Chapter 2

Cigarettes and Disease: Is there a causal relationship?

- '94, representatives from all major tobacco companies came before US Congress House Subcommittee on Health to argue that cigs don't cause health probs such as heart disease/lung cancer - Crux of argument was that no scientific study had ever proven that cig smoking causes these conditions in humans; technically right, only experimental studies can absolutely demonstrate causation, and no such experimental study has ever been/ever will be conducted on humans - During past 50 yrs tho, researchers have used nonexperimental studies to est a link b/w cig smoking and several diseases, esp CVD/lung cancer - Experimental, randomized, placebo-controlled, double-blind studies not required before scientists can infer a causal link b/w the IV (smoking) and the DV (heart disease/lung cancer); draw conclusions that a causal relationship exists if certain criteria are met

Real-World Profile of Sylvester Colligan

- 76 y/o man who had trouble with his right knee for 5 years; doctor diagnosed arthritis by had no treatment that would help - Same physician told him abt an experimental study conducted by Dr. J Bruce Moseley; treatment worked - Two years after the surgery, Colligan reported that his knee hadn't bothered him since - Improvement would not be so surprising, except for one thing: Moseley didn't perform surgery on Colligan - Instead, Moseley gave him anesthesia, made some cuts around Colligan's knee that looked like surgical incisions, then sent Colligan on his way home

Summary thus far

- A placebo is an inactive substance/condition having the appearance of an active treatment; may cause participants in experiment to improve/change beh b/c of their belief in placebo's effectiveness and prior experiences w/ receiving effective treatment - Tho placebos can have a positive effect from the patient's POV, they are a continuing problem for research - Generally, a placebo's effects are ~35% -- effects on reducing pain may be higher whereas its effects on other conditions may be lower - Placebos can influence a wide variety of disorders/diseases - Experimental designs that measure the efficacy of an intervention typically use a placebo so that ppl in control group (who receive placebo) have same expectations for success as do ppl in experimental group - Drug studies usually double-blind - Researchers in psych treatment studies are often not 'blind' concerning the treatment, but participants are (single-blind)

Summary thus far

- Epidemiologists conduct research using designs and terminology that differ from those used by psych researchers; use concepts of risk factor, prevalence, and incidence - Risk factor: any condition that occurs w/ greater frequency in ppl w/ a disease than it does in ppl free from that disease - Prevalence: Proportion of population that has a particular disease at a specific time - Incidence: Measures frequency of new cases of the disease during a specified period of time - Use research methods similar to those used by psychs, but terminology varies - Observational studies similar to correlational studies can be prospective/retrospective - Randomized controlled trials similar to experimental designs in psych - Clinical trials, a common type of RCT, typically used to determine effectiveness of new drugs but can be used in other controlled studies - Natural experiments, which are similar to EPF studies used when naturally occurring conditions allow for comparisons - Stat technique of M-A allows researchers to examine group of studies that have researched the variable of interest and provide an overall estimate of the size of the effect

Summary thus far

- A risk factor is any characteristic/condition that occurs w/ greater frequency in ppl w/ a disease than it does in ppl free from that disease; may be expressed either in terms of the absolute risk (a person's risk of developing a disease independent of other factors) or the relative risk (ratio of risk of those exposed to a risk factor compared w/ those not exposed) - Although risk factor approach alone can't determine causation, epidemiologists use several criteria for determining C/E relationship between a condition and a disease: 1. A dose-response relationship must exist b/w the condition and the disease 2. The removal of the condition must reduce the prevalence or incidence of the disease 3. Condition must precede disease 4. The causal relationship b/w the condition and the disease must be physiologically plausible 5. Research data must consistently reveal a relationship b/w the condition and the diease 6. Strength of the relationship b/w condition and disease must be relatively high 7. Relationship b/w condition and disease must be based on well-designed studies - When findings meet all 7 criteria, scientists can infer a C/E relationship b/w IV and DV

Treatment and the Placebo

- Beecher observed effects of placebos on a variety of conditions ranging from headaches to the common cold; concluded that the therapeutic effect was substantial (~35% of patients showed improvement) - Recent review of 100s of research studies confirms that placebos can lead to noticeable improvements in health outcomes, especially in the context of pain/nausea - Likelihood of improvement largely attributable to the type of pain experienced - Some researchers argue that the p/e is responsible for much of the effectiveness of antidepressant drugs, especially among ppl with mild-moderate symptoms - Some conditions, like broken bones, aren't responsive to placebos - More a placebo resembles an effective treatment, the stronger the placebo effect - Big pills are more effective than little ones, and colored pills work better than white tablets - Capsules work better than tablets, and placebos labeled with brain names work better than generics - 2 doses provoke a larger placebo response than 1 - Injection is more powerful than a pill, and surgery tends to prompt an even larger placebo response than injection - Even cost matters; more expensive ones work better than cheaper pills - Physician/patient expectations also strengthen placebo effects - Placebo responses also relate to other characteristics of the practitioner, such as their reputation, attention, interest, concern, and confidence projected that a treatment will be effective - When researchers tell patient-participants that they could be receiving either a placebo or a real treatment, placebo effects tend to weaken; participant less confident abt receiving an effective treatment - However, if participants CHOOSE which treatment to receive, placebo effects are stronger than those who don't choose placebo - Effect both have a physical and psychological basis; a placebo analgesic alters lvls of brain activity in ways that are consistent with the activity that occurs during pain relief from analgesic drugs - Reduce/cure a remarkable range of disorders/symptoms, including insomnia, low back pain, burn pain, headache, asthma, hypertension, and anxiety - Findings suggest that the underlying phys mechanisms for placebo responses are the same as for drug treatments - Drugs that block the action of analgesic drugs also block the placebo response to analgesic drugs - Can alter neurotransmitters, hormones, and endorphins, potentially producing a variety of perceptual, beh, and phys effects - Can also produce adverse effects called the nocebo effect; nearly 20% of healthy volunteers given a placebo in a double-blind study experienced some negative effect as a result of the nocebo effect - Sometimes these negative effects appear as side effect, which show the same symptoms as other drug side effects such as headaches, nausea/other digestive problems, dry mouth, and sleep disturbances - Presence of negative effects demonstrated that the placebo effect is not merely improvement but includes any change resulting from an inert treatment - Nocebo response also activates specific areas of the brain and acts on neurotransmitters, giving additional support to physical reality - Expectancy is a major component of the placebo effect; people act in ways they THINK they should -- thus ppl who receive treatment w/o their knowledge don't benefit as much as those who know what to expect - Culture influences that placebo response; cultures that place greater faith in medical interventions show stronger responses to placebos that resemble a medical intervention - Learning and conditioning also factors in placebo response - Thru classical/operant conditioning, ppl associate a treatment with getting better, creating situations in which receiving treatment leads to improvements - When patients' positive expectations increase their chances for improvement, the placebo is a valuable adjunct in treatment - P/e are a tribute to the ability of humans to heal themselves, and practitioners can enlist this ability to help patients become healthier - P/e can be a positive factor in med and beh therapies

Cross-Sectional and Longitudinal Studies

- Both approaches to studying developmental issues/understanding how health problems develop over time - Cross-sectional: conducted at only one point in time; studies group of ppl from at least 2 different age groups to determine the possible differences b/w the groups on some measure - Longitudinal: follow participants over an extended period of time; can yield info that C-S studies cannot BECAUSE they assess same people over time, allowing researchers to ID developmental trends and patterns - Drawbacks to longitudinal studies are that they take time, thus making them more costly and frequently require a large team of researchers - Disadvantages to C-S studies compare 2+ separate groups of individuals, which make them incapable of revealing info abt changes in individuals over time

Determining Causation

- Both prospective/retrospective studies can ID risk factors for a disease, by they don't demonstrate causation - Obesity is a risk for hypertension, heart disease, diabetes, and kidney disease; obese ppl are more likely to develop these conditions than ppl who are normal weight - However, some ppl who aren't obese/overweight develop heart and kidney diseases - Risk factor approach as a means of suggesting causation and then examines evidence that cigarette smoking causes disease

Prescribing Placebos May Be Considered Ethical

- Cebocap -- a capsule only available by prescription; can be remarkably effective in relieving a wide range of health problems w/ few serious side effects; a placebo - Though it's unclear how often physicians prescribe Cebocap, many doctors already report prescribing treatments that they consider to be placebos, such as vitamins or antibiotics for a viral infection - Nearly 3/4 of doctors who admit to prescribing a placebo describe it simply as "medicine not typically used for your condition but might benefit you"; this is truthful and preserves the active ingredient of placebos: positive expectations - Critics of this practice argue that the physician is deceiving the patient by withholding the fact that the treatment has no inherent medical benefit - Could a placebo still be effective if the provider fully informed the patient about the nature of the treatment? - Study on IBS; given few effective treatments available for IBS, many view it as ethically permissible to study effects of placebos on IBS symptoms - In one experimental condition, researchers told patients to take placebo pills 2x daily, described them and stated effectiveness; patients in control condition didn't receive any treatment at all - The placebo treatment, even when prescribed in this completely transparent manner, led to fewer symptoms, greater improvement, and better quality of life compared with no treatment; placebos can be both ethically prescribed and effective in treatment - Despite value of placebo controls in clinical research, some physicians and med ethicists consider the use of ineffective treatment to be ethically unacceptable b/c the welfare of patients is not the primary concern; a valid worry if a patient-participant receives a placebo instead of the accepted standard care - Opinion regarding the ethical acceptability of placebo treatment is divided; some find it acceptable and necessary for research and others object the failure to provide an adequate standard of treatment

Research and the Placebo

- Effects present problems in evaluating treatment effectiveness - To conclude that a treatment is effective, it must show a higher rate of effectiveness than a placebo; otherwise, any effects of treatment are probably due only to ppl's beliefs abt the treatment - This standard calls for researchers to use at least 2 groups in a study: one that receives treatment and another that receives a placebo - Both groups must have equal expectations concerning the effectiveness of the treatment; in order to creat equal expectancy, not only must the participants be ignorant of who is getting what, but the experimenters who dispense both conditions must also be 'blind' to which group is which (double-blind design); this can create ethical dilemmas - Psych treatments such as counseling, hypnosis, biofeedback, relaxation training, massage, and a variety of stress/pain management techniques also produce expectancy effects - P/e also applies to research in psych, but double-blind designs aren't easy to perform w/ these treatments - Placebo pills can look the same as pills containing an active ingredient, but providers of psych or beh treatments always know when they are providing a shame treatment while participants don't know if they are receiving active or inactive pulls(single-blind design) - In single-blind designs, control for expectancy is not as complete as in double-blind designs, but creating equal expectancies for participants is usually the more important control feature

Research Methods in Epidemiology

- Epidemiology is a branch of med that investigates factors that contribute to health/disease in a particular population - W/ increase in chronic diseases during 20th century, epidemiologists make fundamental contributions to health by ID'ing risk factors for diseases - A risk factor is any characteristic or condition that occurs w/ greater frequency in ppl w/ a disease than in ppl free from that disease - Epidemiologists study demographic and behavioral factors that relate to heart disease, cancer, and other chronic diseases - Such studies were the first to detect relationship b/w smoking and heart disease - Two important concepts in epidemiology: 1. Prevalence -- refers to proportion of the pop that has a particular disease/condition at a specific time 2. Incidence -- measures frequency of new cases during a specified period, usually one year - W/ both concepts, the number of ppl in the population at risk is divided into either the number of ppl with the disorder (prevalence) or the number of new cases in a particular time frame (incidence) - Prevalence of a disorder may be quite different from the incidence of it - Research in epidemiology uses three broad methods: 1. Observational studies 2. Randomized, controlled trials 3. Natural experiments - Each has its own requirements and yields specific info - Fig 2.2

The Alameda County Study

- Famous example of an epidemiological study - Ongoing prospective study of a single community to ID health practices that may protect against death/disease - Epidemiologists ID risk factors by studying large pops over some period of time and by sifting out beh, demographic, or phys conditions that show a relationship to subsequent disease/death - Lester Breslow and colleagues from Human Pop Lab of California State Department of Public Health began a survey of a sample of households in Alameda County (Oakland), CA - After determining # of adults at these addresses, researchers sent detailed questionnaires to each resident 20+ y/o; nearly 7,000 ppl returned surveys - Answered questions (among others) abt 7 basic health practices: 1. Getting 7/8 hrs of sleep daily 2. Eating breakfast almost every day 3. Rarely eating between meals 4. Drinking alcohol in moderation/not at all 5. Not smoking cigarettes 6. Exercising regularly 7. Maintaining weight near the prescribed ideal - At time of original survey in '65, only cigarette smoking had been implicated as a health risk; evidence that any other six practices predicted health/mortality was weak - To control for possible confounding effects (original health status possibly confounding subsequent death rates), investigators asked residents abt their phys disabilities, acute/chronic illnesses, phys symptoms, and current lvls of energy - Follow-up 5 yrs later revealed residents who practiced six or seven of basic health-related behaviors were far less likely to have died than those who practiced zero to three; this decreased mortality risk was independent of their '65 health status, suggesting that health behs lead to lower rates of death - '74, investigators conducted a major follow=up of living participants and also surveyed a new sample to determine whether the community in general had adopted a healthier lifestyle b/w '65 and '74; 9yr followup determined relationship b/w mortality and the 7 health practices, considered individually and in combination - Cigarette smoking, lack of phys activity, and alcohol consumption strongly related to mortality, whereas obesity and too much/too little sleep were only weakly associated with increased death rates - Skipping breakfast and snacking b/w means didn't significantly predict mortality - Number of close social relationships also predicted mortality: ppl w/ few social contacts were 2.5 times more likely to have died than those with many such contacts - Alameda County Study was the first study to uncover a link b/w social relationships/mortality -- social relationships/support are important factors in phys health - If some health practices predict greater risk of mortality, then a second question concerns how these same factors related to morbidity, or disease - A condition that predicts death need not also predict disease; many disabilities, chronic illnesses, and illness symptoms don't inevitably lead to death - Therefore, it's important to know whether basic health practices and social contacts predict later phys health - Do health practices merely contribute to survival time, or do they also raise an individual's general lvl of health? - To answer this, researchers studied a subset of the original sample of Alameda County participants - In addition to the 5 health practices that related to mortality, this investigation included a Social Network Index that combined marital status, contacts w/ friends/relatives, and membership in church and other organizations - Each of 5 health behs as well as the SNI showed a relationship to changes in health; more specifically, ppl w/ the best health were: 1. Those who didn't smoke 2. Those who drank alcohol moderately 3. People who slept 7 or 8 hours per night 4. People at a normal weight 5. People who scored high on the SNI

The Role of Psychometrics in Research

- Health psych study a number of phenomena that can't be described in terms of simple physical measurements, like weight/length; phenomena include behs and conditions such as stress, coping, pain, hostility, eating habits, and personality - To study each of these phenomena, health psychs must develop new measures that can reliably and validly measure differences b/w ppl - One of psych's most important contributions to beh med and beh health is its sophisticated methods of measuring important psych factors in health - For any measuring instrument to be useful, it must be both reliable (consistent) and valid (accurate); Establishing both is critical to the development of any measurement scale

Summary thus far

- Health psychs use several research methods, including correlational studies, cross-sectional and longitudinal studies, experimental designs and EPF studies - Correlational indicates degree of association b/w two variables, but can never show causation - C-S investigates a group of ppl at one point in time - Longitudinal participants investigated over extended period; though will yield more useful results than C-S, they're time consuming and expensive - Experimental designs, researchers manipulate IV so that any resulting differences b/w experimental and control groups can be attributed to their differential exposure to the IV; typically include a placebo for control group - EPF studies are similar in that researchers compare 2+ groups then then record group differences in DV, but differ in that the IV is preexisting rather than manipulated

Becoming an Informed Reader of Health-Related Research on the Internet

- Internet presents wealth of useful health info; over 80% of users turn to internet for health info - Several questions to ask yourself as you evaluate health information on the internet: 1. Who runs the website? 2. What is the purpose of the website? 3. What is the evidence supporting a claim? 4. Is there adequate info available to evaluate the research design of a scientific study? 5. How is the information reviewed before it's posted? 6. How current is the info? - NIH and CDC websites excellent source for the latest scientific info on a wide variety of health topics

Moseley's Study

- Moseley and colleagues were conducting a study of the effectiveness of arthroscopic knee surgery; procedure widely performed but expensive, and Moseley had doubts about the effectiveness - Decided to perform an experimental study that included a placebo as well as a real arthroscopic surgery - A placebo is an inactive substance/condition that has the appearance of an active treatment; may cause participants to improve/change because of their belief in the placebo's efficacy - Moseley suspected that this type of belief and not the surgery was producing improvements, so designed a study where half of participants received sham knee surgery - Participants agreed to be in either group, knowing they might receive sham surgery; including Colligan, they didn't know for several years whether they were in placebo or arthroscopic surgery group - Discovered that surgery provided no real benefits beyond a placebo effect; those who received sham surgery reported same lvl of knee pain/functioning as those who received the real surgery - These findings join those of hundreds of other studies: A belief in the effectiveness of a treatment boosts the treatment's effectiveness - Though, this effect presents a problem for researchers who want to determine which effects are due to treatment and which are due to beliefs abt treatment

Research Methods in Psychology

- Not long ago, reports of health research appeared mainly in scientific journals, read primarily by physicians; ppl typically heard of this research from their physicians - Today, diverse outlets publicize the 'latest and greatest' health research; w/ this increased publicity comes an increasing problem: Some of this info may not be trustworthy - News media are in the business of getting ppl's attention, so news coverage of health info may mislead by focusing on the most sensational findings - Commercial ads may ignore/distort scientific evidence in order to boost sales - Fortunately, there exists a vast body of health-related info that is relatively objective/free from self-serving claims - Info is produced by researchers trained in beh and BM sciences, who are typically associated w/ universities, research hospitals, gov agencies - Because they use methods of science in their work, this evidence accumulates gradually over an extended period of time; dramatic breakthroughs are rare -- 'revolutionary' claims are most often motivated by financial/other personal interests - Important to know the major ways in gathering/evaluating info; much health-related info comes from studies conducted by beh/BM scientists using a variety of research methods - Choice of methods depends in large part on what questions the scientists are trying to answer -- possible to approach any research topic in a variety of ways - Chapter uses studies of obesity to illustrate common research methods - Findings from each research method add to an understanding of obesity and its relationship to a variety of behs/conditions - When researchers want to ID factors that predict or are related to either disease or healthy functioning, they use correlational studies - When wanting to compare ppl across different age groups, they rely on cross-sectional studies - When wanting info on stability/instability of health status or some other characteristic over a period of time, they'll use longitudinal studies - When wanting to compare one group of participants with another, they use either experimental designs or ex post facto designs

Experimental Designs

- Only design to determine causality - Experiment consists of a comparison of at least 2 groups, the experimental group and control group - Participants in experimental group must receive treatment identical to that of participants in control group except those in experimental group receive one level of the independent variable, whereas ppl in control group receive a different lvl - Independent variable is the condition of interest, which experimenter systematically manipulates to observe its influence on a beh/response (the dependent variable) - Manipulation of IV critical element of experimental design b/c this manipulation allows researchers to control the situation by choosing/creating the appropriate lvls - Assign participants randomly to ensure groups are equivalent at the beginning of the study - Often, experimental condition consists of administering a treatment, whereas control condition consists of withholding that treatment/perhaps presenting some sort of placebo - If experimental group later shows a different score on the DV than the control group, the IV has a cause and effect relationship with DV

The Placebo in Treatment and Research

- Placebo effect represents on of the clearest examples of the link b/w ppl's beliefs and their phys health - Most physicians are aware of the placebo effect, and many may even prescribe placebos when no other effective treatments are available - Strong placebo effects can pose a problem for scientists trying to evaluate if a new treatment is effective - P/e may help individuals who receive them but complicate the job of researchers; can have treatment benefits but research drawbacks

The Risk Factor Approach

- Popularized by the Framingham Heart Study, a large scale epidemiology investigation that began in '48 and included 5,000+ ppl in town of Framingham, Massachusetts - Study has allowed researchers to ID such risk factors for cardiovascular disease (CVD) as serum cholesterol, gender, high blood pressure, cigarette smoking, and obesity; these risk factors don't necessarily cause CVD, but they relate to it in some way - Two types of expression exist for conveying risk: 1. Absolute Risk -- person's chances of developing a disease/disorder independent of any risk that other ppl may have for that disease/disorder; chances tend to be small 2. Relative Risk -- ratio of incidence (or prevalence) of a disease in an exposed group to the incidence (or prevalence) of that disease in the unexposed group - Relative risk of the unexposed group is always 1.00; thus a relative risk of 1.50 indicates that the exposed group is 50% more likely to develop condition - High relative risk for lung cancer among ppl who have a long history of cigarette smoking may suggest that most smokers will die of cancer, though this isn't quite the case; most smokers will not die of lung cancer - 39% of male smokers/40% of female smokers who die of cancer will develop cancer in sites other than the lung - The absolute frequency of death due to heart disease makes a smoker almost as likely to die of heart disease as lung cancer - Smokers have a much higher relative risk of dying from lung cancer than CVD, but their absolute RISK of dying from CVD is much more similar

The Role of Theory in Research

- Psych shares the use of scientific methods to investigate natural phenomena w/ other disciplines - Work of science isn't restricted to research methodology; also involves constructing theoretical models to serve as vehicles for making sense of research findings - Health psychs developed a # of models/theories to explain health-related behs and conditions such as stress, pain, smoking, alcohol abuse, and unhealthy eating habits - Theories are practical tools that give both direction and meaning to their research - Scientific theory is :a set of related assumptions that allow scientists to use logical deductive reasoning to formulate testable hypotheses" - Theories/scientific observations have an interactive relationship - A theory gives meaning to observations, and observations in turn fit into and alter the theory to explain these observations - Theories, then, are dynamic and become more powerful as they expand to explain more and more relevant observations - Term 'model' is more appropriate than 'theory' when theoretical framework is still rudimentary/not yet sufficiently comprehensive to explain a large # of observations; in practice however, researchers sometimes use theory and model interchangeably - Role of theory in health psych is basically the same as it is in any other scientific discipline: 1. A useful theory should generate research (both descriptive research and hypothesis testing) - Goal of descriptive research is to expand on existing theory; deals w/ measurement, labeling, and categorization of observations - A useful theory on psychosocial factors in obesity should stimulate the formulation of a number of hypotheses that, when tested, produce a greater understanding of the psych and social conditions that relate to the development and maintenance of obesity; results would either support of fail to support existing theory -- ordinarily don't enlarge/alter it 2. A useful theory should organize/explain the observations derived from research and make them intelligible - Unless research data are organized into some meaningful framework, scientists have no clear direction to follow in their pursuit of further knowledge 3. A useful theory should serve as a guide to action, permitting the practitioner to predict beh and to implement strategies to change beh - A practitioner concerned w/ helping others change health-related behs is greatly aided by a theory of beh change; psychologists w/ particular theoretical orientations rely on their theories to supply them w/ solutions to the many ?s they confront in their practice

Research Tools

- Psychologists frequently rely on two important tools to conduct research: 1. Theoretical models 2. Psychometric instruments - Many (but not all) psych studies use a theoretical model and attempt to test hypotheses suggested by that model - Many studies rely on measuring devices to assess behs, phys functions, attitudes, abilities, personality traits, and other variables

Meta-Analysis

- Research on the same topic may not yield consistent findings, putting researchers in the position of wondering which outcome is valid - Some studies are larger than others, and when it comes to accepting a result, size matters; tho sometimes even large studies furnish results that seem contradictory - M-A allows researchers to evaluate many research studies on the same topic, even if the methods differed - Results from a M-A include a measure of the overall size of the effect of the variable under study - Ability to offer an estimate of the size of an effect is an advantage - If an effect is statistically sig but small, then ppl shouldn't be encouraged to change their behavior on the basis of such findings -- doing so would provide too few benefits - On the other hand, if an effect is large, then working toward change would be beneficial, even if it's difficult - Ex: a group of researchers evaluated studies conducted over a period of nearly 20 yrs to determine relationship b/w childhood weight and risk for developing Type 1 diabetes - Included case-control and prospective designs - Results indicated that childhood obesity doubles the risk of subsequent Type 1 diabetes - This M-A allowed researchers to integrate findings from a large number of studies and draw conclusions abt the impact of weight on this disease

Randomized Controlled Trials

- Second type of epidemiological research, equivalent to experimental research in psych - Researchers randomly assign participants to either a study group or control group, making the two groups equal on all pertinent factors except the variable being studied - Must also control variables other than those of primary interest to prevent them from affecting the outcome - Must avoid the problem of self-selection -- mustn't let participants choose whether to be in the experimental group or the control group (place them randomly) - A clinical trial is a research design that tests the effects of a new drug/med treatment; many are randomized controlled trials that feature random assignment and control other variables, allowing researchers to determine effectiveness of new treatments - Epidemiologists often regard randomized, placebo-controlled, double-blind trials as the 'gold standard' of research designs - Commonly used to measure the effectiveness of new drugs as well as psych and edu interventions - Ex: all drugs approved by FDA must first undergo extensive clinical trials of this nature, demonstrating that the drug is effective and has acceptable lvls of side effect or other risks - When a controlled clinical trial demonstrates the effectiveness of a new drug/intervention, researchers often publish/publicize findings so that others can adopt the treatment, but others that fail to demonstrate effectiveness are 3x more likely to not be published - Researchers/health care providers/public are more likely to learn abt research that shows a particular treatment to be effective, and less likely to learn abt research that shows the same treatment to be ineffective; several safeguards are now in place to help ensure researchers and health care providers can access all available evidence - Ex: major medical journals require that researchers comply w/ clear guidelines in reporting results of a clinical trial (guidelines known as the Consolidated Standards of Reporting Trials (CONSORT) guidelines); requires researchers to register their clinical trial in a registry prior to the start of the study - Database allows anybody to locate all clinical trials conducted on a treatment, not just one that found a treatment to be effective

Establishing Validity

- The extent to which an instrument measures what it was designed to measure - In context of a psych measure such as stress assessment, a valid stress measure should tell you a person is under high stress, but only when that person experiences high stress - Determine validity of a measuring instrument by comparing scores from that instrument w/ some independent/outside criterion; a standard that exists independently of the instrument being validated - In health psych, that criterion can be a phys measure, like a phys stress response such as elevated blood pressure - Criterion can also be some future event, such as a diagnosis of heart disease or the development of diabetes; an instrument capable of predicting who will receive such a diagnosis and who will remain disease free has predictive validity - Ex: scales that measure attitudes abt body predict the development of eating disorders - For such a scale to demonstrate predictive validity, it must be administered to participants who are currently free of disease

Establishing Reliability

- The reliability of a measuring instrument is the extent to which it yields consistent results - Reliability most frequently determined by comparing scores on 2+ administrations of the same instrument (test-retest reliability) or by comparing ratings obtained from 2+ judges observing the same phenomenon (interrater reliability) - Measuring psych phenomena is less precise than measuring phys dimensions such as length - Perfect reliability is nearly impossible, and researchers most frequently describe reliability in terms of either correlation coefficients/percentages - The correlation coefficient, which expresses the degree of correspondence b/w two sets of scores, is the same stat used in correlational studies - High reliability coefficients (such as 0.80/0.90) indicate that participants have obtained nearly the same scores on 2 administrations of a test; percentages can express the degree of agreement b/w the independent ratings of observers

Natural Experiments

- Third area of epidemiological study, in which researcher can only select the IV, not manipulate it - Similar to EPF designs used in psychology and involve the study of natural conditions that provide the possibility for comparison - When 2 similar groups of ppl naturally divide themselves into those exposed to a risk/pathogen and those not exposed, natural experiments are possible; such a situation occurred in 2008 when NYC passed a law requiring restaurants/fast food chains to post calorie content of food on menus; Newark, a large city just 10 miles west of NYC did not pass such a law - Allowed a comparison of food choices in children/teens who saw the calorie content of menu items w/ those who didn't; test hypothesis that the absence of calorie info encourages children/teens to make unhealthy food choices - Results indicated no significant difference in food choices among the young patrons before or after the law took effect - This natural experiment differs from case-control studies in that it began by examining all patrons of the restaurants, regardless of their good choices/obesity - Differs from randomized controlled trials in that researchers selected the participants according to their city rather than manipulating who would receive calorie info and who wouldn't; in a true randomized controlled trial, researchers would assign participants to calorie info conditions

Summary thus far

- Two important tools aid the work of scientists: 1. Useful theories - Generate research - Predict/explain research data - Help practitioner solve a variety of problems 2. Accurate measurement - Accurate psychometric instruments are both reliable and valid - Reliability is extent to which an assessment device measures consistently - Validity is the extent to which an assessment instrument measures what it's supposed to measure

Ex Post Facto Designs

- What researchers rely on when they can't manipulate certain variables in a systematic manner - One of the many types of quasi-experimental studies; resembles an experiment in some ways but differs in others - Don't involve manipulation of IVs; instead researchers choose a variable of interest and select participants who already differ on this variable, called a subject variable/participant variable - Involve the measurement of DVs - Comparison group in an EPF design is not an equivalent control group, b/c these participants were assigned to groups b/c of their weight rather than by random assignment - W/o random assignment, groups may potentially differ on variables other than weight, such as exercise, diet, cholesterol lvls, or smoking; existence of these other differences means researchers can't pinpoint the subject variable as the cause of differences b/w groups - Findings abt differences in lvl of food preference b/w the two groups can yield some useful information, making this type of study a choice for many investigations - EPF designs common in health psych

Correlational Studies

- Yield info abt the degree of relationship b/w two variables; describe this relationship, making them a descriptive research design - To assess the degree of relationship b/w two variables, researcher measures each of these variables in a group of participants then calculates the correlation coefficient b/w these measures; calculation yields a number that varies b/w -1.00 and +1.00 - Positive correlations: variables increase/decrease together - Negative correlations: one variable increases as the other decreases - Correlations that are closer to -/+ 1.00 indicate stronger relationships than do correlations that are closer to 0.00 - Small correlations (<0.10) can be statistically sig if they are based on a very large number of observations, as in a study w/ many participants; though not random, they offer researcher very little ability to predict scores on one variable from knowledge of scores on the other variable - Allowed group of researchers to study the relationship b/w neighborhood factors and overweight, assessing more than 400 residents of housing developments, measuring phys characteristics of neighborhood and relating those characteristics to obesity - Found negative correlation b/w neighborhood features that supported physical activity and obesity

Observational Methods

-Used to estimate occurrence of a specific disease in a given population - Don't show causes of disease, but researchers can draw inferences abt possible factors that relate to the disease - Similar to correlational studies in psych; both show association b/w 2+ conditions, but neither can be used to demonstrate causation - Two important types: 1. Prospective studies; begin w/ a pop of disease-free participants and follow them over a period of time to determine whether a given condition is related to a later condition; equivalent to longitudinal studies 2. Retrospective studies; opposite approach -- begin with a group of ppl already suffering from a particular disease/disorder, then look backward for characteristics/conditions that marked them as being different from ppl who don't have that problem - Retrospective has an advantage over prospective in that it doesn't require as much time or expense - Case-control studies (type of retrospective) -- cases (ppl who had resolved their diabetes) were compared w/ controls (ppl who continued to show symptoms of diabetes); in comparing two groups, researchers found several differences, including amount of weight loss

Criteria for Drawing C/E Relationships b/w Cigarette Smoking and Heart Disease/Lung Cancer By Nonexperimental Means

1. A dose-response relationship must exist b/w a possible cause and changes in the prevalence of incidence of a disease - Dose-response relationship: direct, consistent association b/w an IV such as a beh and a DV such as a disease -- higher the dose, higher the death rate - Demonstrated a D-R relationship b/w # of cigs smoked per day and # of years one has smoked and the subsequent incident of heart disease/lung cancer/stroke 2. The prevalence or incidence of a disease should decline with the removal of the possible cause - Consistently demonstrated that quitting cig smoking lowers one's risk of CVD and decreases risk of lung cancer 3. The cause must precede the disease - Cig smoking almost always precedes incidence of disease 4. A cause and effect relationship b/w the condition and the disease must be plausible - Must be consistent w/ other data, and it must make sense from a bio viewpoint - Not necessary that underlying connection b/w a beh and a disease be known, but it must be a reasonable possibility 5. Research findings must be consistent 6. The strength of the association b/w the condition and the disease must be relatively high 7. The existence of appropriately designed studies - Tho no experimental designs w/ human participants have been reported on the relationship b/w cigarettes and disease, well-designed observational studies can yield the results equivalent to experimental studies - A large number of these observational studies consistently reveal a closer association b/w cig smoking and both CVD/lung cancer

Chapter Questions

1. What are placebos, and how do they affect research/treatment? - Placebo is any treatment that is effective b/c of patients' expectations based on previous experiences w/ treatment - Positive effects are usually beneficial to patients, but create problems for researchers attempting to determine efficacy of a treatment 2. How does psych research contribute to health knowledge? - At least 5 different ways: 2.1. Long tradition of techniques to change beh 2.2. Emphasis on health rather than disease 2.3 Development of reliable and valid measuring instruments 2.4. Construction of useful theoretical models to explain health-related research 2.5. Various research methods used in psych 3. How has epidemiology contributed to health knowledge? - Concepts of risk factor, prevalence, and incidence - Offer research methods similar to those used in psych; at least 2 basic kinds of research methodology: 3.1. Observational studies 3.2. Randomized controlled trials 3.3. Natural experiments 4. How can scientists determine if a behavior causes a disease? - 7 criteria used for determining a C/E relationship b/w a condition and a disease: 4.1. A dose-response relationship must exist b/w condition and disease 4.2. Removal of condition must reduce prevalence or incidence of disease 4.3. Condition must precede disease 4.4. Causal relationship b/w condition and disease must be phys plausible 4.5. Research data must consistently reveal a relationship b/w condition and disease 4.6 Strength of the relationship b/w condition and disease must be relatively high 4.7. Relationship b/w condition and disease must be based on well-designed studies 5. How do theory and measurement contribute to health psych? - Theories are important tools used by scientists to: 5.1. Generate research 5.2. Predict/explain research data 5.3. Help practitioner solve a variety of problems - Use a variety of measurement instruments to assess behs and theoretical concepts - To be useful, these instruments must be both reliable and valid