CH. 2 Doing Health Psychology Flashcards
Common Rubrics for Health
COMMON RUBRICS FOR HEALTH :
Health Psychology is the leading journal – This journal features many studies that define health in terms of:
-
…the extent to which health-improving behaviors are practiced
- (e.g., how much did the participants in the study exercise in a week?)
-
…psychological well-being
- (e.g., what were the participant’s scores on the Profile of Mood States, a common measure of mood?).
Annals of the Society for Behavioral Medicine and Psychosomatic Medicine – measure many specific physiological outcomes. Ex: What are the levels of immune cells in the blood?
- The point is that we determine if people are healthy by measuring a variety of aspects.
- You will see measures of basic physiological levels of bodies’ various systems
- (e.g., blood pressure, heart rate, or cholesterol level).
- You will see measures of how much people practice healthy behaviors
- (e.g., exercising).
- You will also see many measures of psychological well-being
- (e.g., levels of depression or optimism)
- and how well people practice healthy psychological ways
- (e.g., good coping skills).
The major question for all scientific research:Do the findings replicate? different studies use slightly different measures.
Research Primer
RESEARCH PRIMER:
- Health psychology relies firmly on the SCIENTIFIC METHOD.
- Always good practice to go to the original published article to substantiate the results.
- The key elements of science are:
- that it is empirical (relying on sense observations and data) and
- that it is theory-driven.
Key Steps to Doing Research
KEY STEPS TO DOING RESEARCH:
- Identify a question of interest
- Review what has been published on the topic.
- Determine what is left to be discovered or needed to be researched
- Decide how to conduct that research.
Approach example:
-
Researchers can start with descriptive studies
- (e.g., How much are people exercising?),
-
move on to correlational designs
- (e.g., What is exercising associated with?), and then
-
design interventions
- (e.g., If I introduce a new way of talking about exercise, will the amount of exercise change?).
MEASUREMENT – Focuses on describing characteristics of an individual (e.g., self-efficacy – an individual’s belief in his capacity to execute behaviors necessary to produce specific performance outcomes).
ASSESSMENT – Relates to obtaining information according to a goal (e.g., how much did exercise increase after an intervention).
EVALUATION – How did the person perform relative to a normative standard?
Major Research Designs
Major Research Designs:
DESCRIPTIVE STUDIES – The most basic form of research describes what is going on.
- This basic form of design is exploratory and aims to establish baselines for behaviors.
- Ex: How many people smoke? How prevalent is a certain disease or disorder?
PREVALENCE RATES – The proportion of the population that experiences a particular situation or condition.
- EX: what proportion of the population has a particular disease at a particular time (commonly reported as cases per 1,000 or 100,000 people).
INCIDENCE RATES – The frequency of new cases of the disease during a year.
- EX: The Incidence Rates of Covid have fallen from 50 million/year in 2020 to 20 million/year in 2021 and 5 million in 2022 (note: these are made-up numbers)
Correlational Studies
CORRELATION STUDIES – The most basic form of research design – describes relationships between variables.
- But does NOT indicate CAUSATION – determining what causes the studied variables to have a relationship.
CORRELATION COEFFICIENT – This Is the statistical measure of the association represented with a lowercase r.
- Correlations range from –1.00 to +1.00
-
Values closer to 1 (regardless of sign) signify stronger associations.
-
Positive correlations indicate variables that change in the same direction
- (e.g., higher weight correlates with a higher risk of cardiovascular disease).
-
Negative correlations indicate variables that change in opposite directions
- (e.g., lower socioeconomic status correlates with higher smoking rates).
-
Positive correlations indicate variables that change in the same direction
- Given that there are many factors accounting for any behavior or result, a statistically significant correlation in the .2 to .3 range between any two variables suggests a worthy relationship.
- Statistical analyses with p-values (probability values) less than .05, .01, or .001 (reported as p < .05, p < .01, or p < .001) are significant.
- All results tables will indicate statistically significant differences with an asterisk (*) so you can be on the lookout
- * signifies p < .05
- ** signifies p < .01
- *** signifies p < .001
- The lower the p-value, the better.
- All results tables will indicate statistically significant differences with an asterisk (*) so you can be on the lookout
- A probability value of less than 0.05 suggests that the probability of getting the same result by chance is less than 5 in 100 – You can see why p < 0.001 is a significant level.
- The p-value is influenced by sample size.
- Given that many studies in health psychology have very large samples, even small correlations could be statistically significant.
- The p-value is influenced by sample size.
DIRECT CORRELATION – When only the relationship between two variables is tested.
PARTIAL CORRELATION – There is often more than one variable influencing another; to statistically control for multiple associations, researchers use a partial correlation.
- When you calculate a partial correlation or control for another variable, the relationship between two variables is tested while controlling for a third variable (or more).
- Ex: researchers often statistically control for a research participant’s age when assessing correlations, which essentially acknowledges that the association between the variables of interest (e.g., distress and coping style) could vary for people of different ages.
CORRELATIONAL TABLES – The main variables are listed in the column down the left (e.g., 1. Age, 2. Nativity). The same variables are referred to along the top, and, conveniently, the labels (e.g., Age) do not have to be rewritten.
- Each dash (–) represents the association of a variable with itself. Given that one variable will be perfectly (i.e., 1) correlated with itself, the numeral 1 is not used.
- All the numbers with asterisks represent the statistically significant correlations. The more *s, the higher the statistical significance.
Experimental and Quasi-Experimental Designs
Experimental and Quasi-Experimental Designs:
EXPERIMENTAL DESIGN – helps us determine causality.
- If we really want to know if something caused something else we need to introduce that something and see if it has an effect.
INDEPENDENT VARIABLE – (the manipulated variable) The researcher manipulates the variable that is believed to be important. (The independent variable – believed to be CAUSAL to the Dependent Variable).
DEPENDENT VARIABLE – (the affected variable) and measures how changes in the independent variable influence another variable - the DEPENDENT VARIABLE.
- Experiments have two or more groups, each of which experiences different levels of the independent variable.
RANDOMLY SAMPLED – everyone in the population has an equal chance of being in the study) and
EXTRANEOUS VARIABLES – other variables (beyond the Independent Variable) that may influence the outcome of interest – such as socioeconomic status or other health behaviors.
- These Extraneous Variables must be controlled for. Only then can one be fairly certain that changes in the dependent variable are due to changes in the independent variable.
QUASI-EXPERIMENTAL DESIGN – modifying true EXPERIMENTAL DESIGN to fit the constraints of reality and society.
- In health psychological research, it is often impractical and unethical to manipulate key variables of interest (e.g., making people smoke) so groups that naturally vary in the variable of interest are used instead (e.g., compare groups of people who vary in how much they smoke)
- Because using naturally occurring groups is not a perfect experiment, such designs are referred to as QUASI-EXPERIMENTAL DESIGNS, and the independent variables are called SUBJECT VARIABLES.
Randomized Control Trials
RANDOMIZED, CONTROLLED, or CLINICAL TRIALS (RCTs) – In which one group gets an experimental drug or intervention treatment and a second group unknowingly gets a placebo or nothing.
INTERNAL VALIDITY – ensuring that the active intervention, and not other factors, caused the observed changes in the outcome.
EXTERNAL VALIDITY – How well it will generalize to a large sample.
Cross-Sectional and Longitudinal Designs
Cross-Sectional and Longitudinal Designs:
CROSS-SECTIONAL – Conducted at one point in time.
- Cross-sectional studies often sample a large number of people and examine different cultural groups in the sample comparing men and women, and people of different ethnicities.
LONGITUDINAL – Conducted over a period of time and often involves many measures of the key variables – usually follows the same people or situation over a long period.
-
Research can be:
- PROSPECTIVE – Following disease-free participants over a period to determine whether certain variables (e.g., eating too much fast food) predict disease.
- RETROSPECTIVE – Studying participants with a disease and tracing their histories of health behaviors to determine what caused the disease.
Health psychology measurement involves five major steps
Health psychology measurement involves five major steps:
- __General framework, considering purpose.
- Target population.
- Type of measurement.
- Psychometric characteristics.
- Issues of implementation.
- When we talk about measurement we need to keep in mind the main qualities of measurement.
RELIABILITY – Ex: its internal consistency, inter-rater reliability, test-retest reliability.
SENSITIVITY – The extent to which the instrument may detect small changes occurring over time.
- Is the instrument or measurement sensitive enough to detect the changes you are looking for?
Statistically Savvy
Statistically Savvy:
CORRELATIONAL COEFFICIENTS – Reported using the italicized letter r.
TESTS OF GROUP DIFFERENCES – Reported using the italicized letters F or t.
-
Be aware: Not all associations or changes may be statistically significant. Furthermore, not all statistically significant change may be meaningful change.
- i.e. something may be significant, but it just doesn’t matter with regard to what you’re looking at.
-
Increasing the sample size can make previously insignificant changes significant.
- Only phenomena that have a large EFFECT SIZE will be significant when the sample size increases
Common Statistical Tests
COMMON STATISTICAL TESTS:
- Analyses of variance (ANOVAs)
- Multivariate analyses of variance (MANOVAs)
- Regression analyses
ANOVAs and MANOVAs – Test for differences between group means.
- If you want to test for differences between a number of variables that are related to each other, you would use a MANOVA.
REGRESSION ANALYSES – Used to predict the likelihood of an outcome from a list of variables.
- In regressions, you can actually get a sense of how much of the variance in the dependent variable your predictor variables account for.
For an example, refer to Table 2.3 in the book:
- Regression analysis and their results table is reproduced in Table 2.3.
- Pay the most attention to the p values.
- The variable that significantly predicted cortisol was the model with the condition in it showing a direct effect of time of day on levels of cortisol.
- Neither sex nor intentions are significant variables.
Logistic Regression
LOGISTIC REGRESSION – This analysis predicts the probability of the occurrence of an event.
ODDS RATIO – The ratio of the odds of an event occurring in one group to the odds of it occurring in another group.
- In other words: The odds of the probability of an event occurring in one group, divided by the probability of it not occurring.
-
Ex: Are men more likely to have a heart attack than women? (See Chapter 14 for the answer.)
- An odds ratio of 1 suggests the phenomenon (e.g., a heart attack) is equally likely in both groups.
- An odds ratio greater than 1 suggests the phenomenon is more likely to occur in the first group.
HAZARD RATIO – Is the comparison between the probability of events taking place in a treatment group compared to the probability of the events taking place in a control group.
- The hazard ratio essentially provides a statistical test of the efficacy (effectiveness) of a treatment
- The statistic that is now overtaking the journals in health psychology is the hazard ratio.
- Both the odds ratio and the hazard ratio relate to RELATIVE RISK. The probability of seeing a certain event in some group is called RISK.
RELATIVE RISK – Ratio of incidence or prevalence of a disease in an exposed group to the incidence or prevalence of the disease in an unexposed group.
ABSOLUTE RISK – Person’s chance of developing a disease independent of any risk that other people may have.
Structural Equation Modeling
STRUCTURAL EQUATION MODELING – Model multiple relationships simultaneously. You can draw a structure of variables and hypothesize how they are related.
Context and Level of Analysis
CONTEXT AND LEVEL OF ANALYSIS:
PERSISTENT – Wise, smart – INSIGHTFUL – Loving – SUCCESSFUL – Leaders – VISIONARY – Creative – CLUTCH – Powerful – STRONG – RELENTLESS – Enduring, rock
- The order in which the different descriptors came to your mind gives you a good idea of the aspects of yourself that are most important to you right now.
CONTEXT – First, the order in which we use words to describe ourselves often depends on the context or the environment in which we are.
- Ex: If asked to define yourself at a Yankees game, in a bar, at a family reunion, on Christmas morning, at work, etc, your answers will vary influenced by the context.
LEVEL OF ANALYSIS – This means that our views of ourselves reside at different levels of conscious awareness.
ANCOVAs and MANCOVA
ANCOVAs and MANCOVAs – Where the “C” stands for covariance.
- Beyond just controlling for variables(ANOVA, MANOVA), health psychological research aims to test for the different ways that third variables can influence relationships between two other variables (COVARIANCE).
MODERATOR – a variable that changes the magnitude (and sometimes the direction) of the relationship between an antecedent variable and an outcome variable.
-
Ex: Older people tend to be more health conscious than younger people. People high in social support tend to cope better than people low in social support. In each of these cases the variable—income, age, and social support—are called moderators.
- In the example of social support, the number of stressors can be the antecedent variable and well-being is the outcome.
- A simple direct effect would be that people with more stressors are unhappier (a positive correlation). However, things are more complex than that. In any group of people, some individuals will have more social support than others.
- In the example of social support, the number of stressors can be the antecedent variable and well-being is the outcome.
- Let’s measure social support and divide the people into a high support and a low support group. We would find that people with more support are happier than people with less social support. Social support has moderated or buffered the relationship between stress and well-being.
- Such moderating effects of support are now well established and can be seen in a variety of life examples.
MEDIATOR – Is the intervening process (variable) through which an antecedent variable influences an outcome variable.
- Mediation can be described as a relationship where an independent variable changes a mediating variable, which then changes a dependent variable.
-
Ex: Coping behaviors in general and specific health behaviors are common mediators. Look at Figure 2.5. Instead of stress directly making you feel good or bad, it may influence your health behaviors (e.g., you drink alcohol or eat more) that in turn influence whether you feel good or bad.
- Here, health behaviors have mediated the relationship.
-
Ex: Coping behaviors in general and specific health behaviors are common mediators. Look at Figure 2.5. Instead of stress directly making you feel good or bad, it may influence your health behaviors (e.g., you drink alcohol or eat more) that in turn influence whether you feel good or bad.
- A large body of literature in health psychology concerns interventions aimed at improving well-being by enhancing coping, based on the assumption that effective coping is a mediator.
- It is easy to see whether mediation is taking place by comparing the correlation between the antecedent and outcome variables before and after the potential mediator is entered into the statistical analysis.
- If the variable you are studying is a mediator, the relationship between the antecedent and outcome variable significantly changes (gets lower) once the mediator is in the analysis.
- Ex: If you are stressed and you take a nap, you will probably wake up feeling better. If you are stressed and you do not take a nap, you may feel worse. In this example, sleep is said to mediate the relationship between your stress level and how you feel.
-
Most variables that health psychologists study (e.g., coping styles and social support) can be both mediators and moderators.
- The role of the variable depends on the study.
- As a rule, mediators are changed by the stressor and correspondingly change the outcome.
- Ex: If more stress leads to you asking for more social support, which leads to you feeling better, social support is a mediator.
-
Ex: If a longitudinal study shows that those with more stress exercise more and this makes them feel better, then exercise is a mediator.
- HOWEVER, If a correlational study shows that the group of people who exercise more are less distressed than a similar group of stressed individuals who exercise less, then exercise is a moderator.
- As a rule, mediators are changed by the stressor and correspondingly change the outcome.
- The role of the variable depends on the study.
- In the first case (mediation), exercise follows the stressor, changing in level (e.g., increasing) and influencing the outcome.
- In the second case (moderation), we are looking at two separate groups of exercisers.
The only variables that cannot be both mediators and moderators are those that cannot change as a function of the stressor or antecedent variable.
- Ex: Age, ethnicity, and race are examples of moderators that cannot be mediators (e.g., being more stressed cannot change your age)