressure ​ New faculty → made conscious effort no push view on student, but older students were less restrained Problem: Is it a long lasting change? ​ 25 years later - locate 94% women and study them again (some interviews, or email if not possible) ​ New questions were devised bc different issues Problem: Find comparison Group ​ Women compatible (age/income/geographical location) to this group → same interview/questions ​ → bennington expressed more liberal views than control Problem: Maybe in family ​ Compare sistster & family & sister in law (exclude ehtinity/affluence and parental attitudes/social class) Result → Bennington → + liberal than sisters and sisters in law did → change in attitude was genuine and long lasting ​ BUT women of bennington select husbands with more liberal attitudes than non-bennington women & close friends with same values EXAMPLE: CULTURAL PSYCHOLOGY: EXPLANATION OF ACTIONS Miller (1984) ​ Interview people 2 cultures → Hindus living in India & american living in US ​ Asked think examples actions performed by friends and explain why actions were performed ○​ American explained with → he/she is that kind of person ○​ Indian respond with outside situational factors → did because situation was such and such 2 cultures → differ as to kind of explanation for someone actions that come readily in mind ​ Americans think ab personality variables ​ Indians look at situation Problem: american might think situation where personal causes & india determined by situation ​ took scenarios generated by indian and american comments ○​ indians given situations explanation for actions ○​ americans explained same actions in same situation by reference to personality of the actor Result: American prefer have one kind of explanation and indian another for same event PARTICIPANT OBSERVATION Researcher participants in certain group of people to study the dynamics of that group  Advantages Disadvantage Valuable info that not easily accessible Observer Intensely Involved Observer effect EXAMPLE: SOCIAL PSYCHOLOGY: THE ‘SEEKERS’ Festinger et al. (1956) Procedure Marian Keech think receive message from outer space (flood destroy world in december) Small followers (normal people, no media publicity) believe her → committed, some quit jobs Observe their meetings No flood arrives → belief in group strengths and start serving from media attention Problem: believed more at beginning ​ No were normal, no standing out Conclusion & Results Using theory cognitive dissonance → uncomfortable if actions in conflict with public beliefs Social pressures → can promote attitude change to convince themselves, need to convince others DIRECT OBSERVATION OF BEHAVIOUR: ​ Behaviors is being observed in ○​ Natural environment ○​ Artificial environment Advantages Disadvantages Observe exactly what interested in in real Less control on participants compared to world experimental studies EXAMPLE: ETHOLOGY: SEX AND STICKLEBACK Ethology → study animal and human behaviour in natural setting → each animal has reproductive behaviour ​ Typical reproductive behaviour → three-spined stickleback (freshwater fish) ​ Seemingly simple, but is complex Three-spined stickleback (freshwater fish) Mating Ritual: 1.​ Male into reproductive condition ( changes in color bc of reproductive hormone) 2.​ Build underwater nest 3.​ Patrol entrance a.​ If male approaches → make characteristic ‘threat display’ → standing on head with fins outspread b.​ If female approaches → courts with a zig zag dance 4.​ If successful → do a choreographed mating sequence together (responds) 5.​ Male leads female to nest 6.​ She deposits eggs 7.​ He deposits sperm 8.​ He chases her away & take over job caring eggs Does take practice to do these Procedure: ​ Releasing intruder male (red underbelly) → will be threatened ​ Releasing male → will be courted Result: Act in characteristic way Concept of instinct behaviour had strong comeback HUMAN ETHOLOGY 1: FACIAL EXPRESSIONS: Observation → see human behaviour Can different cultures recognize facial expressions of members of different cultures? Yes, but there are differences ​ Japan → to mask any negative emotion with polite smile and eyebrow flash is rude So entire culture becomes a case study, not only individual EVEN if facial expressions can develop without learning, can still be modified by learning experiences encounter in society grew up in  Woman is → deaf and blind How do I learn facial expression (smiling), maybe instinct? Humans more complicated → taught display rules by society grow up in, eg. what appropriate expression HUMAN ETHOLOGY 2: BOOK CARRYING: Differences observed in USA & Canada → college and libraries → more general Women Men Carry books arm bent, cradling against Carry unsupported with arm straight cest TESTING HYPOTHESIS WITH OBSERVATIONS: Observational research → used to test specific hypothesis EXAMPLE FROM CLINICAL PSYCHOLOGY: SMOKING, OBESITY AND SELF HELP Certain behavioural disorders → easy to stop BUT difficult to not do ​ 80% and more smokers and overeaters relapse within first year ​ Most relapses come from clinics Schachter (1982) Procedure: Sample → psychology department where taught & in small New England town where spent summers and knew residence with smoking/weight/both problems and Small scale survey and questionnaire/interview and ask: ​ What they did about it  40% people who quit → ramined smoke free for years = similar to weight issues Problem: Why are clinics less encouraging? ​ Those with severe problems go there ​ No time & right mental health to deal with this problem ​ Asked if sight help → these ones tend to be most overweight or heaviest ex smokers Results: Very discouraging statistical outlook for people with problems can be misleading → can be sampling bias Cases seek professional help are ones whose problems are difficult to overcome EXAMPLE FROM SOCIAL DEVELOPMENT: MATERNAL RESPONSIVENESS AND INFANT CRYING Direct observation of behaviour → test theory, or a pair of them Baby fed, changed and put in crib but cries, What does the parent do? If Respond (Attachment Theory) If Ignore (Reinforcement Theory) Attention caretaker can bring warmth, John Bowlby → baby learns no manipulate jiggling, contact comfort acts as reward → you, also learning no depend on reinforcer → if baby cries brings that responding to signals he gives comfort, crying will be reinforced and persistent Dr.Benjamin Spock (world famous Erick Erickson → baby fail to form ‘basic pediatrician) → - if come running to baby trust’ in you and people every time cries, will learn to manipulate you View on grandparent → going to spoil baby Baby not be reinforced by crying but reinforced not dependent on what baby does, may form conditioned helplessness - Martin Seligman Other grandparent → go with impulses Reinforcement theory → advice is to let baby cry ​ Stop if unrewarded and after have a baby who cries only if is in distress Attachment theory → confort baby ​ only tended when needs feeding/changing like baby mom → become insecure and a crying baby Predictions: If Attachment Theory is Right If Reinforcement Theory is Right have secure, non fussy babies Have babies that no cry a lot Unresponsive, tend have insecure and Should be positively correlated with fussy babies → amount crying negatively parents responsiveness → responsive correlated with parent responsiveness parents should tend to have babies who cry less More responsive parent, less baby cry THE BELL AND AINSWORTH STUDY (1972): Procedure: Spent afternoon with mother-infant during child's 1st year Observation study, no changes Responsiveness mother operationalized in many ways → counted: -​ Number cries  -​ Number times mother responded -​ Time measured When cry hear → observer start stopwatch and stop it when mother left to deal with baby ​ reliable bc others starts the records Results: Negative → more responsive mothers had less fussy babies Good example how correlational data used for test predictions: ​ emphasise role communication and responsiveness to communication → good for attachment theory ​ Principle correlational data no establish what causes what Other ways interpreting data → maybe true unresponsive mother make baby fussy, or other way around -- baby who is fussy exhausts mother and make her unresponsive Problem: Other factors ​ might be lot of noise at home ​ baby + mother would be tense VALUE OF OBSERVATIONAL RESEARCH: Questions are open ended ones As explore, more specific questions come to mind Thus whereas correlation does not establish casualty, correlation studies can sometimes see confirm predictions of causal theory OBSERVATION AND DESCRIPTION II: SOME TECHNICAL PROBLEMS Chapter about gathering misleading data and prevent this from happening Sampling Bias → gather data from different organisms/events, no having sample representative of population Observer effects → distort what observe, or people might change their way of acting bc know being observed Inferential Bias → make inferences from data in biassed way SAMPLING: With representative sample of population → eg. college students not representative of US population ​ Survey needs draw conclusion ab specific population based on findings, more than one experiment Analyse sampling used: ​ People? ​ Group? ​ Animals? SAMPLE AND POPULATION: Want to know general behaviour of population, not specific, individual cases ​ Sample → represents whole population ​ Can be biassed or unbiased Problem sampling bias: data is useless ​ Most biased is a self selected sample ​ Representation of population whole, more generalisable as possible Avoid Bias: ​ define our population and seek representative sample from population RANDOM SAMPLING: ​ every member of population has equal chance of being selected ​ manually (selecting from hat) or with computer (generate series number randomly): need to ○​ List all members of population ○​ Assign each member a number (or name) ○​ Draw random sample of numbers ○​ Find person corresponding to number ​ If involved in an interview and no want to participate → select more subjects at random Problems: ​ No guarantee represent a population ​ Make unlikely that sample will be very different from population which is drawn ​ Larger the sample, more unlikely any large differences become Problem: sample representative of one population may not be representative of another ​ Cannot be generalised to other population bc of exceptions VARIANTS OF RANDOM SAMPLING: ​ Multistage sampling → eg want a list of all college students US: 1.​ List all colleges in US  2.​ Take random sample of college 3.​ Random sample of students ​ Systematic sampling: 1.​ No keep track of individuals 2.​ Observe /interview every person 3.​ Quicker than random sampling 4.​ Managed to randomly select OTHER APPROACHES TO SAMPLING: For purposes like: ​ Political ​ Opinion ​ Polling representative Use: PURPOSIVE SAMPLING: ​ Take in consideration certain characteristics where are relevant to the question ○​ Eg. Schachter looked for peoples with smoking and weight problem ​ Cases observe selected because question ask, not because representative of any particular population CONVENIENCE SAMPLING: ​ Most common ​ Select for accessibility and convenience ​ Can make an unrepresentative sample: ○​ variety need a random sampling ○​ generalised to people similar to subjects of experiment → based on assumption people are alike, but does not always apply ○​ Eg study on memory → college students have experience of learning this for remembering them later (by memory) → not apply to adults ​ Has its dangers: ○​ can miss differences in ​ social relations ​ facial expression ​ strategies in remembering things ​ Generality of conclusions come from consistency of findings, not the representativeness of subjects ​ This does not mean that no satisfaction → very large proportion experimental findings from limited database WHEN DO WE NEED REPRESENTATIVE SAMPLES? Any study has certain generality we can generalise to population of subjects like these What ab random sampling → need them when there is real-world population out there, specific existing one, not the one we define to which we want to generalize our findings Eg. political polling ​ Foolish to selected sample of affluent community member ​ Interest in whole population of voters ​ Need to make a representative sample ​ Need to decide case by case OBSERVER EFFECTS: Observer effects → when subject no behave normally → problem when analysing the results Eg → if adult is watching a child: ​ Could be con best behaviour ​ Typical behaviour with parent EXAMPLE FROM ANIMAL BEHAVIOR: THE CASE OF CLEVER HANS Mr. Ostler & Clever Hans Procedure: In germany Ostler discovers horse does math → as a 5th/6th grader Checked it out scientists until Pfungst whisper in ears horse 4+7 No know how to answer → horse looked at audience which would tense and give him the answer → reading unintended signals Called Clever Hans effect Result: Audience would give unintended cues that affect his performance A CLEVER HANS EFFECT IN HUMANS: FACILITATED COMMUNICATION Autism → disorder appearing early childhood, characterised by severe deficiencies in social interaction & language etc ​ Even if did/could not speak able to type coherent messages on keyboards (hands held over keyboard by facilitator) ​ Renowned hope ​ BUT case of unconscious cueing → facilitator opportunities to influence what he/she types Procedure: Showed drawing to facilitator and child types his name Neither could see others drawing Result:  Child types out name that fit drawing of facilitator → response determined by facilitator → cuing the child in someway without being aware Some adults (teachers & parents) where incarcerated by accusations of sexual abuse on children and were released Possibility observer effects in facilitated communication → obvious to competent researcher → bc minimal training Many large studies show that child performance depended on facilitator → Clever Hans effect EXAMPLE FROM MEMORY: LEADING QUESTIONS Many research situations → depend on question answer scenarios Face problem → depend on how the questionnaire puts the question Loftus and Palmer 1974 Procedure: Watched videotape auto collision (Same for 3 conditions) Were then asked ‘how fast was the car going when they hit/smashed each other?’ Memories of the tapes were invented → asked if remembered if car stopped at stop sign → many remembered but was not there Other asked if there was broken glass → more subjects in smashed than hit condition remembered seeing broken glass when there was not Result: The group ‘smashed’ → gave higher estimates of speeds seen earlier Leading question can & do discord what subject thinks he remembers → interest in its own right but raises possibility that something may occur Give us distorted because of what researcher says, way they were questioned Researcher gave unintended cues that affect his performance CONTROLLING OBSERVER EFFECTS: What can be done? HIDING: From subjects, to avoid influencing them ​ Behind one way mirror ​ Can be done of people who eat in restaurants/drink at bars → posing as other customers WAITING IT OUT: Kids get used to presence, until lose interest and hope that they forget and do their business: ​ Used by Jane Goodall as tactic → began instructing herself to chimp in jungle → eventually though chimps got used to having her around DECEPTION: Might pretend to be part of something to observe behaviour → called deception Necessary to mislead people we are observing ​ ‘Cloak of legitimacy’ → from deception → observe people in bar and restaurants → pretending and attempting to deceive UNOBTRUSIVE MEASURES: Not to be there when behaviour occurs → look at consequences of behaviours that can be observed later ​ See how police records can be used to test hypotheses about human aggression and influences ​ By observing reduces consequences → unobtrusive measures ‘BLIND’ OBSERVERS: Observers know what result of study ‘ought’ to be and communicate their expectations to subject Eg. clever hands ​ When no know observers, then no one tells him the answer ​ Or with children and facilitator Experiments conducted by students/technicians who are kept blind → simply rule out clever hans effect so no one expects subjects to behave and give them clues OBSERVER BIAS: → tendency of researcher to unintentionally influence or interpret data in way that aligns with expectation/select data EXAMPLE FROM PHYSICS: ‘N-RAYS’ More likely to see events that confirm theory that ones that do not 1903, Professor Blondlot → says discover new form radiation called N-rays No other scientists verify his findings While researcher removed prism supposed to isolate N-rays (no show them anymore) without hosts knowing → still could see the rays → driven by expectations EXAMPLE PSYCHOLOGY: UFOS Hayman stage magician → concerned by nonsense offered as facts by fakers (fake news) Appears as guest on talk show and informed audience that while driving from New Jersey to New York seen a V-formation flying (ufo) One ‘eyewitness’ → called confirm his sighing which was not real Looked in sky and expected to see a ufo so they did CONTROLLING OBSERVER BIAS: BLIND OBSERVING: Control observer bias & effect ​ IF clinicians interview people not know person/patient → expectation patients not affected by their judgement ​ Eg. Bell and Ainsworth study → they did not know until it was over → minimize bias and effects observer OBJECTIVE MEASURES: Objective means → risk bias is reduced ​ Eg.Harlow experiment → global judgement opened door to bias in judgement ​ INSTEAD → turned into physical measure → how much time each baby monkey spent with each model mother, so observer bias was reduced ​ Eg. Bell and Ainsworth study → baby cried → observer click a stopwatch and again if mother when to baby MULTIPLE OBSERVERS: Multiple people agree what happened, more likely it is what really happened ​ Especially in subjective judgements ​ Determining interobserver reliability is in part a check on observer bias ​ No get rid of biases, simply shows how serious problem is ​ If researchers no agree → approach problem with more objective methods or better trained observers OBSERVER BIAS IN EVERYDAY LIFE: Humans strongly biased toward finding meaningful patterns even in random stimuli → hear messages in seashell or see the main in the moon CORRELATION AND CAUSALITY: Inferential bias → even if data are sound, we many unsound inferences from them ​ Particular instance where this happens → confusion of correlation and causality ​ Correlation not imply causality ​ Eg. Keith Stanovich (1998) → A ‘little green men’ has revolutionised family planning research ○​ Discovered toaster method of birth control → more electrical appliances there are in household, fewer children you have ○​ If I want to limit number children, get many toasters ○​ Nonsensical conclusion but no more nonsensical than similar conclusion that are down from similar data, even professions → both variables affect the income ​ Wealthier people have less children than poorer people → more disposable income → more likely spend money on luxuries ​ Most unlikely that large number electrical appliances causes small number of babies ​ Even though the two have no cause-and-effect relationship Problems of Earlier Study: ​ Bell and ainsworth study of maternal responsiveness ○​ Responsiveness was negatively correlation with fussiness ○​ Reinforcement hypothesis → predicts opposite relationship but NO establish attachment hypothesis is correct There are 3 possibilities and data no distinguish among them: 1.​ Could be one realises on the other → 2.​ Could be the opposite 3.​ Can not be affecting one another but there is a third variable which is affecting the other two Eg: High self esteem is prerequisite for success at school - Dawes 1994 X= self esteem Y = academic performance 1.​ X affects Y 2.​ Y affects X 3.​ Secret option Z (could be genetics, good nutrition) that affects both X and Y Useful way of thinking though interpretation of any correlation data → think ab 3 possibilities