Psychology_ frontiers and applications-54-84.pdf

CHAPTER 2 CHAPTER OUTLINE Studying Behaviour Scientifically Replicating and Generalizing the Findings SCIENTIFIC PRINCIPLES IN PSYCHOLOGY Scientific Attitudes Gathering Evidence: Steps in the Scientific Process Two Approaches to Understanding Behaviour Defining and Measuring Variables METHODS OF...

CHAPTER 2 CHAPTER OUTLINE Studying Behaviour Scientifically Replicating and Generalizing the Findings SCIENTIFIC PRINCIPLES IN PSYCHOLOGY Scientific Attitudes Gathering Evidence: Steps in the Scientific Process Two Approaches to Understanding Behaviour Defining and Measuring Variables METHODS OF RESEARCH Descriptive Research: Recording Events Focus on Neuroscience: The Neuroscience of the Human Brain at Work Correlational Research: Measuring Associations between Events Experiments: Examining Cause and Effect Frontiers: Does ESP Exist? ETHICAL PRINCIPLES IN HUMAN AND ANIMAL RESEARCH Ethical Standards in Human Research Ethical Standards in Animal Research CRITICAL THINKING IN SCIENCE AND EVERYDAY LIFE Applications: Evaluating Claims in Research and Everyday Life Research Foundations: Very Happy People THREATS TO THE VALIDITY OF RESEARCH Confounding of Variables Placebo Effects Experimenter Expectancy Effects I have no special talents. I am only passionately curious. —Albert Einstein What are the issues here? What do we need to know? Where can we find the information to answer the questions? In March 1964, a young woman named Kitty Genovese was attacked by a knife-wielding assailant as she was returning from work to her New York City apartment. She was stabbed repeatedly and raped. The 3:00 a.m. attack lasted about 30 minutes, during which time her screams and please for help were heard by 38 of her neighbours. Many went to their windows to find out what was happening. Yet nobody assisted her, and by the time anyone called the police, she had died. On May 12, 2001, Katerine, a 17-year-old Canadian girl, was severely beaten and left half-naked in a parking lot near a west-end subway station in Montreal. Katerine remained there in a coma for hours, ignored by pedestrians, who blithely walked past the girl. Even more upsetting was that employees working at a call centre, Sitel Canada, next to the parking lot, could see Katerine from their windows. When several employees wanted to call 911 and go to the girl’s assistance, they were told not to do so by their supervisor. Katerine remained in a coma for months, and when the police published her photo and pleaded for help from bystanders, no one came forward. Sitel Canada did eventually fire the supervisor. S cience frequently has all the mystery of a detective story. Consider the psychological puzzle of bystander intervention. If you were in trouble and needed help from bystanders, would you receive it? Many citizens act decisively to help someone in need. But, as other tragedies illustrate, people do not always come to the aid of others. Why do bystanders sometimes risk injury and death to assist a stranger yet at other times fail to intervene— even when helping or calling the police entails little personal risk? We will return to this puzzle shortly. In this chapter, we explore principles and methods that form the foundation of psychological science. These principles also promote a way of thinking—critical thinking—that can serve you well in many aspects of your life. SCIENTIFIC PRINCIPLES IN PSYCHOLOGY At its core, science is an approach to asking and answering questions about the universe around us. Certainly, there are other ways we learn about the world and ourselves—through reason, intuition, and common sense; religion and spirituality; the arts; and the teachings of family, friends, and others. What distinguishes science from these approaches is a process guided by certain principles: the scientific method. The scientific method was used by physicists and chemists for several centuries to make great progress in determining the laws of the physical sciences. Psychologists took much longer to adopt the scientific method for behavioural science because the subject matter is not as tangible as that in the physical sciences. Calculating the speed of a thrown ball by measuring time and distance is much more straightforward than measuring a person’s intelligence. In this chapter, we explore in detail the principles and methods that form the foundation of psychological science. Scientific Attitudes Curiosity, skepticism, and open-mindedness are driving forces behind scientific inquiry. Like a child who constantly asks “Why?” the good scientist has an insatiable curiosity. And like a master detective, the good scientist is an incurable skeptic. Each claim is met with the reply “Show me your evidence,” and even when a mystery appears to be solved, the good scientist asks, “Might there be a better explanation?” Scientists also must remain open-minded to conclusions that are supported by facts, even if those conclusions refute their own beliefs. Following the Kitty Genovese murder, two psychology professors in New York City, John Darley of New York University and Bibb Latané of Columbia University, met for dinner. They were so curious about how 38 people could witness such a violent crime and not even call the police that they decided to investigate further. Darley and Latané also were skeptical of the media’s “bystander apathy” explanation; they believed it was unlikely that every one of the bystanders could have been apathetic. They noted that the bystanders could see that other neighbours had turned on their lights and were looking out their windows. Each bystander might have been concerned about Kitty Genovese’s plight but assumed that someone else surely would help or call the police. Darley and Latané reasoned that the presence of multiple bystanders produced a diffusion of responsibility, a psychological state in which each person feels decreased personal responsibility for intervening. They performed several experiments to test their explanation. Gathering Evidence: Steps in the Scientific Process Science involves a continuous interplay between observing and explaining events. Figure 2.1 shows the following five steps, which reflect how scientific inquiry often proceeds. Step 1: Identify a Question of Interest. Curiosity sparks the first step: identifying a question of interest. From personal experiences, news events, scientific articles, books, and other sources, scientists observe something that piques their interest, and they ask a question about it. Darley and Latané 1. What key scientific attitudes did Darley and Latané display? 2. How does Darley and Latané’s research illustrate the basic steps of the scientific process? 36 CHAPTER T WO Examining bystander intervention: Why do people sometimes fail to help a victim in need during an emergency, even when there is little or no personal risk? What factors increase or decrease the likelihood that a bystander will intervene? 1. IDENTIFY Question of Interest Kitty Genovese is murdered. The attack lasts over 30 minutes. Neighbours watched but failed to call the police until it was too late. The public is shocked. Why did no one help? 2. GATHER Information and FORM Hypothesis A difusion of responsibility may have occurred. Hypothesis: IF multiple bystanders are present, THEN a difusion of responsibility will decrease each bystander’s likelihood of intervening. 3. TEST Hypothesis by Conducting Research • Create an “emergency” in a controlled setting. • Manipulate (control) the perceived number of bystanders. • Measure whether and how quickly each participant helps the victim. 4. ANALYZE Data, Draw Tentative Conclusions, and REPORT Findings The data reveal that helping decreases as the perceived number of bystanders increases. The hypothesis is supported. (If data do not support the hypothesis, then revise the hypothesis or procedures and retest.) 5. BUILD a Body of Knowledge; Ask Further Questions; Conduct More Research; Develop and Test Theories Additional experiments support the hypothesis. A “Theory of Social Impact” is developed based on these findings. Test the theory directly by deriving new hypotheses and conducting new research. FIGURE 2.1 Using the scientific method. observed that nobody helped Kitty Genovese and then asked, “Why?” 3. What is a hypothesis? Step 2: Gather Information and Form Hypothesis. Next, scientists determine whether any studies, theories, and other information that might help answer their question already exist. Then they form a hypothesis. Noting that each bystander probably knew that other bystanders were witnessing Kitty Genovese’s plight, Darley and Latané proposed that a diffusion of responsibility reduced the likelihood that any one bystander would intervene. This tentative explanation is then translated into a hypothesis, a specific prediction about some phenomenon that often takes the form of an “If-Then” statement: “In an emergency, IF multiple bystanders are present, THEN the likelihood that any one bystander will intervene is reduced.” Step 3: Test Hypothesis by Conducting Research. The third step is to test the hypothesis by conducting research. Darley and Latané (1968) staged an “emergency” in their laboratory and recorded people’s responses. Male undergraduate participants were told that they would be discussing college experiences. To ensure privacy, they would be in separate rooms, would communicate through an intercom system, and the experimenter would not listen to their conversation. The students understood that they would take turns speaking for several rounds. In each round, a student would have two minutes to speak, during which time the others would be unable to interrupt or be heard, because their microphones would be turned off. As the discussion began over the intercom, a speaker described his difficulties adjusting to university life and disclosed that he suffered from seizures. During the next round, this same speaker began to gasp, saying, “ ‘. . . Could somebody-er-er— help . . . [choking sounds] . . . I’m gonna dieer-er—I’m gonna die-er—help . . . seizure’ [chokes, then silence]” (Darley & Latané, 1968, p. 379). Unbeknownst to the students, they actually were listening to a recording. The recording ensured that all of them were exposed to the identical “emergency.” To test how the number of bystanders influences helping, Darley and Latané assigned students to one of three conditions on a random basis. Each student was alone but was led to believe that, on the intercom system, (1) he was alone with the victim, (2) another listener was present, or (3) four other listeners were present. The students believed that the seizure was real and serious. But did they help? Studying Behaviour Scientifically Step 4: Analyze Data, Draw Tentative Conclusions, and Report Findings. At the fourth step, researchers analyze the information (called data) they collect, draw tentative conclusions, and report their findings to the scientific community. As Figure 2.2 shows, Darley and Latané found that all participants who thought they were alone with the victim helped within three minutes of the seizure. As the number of presumed bystanders increased, the proportion of actual participants who helped decreased, and those who did help took longer to respond. These findings support the diffusion of responsibility hypothesis and illustrate how research can contradict such common-sense adages as “There’s safety in numbers.” Darley and Latané then submitted a report describing their research to a scientific journal. Expert reviewers favourably judged the quality and importance of the research, so the journal published the article. Publishing research is essential to scientific progress. It allows fellow scientists to learn about new ideas and findings, to evaluate the research, and to challenge or expand on it. Cumulative proportion helping (%) Step 5: Build a Body of Knowledge. At the fifth step, scientists build a body of knowledge about the topic in question. They ask further questions (e.g., What other factors affect bystander intervention?), formulate new hypotheses, and test those hypotheses by conducting more research. As evidence mounts, scientists may attempt to build theories. A theory is a set of formal statements that explains how and why certain events are related to one another. Theories are broader than hypotheses, and in psychology theories typically specify lawful relations between Participant is the only bystander 100 80 Participant plus 1 other bystander 60 40 Participant plus 4 other bystanders 20 0 60 120 180 240 Time from beginning of seizure (seconds) FIGURE 2.2 Some participants believed that they were alone with a student who presumably was having a seizure. Others believed that either one or four more bystanders were present. Participants who believed they were alone with the victim were more likely to intervene and do so more quickly. Adapted from data from Darley & Latané, 1968. 37 certain behaviours and their causes. For example, dozens of experiments reveal that diffusion of responsibility has occurred across many situations (Latané & Nida, 1981). Latané then combined the principle of diffusion of responsibility with other principles of group behaviour to develop a broad theory of social impact, which has been used to explain a variety of social behaviours (DiFonzo et al., 2013; Latané & Bourgeois, 2001). Scientists use theories to develop new hypotheses, which are then tested by conducting more research. In this manner, the scientific process becomes self-correcting. When research consistently supports the hypotheses derived from a theory, confidence in the theory increases. If predictions made by the theory are not supported, then it will need to be modified or, ultimately, discarded. Two Approaches to Understanding Behaviour Humans have a strong desire to understand why things happen. Why do scientists favour the preceding step-by-step approach to understanding behaviour over the approach typically involved in everyday common sense—hindsight? Hindsight (After-the-Fact Understanding) Philosopher Søren Kierkegaard stated, “Life is lived forwards, but understood backwards.” Perhaps the most common method we use to try to understand behaviour in our everyday life is hindsight (i.e., afterthe-fact) reasoning. Indeed, when you report the results of some of the research described in this book, you might hear the following comment: “Big deal— I knew that all along.” However, there is a problem with using hindsight explanations based on common sense and folk knowledge to understand behaviour. For example, suppose two high school sweethearts promised each other undying love before going off to different universities and when they came home for the holidays they broke up. The well-known proverb “Out of sight, out of mind” can explain this result. But, suppose, instead, the opposite happened. When the sweethearts came home, they were more in love than ever and got married. The proverb “Absence makes the heart grow fonder” explains this result. So which proverb is true; or, are there other reasons for their behaviour—for example, the breakup occurred when sweethearts met new partners at university? The main problem with relying solely on hindsight reasoning is that related past events can be explained in many creative, reasonable, and sometimes contradictory, ways. There is no sure way to determine which—if any—of the alternatives is correct. Despite this problem, hindsight reasoning can provide valuable insights and is often the 4. What is a theory? How does it differ from a hypothesis? 5. Explain the major drawback of hindsight understanding. 6. What approach to understanding do scientists prefer? Why? 38 CHAPTER T WO foundation on which further scientific inquiry is built. For example, Darley and Latané’s diffusion of responsibility explanation was initially based on after-the-fact reasoning about the Kitty Genovese murder. Understanding through Prediction, Control, and Theory Building 7. Describe the characteristics of a good theory. Whenever possible, scientists prefer to test their understanding of “what causes what” more directly. If we understand the causes of a given behaviour, then we should be able to predict the conditions under which that behaviour will occur in the future. Furthermore, if we can control those conditions (e.g., in the laboratory), then we should be able to produce that behaviour. Darley and Latané’s research illustrates this approach. They predicted that because of a diffusion of responsibility, the presence of multiple bystanders during an emergency would reduce individual assistance. Next, they carefully staged an emergency and controlled participants’ beliefs about the number of bystanders present. Their prediction was supported. Understanding through prediction and control is a scientific alternative to hindsight understanding. Theory development is the strongest test of scientific understanding because good theories generate an integrated network of predictions. A good theory has several important characteristics: plants would make them sick, without understanding principles of human physiology. But prediction based on understanding (i.e., “theory building”) has important advantages: It satisfies our curiosity, increases knowledge, and generates principles that we can apply to new situations. Defining and Measuring Variables Psychologists study variables and the relations among them. A variable, quite simply, is any characteristic or factor that can vary. People’s sex, height, hair colour, age, income, and grade point average (GPA) are variables: They vary from one person to another, and many also vary within a given person over time. Many variables that psychologists study represent abstract concepts that cannot be observed directly. For example, “self-esteem,” “stress,” and “intelligence” are concepts that refer to people’s internal qualities. We might say that Tyra has high self-esteem, Shaun is intelligent, and Claire feels stressed, but how do we know this? We can’t directly look inside their heads and see “self-esteem,” “stress,” and “intelligence”; yet such concepts must be capable of being measured if we are to study them scientifically. Because any variable may mean different things to different people, scientists must define their • It incorporates existing facts and observations within a single broad framework. In other words, it organizes information in a meaningful way. • It is testable. It generates new hypotheses and predictions whose accuracy can be evaluated by gathering new evidence (Figure 2.3). • The predictions made by the theory are supported by the findings of new research. • It conforms to the law of parsimony: If two theories can explain and predict the same phenomena equally well, the simpler theory is the preferred one. Even when many successful predictions support a theory, it is never regarded as an absolute truth. It is always possible that future observation will contradict it, or that a newer, more accurate theory will displace it. The displacement of old beliefs and theoretical frameworks by new ones is the essence of science (Klahr & Simon, 1999). Finally, although scientists use prediction as a test of “understanding,” this does not mean that prediction requires understanding. Even a child can predict that thunder will follow lightning without knowing why it does so. Our primeval ancestors undoubtedly could predict that eating certain FIGURE 2.3 Is the scientist’s claim of discovering an “eternal life potion” a testable hypothesis? Yes, because it is possible to show the hypothesis to be false. If people drink it but still die, then we have refuted the hypothesis. Therefore, it is testable. It is, however, impossible to absolutely prove true. Even after living for a million years, a person who drank the potion could die the next day. Copyright © 2004 by Sidney Harris. ScienceCartoonsPlus.com . Reprinted with permission. Studying Behaviour Scientifically terms clearly. And when conducting research, scientists must also define variables operationally. An operational definition defines a variable in terms of the specific procedures used to produce or measure it. Operational definitions translate abstract concepts into something observable and measurable. To illustrate, suppose we want to study the relation between stress and academic performance among college students. How shall we operationally define our variables? “Academic performance” could mean a single test score, a course grade, or one’s overall GPA. For our study, let’s operationally define it as students’ final exam scores in an introductory chemistry course. We also have many options for operationally defining exam stress. How might you operationally define “exam stress” at a biological, Levels of Analysis psychological, and environmental level of analysis? Think about this, and then see Figure 2.4. Measurement is challenging because psychologists study incredibly varied and complex processes. Some processes are directly observable, but others are not. Fortunately, psychologists have numerous measurement techniques at their disposal (Figure 2.5). 8. Why are operational definitions important? Self-Reports and Reports by Others Self-report measures ask people to report on their own knowledge, beliefs, feelings, experiences, or behaviour. This information is often gathered through interviews or questionnaires. The accuracy of self-report measures hinges on people’s ability and willingness to respond honestly, especially Measuring Exam Stress Of course, we do not have to limit ourselves to one operational definition of stress or of academic performance. By incorporating multiple levels of analysis, we might ENVIRO ONMENTAL measure students’ pre-exam stress hormones and self-reported worry, their nervous habits during the exam, and the exam’s difficulty. We can then • We W can an me meas asure as re as aspeccts of of the th ac dem aca d ic env enviro ironme iro nment men tha that create t gre te g ate a r examine how these different stress measures relate to immediate exam orr le lesserr deman les m ds on ma o stu studen dents, den t such h as a the performance and students’ overall GPAs. This strategy of measuring a difficulty of the t exa am, the h ov vera e lll course er cou urse gradi ding di ng conceptual variable (i.e., a “construct”) in multiple ways can yield a scale, ti t me pre pr ssures and d room r no oise is du uring i the e much more complete picture than does using a single type of measure. exam, exa m, and th the e leve evell of of achi achi chieve evemen eve m t expecta atio tions set by the sttude uden nts’ pare arents nts. nts BIOLOGICAL • Bef Before ore,, duri du ng, g, an and d afte afterr the the final nal exam, m, we can n ph phys ysiolo ogic gicall allyy mea easur sure e stud stude ents’ str stress es hormo rmone e levels lev e , hea heartt rate rate, resp espira ratio ti n rate rate,, musc m scle le tensio ten on, and nd sw weat eating ing. PSYCHOLOGICAL • Two Tw w weeks ks before the t e fina nal exam, m, w we can an adm mini in sste st r a personality ty te tesst to st studen de ts tha hatt measuress the th ir self-reporte ted, genera er l leve ev l off anxiety ov ver taking exam ms. • Jus Justt before b re e the fi final na ex exam, we e can ask studen nts to o report rep rt th the eir ir level of wo worry, tens ension, and anxiety. y • During th the exam x , we can dir irect ectly ly observe nervouss behavi viour urs, s such as fing ngern er ail biiting, foo foott wigg gling, and d hairr pul pullin ling. If you were designing a research study, what measures would you choose to operationally define exam stress? FIGURE 2.4 39 40 CHAPTER T WO (a) (b) (c) FIGURE 2.5 (a) Self-report, (b) physiological, and (c) behavioural measures are important scientific tools for psychologists. 9. Describe the major ways psychologists measure behaviour, and suggest a limitation of each method. 10. What is unobtrusive measurement? when research questions focus on sensitive topics, such as sexual habits and drug use. Participants’ self-reports may be distorted by a social desirability bias, the tendency to respond in a socially acceptable manner rather than according to how one truly feels or behaves. University of British Columbia researcher Delroy Paulhus (1991) suggests that researchers can minimize the social desirability bias by wording questions so that social desirability is not relevant or, if that is impossible, by guaranteeing respondents anonymity and confidentiality so they can respond honestly without fear of future consequences (e.g., questions about taking drugs or having unsafe sex). These measures presume respondents give honest responses. Paulhus, Harms, Bruce, and Lysy (2003) developed the Over-Claiming Questionnaire (OCQ) to measure the degree of a respondent’s social desirability bias. They had respondents rate their familiarity with a large number of items on a questionnaire, of which 20 percent didn’t exist (e.g., cholarine). The social desirability bias is high when respondents confidently claim familiarity with a large number of nonexistent items. We also can gather information about someone’s behaviour by conducting interviews with or administering questionnaires to other people, such as parents, spouses, and teachers, who know the person. For example, job supervisors might be asked to rate workers’ competence or motivation. As with selfreports, researchers try to maximize participants’ honesty in reporting about other people. Measures of Overt Behaviour Another measurement approach is to record overt (i.e., directly observable) behaviour. In an experiment on learning, we might measure how many errors a person makes while performing a task. In an experiment on drug effects, we might measure people’s reaction time—how rapidly they respond to a stimulus (such as the turning on and off of a light)— after ingesting various amounts of alcohol. In the bystander emergency experiment, Darley and Latané (1968) recorded whether and how quickly college students helped a seizure victim. In experiments on “thinking” in preverbal infants, researchers have measured how long infants look at various familiar and novel visual targets or if they turn to look at offcentred familiar versus novel sound sources. Psychologists also develop coding systems to record different categories of behaviour. While a parent and child jointly perform a task, we might code the parent’s behaviour into such categories as “praises child,” “assists child,” and “criticizes child.” Observers must be trained to use the coding system properly so that their measurements will be reliable—consistent observations. If two observers watching the same behaviours repeatedly disagree in their coding (e.g., one says the parent “praised” and another says the parent “assisted”), then the data are unreliable and of little use. Humans and other animals may behave differently when they know they are being observed. To counter this problem, researchers may disguise their presence or use unobtrusive measures, which Studying Behaviour Scientifically record behaviour in a way that keeps participants unaware that certain responses are being measured. For example, if we ask people to report their mood on a questionnaire, then they are aware that we’re measuring their mood. Instead, we could have people perform tasks that assess their moods in ways that are not obvious to them, such as rating pleasant and emotionally neutral pictures, and reading various types of words (Kiecolt-Glaser et al., 2008). Robinson et al. (2012) report that thermal imaging also is a good predictor for mood state. Psychologists also gather information about behaviour by using archival measures, which are records or documents that already exist. For example, to evaluate the effectiveness of a program to reduce schoolchildren’s disruptive classroom behaviours, researchers have examined school records, which contained such things as student suspensions and number of trips to the principal’s office, that were gathered both before and after the program was implemented (Pelham et al., 2005). Psychological tests. Psychologists develop and use specialized tests to measure many types of variables. For example, personality tests, which assess personality traits, often contain questions that ask how a person typically feels or behaves (e.g., “True or False: I prefer to be alone rather than attend social gatherings.”). In essence, such tests are specialized self-reports. Other personality tests present ambiguous stimuli (e.g., pictures that could have different meanings), and personality traits are judged based upon how a person interprets these stimuli. Other psychological tests consist of performance tasks. For example, intelligence tests may ask people to assemble objects or solve arithmetic problems. Neuropsychological tests help to diagnose normal and abnormal brain functioning by measuring how well people perform mental and physical tasks, such as recalling lists of words or manipulating objects (Abramowitz & Caron, 2010; Goodale & Milner, 1992). Physiological measures. Psychologists also record physiological responses to assess what people are experiencing. Measures of heart rate, blood pressure, respiration rate, hormonal secretions, and brain functioning have long been the mainstay of biopsychologists, but these measures have become increasingly important in many other areas of psychology (see this chapter’s Focus on Neuroscience feature). Physiological responses can have their own interpretive problems, the main one being that we don’t always understand what they mean. For example, if a person shows increased heart rate and brain activity in a particular situation, what emotion or thought is being expressed? Nevertheless, our knowledge about links between patterns of physiological activity and specific psychological processes is rapidly expanding (Rolls, 2010). In sum, psychologists can measure behaviour in many ways, and each has advantages and disadvantages. To gain greater confidence in their findings, researchers may use several types of measures within a single study. METHODS OF RESEARCH Like detectives searching for clues to solve a case, psychologists conduct research to gather evidence about behaviour and its causes. The research In Review • The scientific process proceeds through several steps: (1) asking questions based on some type of observation; (2) gathering information and formulating a testable hypothesis; (3) conducting research to test the hypothesis; (4) analyzing the data, drawing tentative conclusions, and reporting one’s findings to the scientific community; and (5) building a body of knowledge by asking further questions, conducting more research, and developing and testing theories. • In everyday life, we typically use hindsight to explain behaviour. Hindsight is flawed because there may be many possible explanations for behaviour and no way to ascertain which one is correct. Psychologists prefer to test their understanding through prediction, control, and theory building. • A good theory organizes known facts, gives rise to additional hypotheses that are testable, is supported by the findings of new research, and is parsimonious. • An operational definition defines a concept or variable in terms of the specific procedures used to produce or measure it. • To measure behaviour, psychologists obtain peoples’ self-reports and reports from others who know the participants, directly observe behaviour using unobtrusive measures, analyze archival data, administer psychological tests, and record physiological responses. 41 42 CHAPTER T WO method chosen depends on the problem being studied, the investigator’s objectives, and ethical principles. source of new ideas and hypotheses that subsequently may be examined by using more controlled research methods. Case studies have provided important insight into such diverse topics as brain functioning (see this chapter’s Focus on Neuroscience feature), child development, mental disorders, and cultural influences. Consider the following example. Normally, human infants gain weight rapidly after birth; not doing so can have a negative impact on their later physical and intellectual development. When medical causes are ruled out, this “failure to thrive” is related to poverty and/or parenting neglect. Researchers at Surrey Place Centre in Toronto developed a training program for mothers with intellectual disabilities who are at high risk for infant neglect (Feldman, Garrick, & Case, 1997). First, the researchers recorded the child’s weight weekly to diagnose failure to thrive (baseline). Next, the mothers were instructed in feeding and nutrition for six weeks (treatment). Finally, the child’s weight was recorded over the next three years (follow-up). The results are shown in Figure 2.6. Clearly, during the baseline, this child remained small relative to the general population, rapidly gained weight to a safe level during treatment, and continued to thrive for the next few years during the follow-up, suggesting that the treatment program was very effective. Case studies have several limitations. First, they are a poor method for determining cause–effect relations. In our failure-to-thrive case study, the Descriptive Research: Recording Events The most basic goal of science is to describe phenomena. In psychology, descriptive research seeks to identify how humans and other animals behave, particularly in natural settings. It provides information about the diversity of behaviour and may yield clues about potential cause–effect relations that are later tested experimentally. Case studies, naturalistic observation, and surveys are research methods commonly used to describe behaviour. Case Studies: Treating Cases of Failure to Thrive (Starvation) in Human Infants A case study is an in-depth analysis of an individual, a group, or an event. By studying a single case in detail, researchers typically hope to discover principles of behaviour that are true for people or situations in general. Data may be gathered through observation, interviews, psychological tests, physiological recordings, and task performance, or from archival records. Case studies have several advantages. First, when a rare phenomenon occurs, this method enables scientists to study it closely. Second, a case study may challenge the validity of a theory or widely held scientific belief. Third, a case study can be a vibrant Change in weight 2.0 Child’s weight in kilograms where “0” = percentile 11. What is a case study? Identify its advantages. Baseline (family physician) 1.5 Treatment (parent training and physician) Follow-up 1.0 0.5 0 –0.5 –1.0 –1.5 –2.0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 44 46 85 Age of child (months) FIGURE 2.6 An example of a case study of a failure-to-thrive infant who stopped growing shortly after birth. Weight on the y-axis is given in kilograms relative to the 5th percentile (95 percent of the infants of that age are above the horizontal line at 0 on the y-axis). During the Baseline, the infant was 0.8 kg below the 5th percentile; but, during Treatment, when the parent received training in feeding and nutrition, the infant rapidly gained weight to a safe level. The infant remained at a normal weight for the next few years (Follow-up). Figure adapted from Feldman, Garrick, & Case, 1997. Studying Behaviour Scientifically 43 Focus on Neuroscience THE NEUROSCIENCE OF THE HUMAN BRAIN AT WORK Neuroscientists use various techniques to identify localization of behavioural function in specific areas of the brain. They have used the case study method to test patients with damage to a specific area of the brain to uncover which of their abilities are preserved and which are lost, in an attempt to identify the functions of different brain structures (e.g., see Chapter 3; classic study of Phineas Gage). A modern example is the work of Mel Goodale, at the University of Western Ontario. In the 1990s, Goodale and Milner (1992) studied a patient (D.F.) who experienced carbon monoxide poisoning. D.F. suffered damage to one cortical visual processing area (the ventral stream) but had no damage in another cortical visual area (the dorsal stream; see Figure 2.7). When D.F. was shown objects (e.g., a rod), she couldn’t identify them, their shape, or their orientation (i.e., she had lost visual object recognition). But when D.F. was asked to grab the rod presented in different orientations and positions, she showed normal anticipatory hand opening, rotated her hand into the correct orientation, accurately reached out and grabbed the rod (i.e., she had retained normal visually guided reaching). The case study of D.F. provided compelling evidence from purely behavioural data that visual object recognition and action are processed independently by the ventral and dorsal streams, respectively. As discussed in Chapter 1, recent advances in brainimaging technology have allowed neuroscientists to monitor neural activity in the intact brain of a person during mental or physical tasks (e.g., Talbot, 2003). PET and fMRI scans actually measure changes in local blood flow or oxygen content, which have been shown to reflect local neural activity (Logothetis et al., 2001). Neuroscientists have used brain-imaging technology to map the neural activity of clinical patients with psychiatric disorders (e.g., schizophrenia, Alzheimer’s disease) and patients suffering from brain damage (e.g., Thompson et al., 2000; Partain, 2006), as well as to explore the development of normal and abnormal brain functions. Using imaging techniques, scientists have identified neural pathways treatment may have caused the infant’s weight gain; alternatively, some other change in the mother’s and infant’s lives could have been responsible, or the symptoms may have ended simply because of the passage of time. Second, case study findings may not generalize to other people or situations. In the failure-to-thrive case study, perhaps the treatment was only beneficial for that particular mother. To establish the generalization of a principle (e.g., involved in various mental operations, emotional regulation, language perception and production, and visual perception and action. For example, James, Culham, Humphrey, Milner, and Goodale (2003) used fMRI scans to measure the activation level in D.F.’s ventral and dorsal streams during both visual recognition tests and object-directed grasping tasks. As predicted from the behavioural results, D.F.’s loss of visual object recognition was associated with absent/abnormal ventral-stream activation, while her dorsal stream regions showed normal activation during object grasping tasks. Finally, Valyear, Culham, Sharif, Westwood, and Goodale (2006) used fMRI to study the activity in the normal human brain and found the predicted differential ventral and dorsal stream activation during the performance of visual object recognition tasks versus object grasping (orientation) tests. Clearly brain-imaging technology gives neuroscientists a powerful tool to study the localization of function. Dorsal Dorsal stream Anterior Posterior Ventral stream Ventral FIGURE 2.7 Navigating the brain. When discussing anatomy, we often use the standard terminology shown here. Thus, the dorsal stream runs along the upper surface of the cortex; the ventral stream runs along the bottom surface. In addition, we often use the terms medial to describe a structure toward the centre of the brain and lateral to indicate one toward the brain’s outer surface. maternal education can reduce instances of failure to thrive), investigators must conduct more case studies, use other research methods, and test a variety of cultural groups. Third, observers may not be objective in gathering and interpreting the data. While this issue may not apply to our failure-tothrive example (i.e., weight measures are relatively objective), measurement bias (also called observer bias) can occur in any type of research. Case studies 12. What are the major limitations of case studies? 44 CHAPTER T WO are particularly worrisome because often they are based on an observer’s subjective impressions. A skeptical attitude requires that claims based on case studies be followed up by more controlled methods before they are accepted. In everyday life, we should adopt a similar skeptical view. When encountering claims based on a case example or an anecdote, remember that the case may be atypical or the person making the claim may be biased. Try to seek out other evidence to evaluate the claim. Naturalistic Observation: Bullying in Canadian Schoolyards 13. What is naturalistic observation, and what is its major advantage? 14. What problems can occur when conducting naturalistic observations? In naturalistic observation, the researcher observes behaviour as it occurs in a natural setting, and attempts to avoid influencing that behaviour (Figure 2.8). For example, by observing African chimpanzees in the wild, British researcher Jane Goodall (1986) and other scientists found that chimpanzees display behaviours, such as making and using tools, that were formerly believed to lie only within the domain of human capabilities (Lonsdorf, 2006). Naturalistic observation is also used to study human behaviour. Consider bullying in schools, a topic that has received increasing attention from psychologists (Kanetsuna, Smith, & Morita, 2006). Were you ever bullied at school? If so, did any schoolmates step in to help? In a three-year study, psychologists recorded children’s playground interactions during recess and lunch periods at two elementary schools in Toronto (Hawkins, Pepler, & Craig, 2001). Their main goal was to describe peer interventions during episodes of schoolyard bullying. How often do schoolmates intervene? What strategies do they use? Are peer interventions effective? To answer these questions, the researchers developed coding systems so that the children’s behaviour could be classified into meaningful categories. To illustrate, here are three of ten categories representing different intervention strategies: • Verbal Assertion: Verbally requesting that the bullying stop (e.g., “Stop it,” “Knock it off ”) • Physical Assertion: Physically separating the bully and victim, but not physically attacking either one • Physical Aggression: Hitting, pushing, shoving, or otherwise physically engaging the bully or victim Overall, of the 306 bullying episodes observed, schoolmates were present 88 percent of the time but intervened in only 19 percent of the episodes. In order, the three most common types of intervention were verbal assertion alone, physical aggression alone, and verbal assertion combined with physical assertion. Like case studies, naturalistic observation does not permit clear causal conclusions. In the real world, many variables simultaneously influence behaviour, and they cannot be disentangled with this research technique. Bias in how researchers interpret what they observe is also possible. Finally, even the mere presence of an observer may disrupt a person’s or animal’s behaviour. Thus, researchers may disguise their presence so that participants are not aware of being observed. Fortunately, when disguise is not feasible, people and other animals typically adapt to and ignore the presence of an observer as time passes. This process is called habituation, and researchers may delay their data collection until participants have habituated to the observers’ presence. Survey Research: Does Your Own Personality Match the Canadian National Character? In survey research, information about a topic is obtained by administering questionnaires or interviews to many people. Political polls are a well-known example, but surveys also ask about participants’ behaviours, experiences, and attitudes on wide-ranging issues. For example, Terracciano et al. (2005) addressed the following question: Does the stereotypical “national character” of a culture (with regard to neuroticism, extraversion, openness to experience, agreeableness, and conscientiousness) actually match the averaged scores for individual members of that culture on those personality characteristics? Eighty-six researchers administered two questionnaires (one measured “national character”; the other measured the respondent’s own personality) to individuals from 49 cultures around the world, including students at York University (by L.E. Ayearst), the University of British Columbia (by D.L. Paulhus), and the University of Winnipeg (by P.D. Trapnell). Respondents’ responses showed that there was a consensus on each culture’s national character. However, the national character personality profile was significantly different from the averaged personality scores of individual members on the same characteristics. To illustrate, the national characters of the Canadian and U.S. samples were similar for neuroticism (e.g., anxiety, hostility, depression, impulsiveness) and agreeableness (e.g., altruism, compliance, modesty); but the averaged individuals’ ratings on themselves were much higher for agreeableness and much lower on neuroticism for Canadians than for the U.S. respondents, and both profiles were significantly different from their respective national characters. Terracciano et al. concluded that while the cultural stereotype of a national character may define national identity, Studying Behaviour Scientifically FIGURE 2.8 Psychologists conduct naturalistic observations in many settings, including the schoolyard. it does not reflect the actual, assessed personality traits of members of that culture. Terracciano et al. surveyed only 3989 adults. So how is it possible to obtain accurate estimates of the stereotypic national character of various cultures? Two key concepts in survey research are population and sample. A population consists of all the individuals about whom we are interested in drawing a conclusion. Terracciano et al. wanted to know the stereotypic national character and actual personality characteristics of adult populations in 49 cultures. Clearly, it would be impossible to study everyone. Therefore, they surveyed a sample, that is, a subset of individuals drawn from the larger population of interest. To draw valid conclusions about a population from a survey, the sample must be representative: A representative sample is one that reflects the important characteristics of the population (Figure 2.9). A sample composed of 80 percent males would not represent a student body in which only 45 percent are men. To obtain a representative sample, survey researchers typically use a procedure called random sampling, in which every member of the population has an equal probability of being chosen to participate in the survey. A common variation of this procedure, called stratified random sampling, is to divide the population into subgroups based on such characteristics as gender or ethnic identity. If the population is 45 percent male, then 45 percent of the spaces in the sample would be allocated to men and 55 percent to women. Random sampling is then used to select the individual women and men who will be in the survey. When a representative sample is surveyed, we can be confident (though never completely certain) that the findings closely portray the population as a whole. This is the strongest advantage of survey research. Modern political opinion polls use such excellent sampling procedures that, just prior to elections, they can reasonably predict who will win a national election from a sample of about 1000 people. In contrast, unrepresentative samples can produce distorted results. Other things being equal, large samples are better than small ones, but it is better to have a smaller representative sample than a larger, unrepresentative one. A famous example is a mail survey of almost two million voters in 1936, which was carried out by Literary Digest magazine. This survey predicted that U.S. Republican presidential candidate Alf Landon would easily defeat Democratic candidate Franklin Roosevelt. When the election took place, Roosevelt won in a landslide! How could a prediction based on two million people be so massively wrong? The answer is that the survey’s sample was unrepresentative of the population that actually voted. The researchers obtained names and addresses from telephone directories, automobile registration lists, and magazine subscription lists. In 1936, most poor Americans did not have telephones, cars, or magazine subscriptions. Thus, the sample under-represented poorer socioeconomic groups and over-represented wealthier people: bad sample, bad prediction. In sum, always consider the nature of the sample when interpreting survey results. The World Wide Web revolution in the 1990s produced a massive electronic interconnection of people around the world on the Internet, allowing psychologists to collect questionnaire data from thousands of subjects quickly and at virtually no cost compared with paper-based questionnaires and telephone surveys (see Kraut et al., 2004). For example, Nosek, Banaji, and Greenwald (2002) measured attitudes toward and stereotypes of social groups of over 1.5 million “drop-in” respondents at their website who were recruited through news media, links from other Internet sites and search engines, and word of mouth. Internet questionnaires can be problematic because researchers do not have much control over data quality; respondents can lie about their ages, identities, and genders, and anonymity permits respondents to answer frivolously or maliciously. Also, sample bias can occur because, unlike randomly dialing telephone numbers, there is no method for randomly sampling the population of Internet users. Kraut et al. noted that Internet users in 2002 were more likely than the general U.S. population to be young, to be white, and to have children. Gosling, Vazire, Srivastava, and John (2004) evaluated these concerns by comparing personality questionnaire data from over 360 000 Internet respondents with that from 510 publications where 45 15. Explain what representative sampling is, and why survey researchers use it. 16. What are some advantages and disadvantages of survey research? 46 CHAPTER T WO Unrepresentative sample Population E F G H D A C B A H B G F Sampling procedure C E H D A B G Representative sample F C E D FIGURE 2.9 A representative sample possesses the important characteristics of the population in the same proportions. Data from a representative sample are more likely to generalize to the larger population than data from an unrepresentative sample. traditional questionnaires were used. Gosling et al. acknowledged that Internet samples are not without flaws—their Internet sample was not representative of the general population—but their sample was more representative than the undergraduate psychology student samples used in many questionnaire studies. Moreover, Internet survey results have been shown to be less influenced by issues such as missing data and socially desirable responding than those of paper-based surveys (Truell, Bartlett, & Alexander, 2002; Wood, Nosko, Desmarais, Ross, & Irvine, 2006), suggesting that the Internet can be a useful research tool. In scientific research, surveys are an efficient method for collecting a large amount of information about people’s opinions, experiences, and lifestyles, and they can reveal changes in people’s beliefs and habits over many years. But there also are several major drawbacks to surveys. First, survey data cannot be used to draw conclusions about cause and effect. Second, surveys rely on participants’ selfreports, which can be distorted by social desirability bias, interviewer bias, peoples’ inaccurate perceptions of their own behaviour, and misinterpretation of survey questions. Third, unrepresentative samples can lead to faulty generalizations about how an entire population would respond. And finally, even when surveys use proper random sampling procedures, once in a while—simply by chance—a sample that is randomly chosen will turn out not to be representative of the larger population. Overall, in properly conducted professional and scientific surveys, this happens less than 5 percent of the time, but it does happen. Note, in this chapter, while we give examples of descriptive statistics (e.g., line graphs in Figures 2.2 and 2.6, pie graphs in Figure 2.9, and percentage differences among groups), we focus on experimental design of studies rather than statistical procedures. A comprehensive introduction to statistical methods relevant to survey results and group observations are covered in the Appendix following Chapter 17. Thinking critically SHOULD YOU TRUST INTERNET AND POP MEDIA SURVEYS? Tom fills out a political-attitude survey posted on the Internet. Claire mails in a dating-satisfaction survey that came in a fashion magazine to which she subscribes. Sam responds to a local TV news phone-in survey on a tax issue (Call our number, press 1 to agree, 2 to disagree). For each survey, can the results be trusted to reflect the general public’s attitudes? Think about it, and then see the Answers section at the end of the book. Correlational Research: Measuring Associations between Events What factors distinguish happily married couples from those headed for divorce? Do first-born children differ in personality from later-born children? Is monetary wealth related to happiness? These and countless other psychological questions ask about Studying Behaviour Scientifically 47 In Review • The goal of descriptive research is to identify how organisms behave, particularly in natural settings. Case studies involve the detailed study of a person, group, or event. Case studies often suggest important ideas for further research, but they are a poor method for establishing cause–effect relations. • Naturalistic observation can yield rich descriptions of behaviour in real-life settings and permits examination of relations between variables. Researchers must avoid influencing the participants they observe. associations between naturally occurring events or variables. To examine such relationships, scientists typically conduct correlational research, which in its simplest form has three components: 1. The researcher measures one variable (X), such as people’s birth order. 2. The researcher measures a second variable (Y), such as a personality trait. 3. The researcher statistically determines whether X and Y are related. Remember that correlational research involves measuring variables, not manipulating them. Naturalistic observation and surveys often are used not only to describe events, but also to study associations between variables. For example, in the naturalistic observation study of schoolyard bullying, the researchers examined associations between the children’s sex and peer intervention (Hawkins et al., 2001). They found that girls were more likely to intervene when the bully and victim were female, and boys were more likely to intervene when the bully and victim were male. Other types of studies also fall under the correlational umbrella, as this chapter’s Research Foundations feature illustrates. Correlation Does Not Establish Causation As described in the Research Foundations feature, Diener and Seligman (2002) found that very happy people had stronger, more satisfying social relationships than unhappy people (Figure 2.10a). It is tempting to conclude from these findings that stronger social relationships cause people to be happier, but correlational research does not allow us to draw such a conclusion. First, the direction of causality could be opposite; perhaps being happy causes • Surveys involve administering questionnaires or interviews to many people. Most surveys study a sample of people that is randomly drawn from the larger population. Representative samples allow researchers to estimate the responses of the entire population. Unrepresentative samples can lead to inaccurate estimates. Survey results also can be distorted by interviewer bias or biases in the way participants report about themselves. people to have stronger social relationships. For example, maybe happiness makes a person more receptive to going out and forming close relationships. In correlational research, you must consider the possibility that variable X (social relationships) has caused variable Y (happiness), that Y has caused X, or that both variables have influenced each other. This interpretive problem is called the bidirectionality (i.e., two-way causality) problem (Figure 2.10b). Second, the association between social relationships and happiness may be artificial, or what scientists call spurious (not genuine). Although social relationships and happiness are statistically related, it may be that neither variable has any causal effect on the other. A third variable, Z, may really be the cause of why some people have better social relationships and also why those people are happier. For example, Z might be a certain personality style. Recall that very happy people in Diener and Seligman’s study were, in general, more outgoing and agreeable and tended to worry less. Perhaps this personality style makes it easier for people to establish good social relationships. At the same time, this style may help people soak up more joy from life and therefore feel happier. Thus, on the surface it looks as if social relationships and happiness are causally linked, but in reality this may be due to only Z (in this case, personality style). This interpretive problem is called the thirdvariable problem: Z is responsible for what looks like a relation between X and Y (Figure 2.10c). As Z varies, it causes X to change. As Z varies, it also causes Y to change. The net result is that X and Y change in unison, but this is caused by Z, not by any direct effect of X or Y on each other. In sum, we cannot draw causal conclusions from correlational data, which is the major disadvantage of correlational research. 17. Explain the main goal of correlational research and how it is achieved. 18. Why are we unable to draw causal conclusions from correlational findings? 48 CHAPTER T WO (a) Social relationships and happiness are correlated Greater happiness (Y) Better social relationships (X ) Thinking critically (b) Bidirectionality problem Does X cause Y ? Greater happiness (Y) Better social relationships (X ) Does Y cause X ? Greater happiness (Y) Better social relationships (X ) There may be no causal relation between X and Y DOES EATING ICE CREAM CAUSE PEOPLE TO DROWN? Nationally, ice cream consumption and drownings are positively correlated. Over the course of the year, on days when more ice cream is consumed, there tend to be more drownings. Are these two variables causally related? What causal possibilities should you consider? Think about it, and then see the Answers section at the end of the book. (c) Third-variable problem Better social relationships (X ) being fired). Likewise, students’ test anxiety and exam performance are negatively correlated (i.e., students with higher levels of test anxiety tend to perform more poorly on exams). Greater happiness (Y) Personality style (Z) FIGURE 2.10 (a) Students who have better social relationships are happier. But why does this association occur? (b) Good social relationships could cause people to become happier or, conversely, being a happier person could