Ch4 :Research methods Flashcards
science
“to know”
systematic pursuit of knowledge through observation
what does science involve?
forming a theory then systematically gathering data to test the theory
theory
set of propositions developed to explain what is observed
is falsifiable
hypothesis
specific predictions about what will occur if theory is correct
three (main) methods of research in psychopathology
- case study
- correlation
- experiment
Case study
collection of detailed biographical information- about one person at a time
(developmental milestones, family history, medical history, environment, etc)
what are case studies useful for (4)
- excellent source of hypotheses
- can provide info about novel cases or procedures, , especially for rare disorders
- rich description
- can disconfirm a relationship believed to be universal (disprove hypothesis)
what is a case study not useful for (3)
- lack control and objectivity (paradigm may not be useful)
- can’ t provide causal evidence b/c cannot rule out alternative hypotheses
- paradigm can influence observations
correlational method
- do variable x and variable y vary together
- are they related in a systematic way?
- variables measured but not manipulated- studied as they exist in nature
does correlation = causation?
no
in a correlational study, all participants are measured the same way, nothing is manipulated
in causal experiments, participants are randomly assigned to different levels that are manipulated, then reactions are measured.
correlation coefficient
“r” -1.0 to 1.0
measures the strength of relationship- both magnitude and direction
strength of r
higher the absolute value, stronger the relationship (-1.00 and 1.00 are strongest)
direction of r
positive or negative
positive r = higher scores on variable x associated w/ higher scores on variable y
negative r = higher scores on variable x associated w/ lower scores on variable y
statistical significance
a statistically significant correlation is large enough that it is unlikely to have occurred by chance
measured by alpha level (p <0.5)
means there is a 5 percent or less chance that the finding obtained by chance alone
in general, as absolute size of correlation coefficient increases, more likely to be statistically significant
what is statistical significance influenced by
- size of the relationship between variables
- number of participants in a study
if fewer people are studied, the correlation coefficient needs to be larger to reach stat sig.
clinical significance
whether a relationship between variables is large enough to matter–
is the association meaningful as well as statistically significant
whether the effect is large enough to be meaningful in predicting/ treating a psychological disorder
Neumonic: doctor looking at giant bump on head saying this is clinically significant
directionality problem
Correlation doesn’t imply causation
- can’t say which (if either) variable causes the other
longitudinal design
researcher tests whether causes are present before disorder has developed
helps w/ directionality problem b/c shows cause precedes effect
(in contrast to cross-sectional design)
Neumonic: line poking through someone with a disease
cross-sectional design
researcher measures causes and effects at same point in time
(in contrast to longitudinal design)
Neumonic: a cross section of somebody with a disease, holding clock
high-risk method
method in longitudinal design to only study people who are high-risk for something, to look at variables that precede getting said thing
third-variable problem
Correlation doesn’t equal causation-
Correlational research can’t rule out third variables (confounds) that may be responsible for correlation
epidemiology
study of the distribution of disorders in a population and possible correlates
designed to be representative of population being studied
Neumonic: demons giving (distributing) epipens to people in a population
three features of a disorder (epidemiology)
- prevalence
- incidence
- risk factors (correlates)
RIP
prevalence (features of a disorder)
the proportion of people w. disorder either currently or during their lifetime
incidence (features of a disorder)
proportion of people who develop new cases of disorder in some period (usually a year)
risk factors/correlates (features of a disorder)
Variables that are correlated w. the presence of the disorder
what type of study are epidemiological studies?
correlational studies- they examine how variables relate to each other w/out manipulating variables
The National Comorbidity Survey-Replication
- Large-scale national survey
- used structured interviews to collect info on prevalence of several diagnoses (in epidemiology)
Neumonic: NCIS doing a survey of the epipen demons
behavior genetics
estimates genetic predisposition (concordance) for a disorder by looking at relatives disorders
difference btwn behavior genetics and molecular genetics
behavior genetics- looks at similarities in relatives psychopathology
molecular genetics- identifying specific genes that contribute to presence of disorder
similarity btwn behavioral genetics and molecular genetics
both rely on correlational techniques in the study of psychopathology
three common methods used in behavior genetics
-family method
-twin method
-adoptees method
family method (in behavioral genetics)
- used to study genetic predisposition among family members
- first-degree relatives (50% genes shared) compared to second-degree relatives (25% shared genes)
concordance
when relatives match on presence or absence of disorder- said to be concordant
index cases/ probands
collection of sample of persons w/ diagnosis in question
(in behavioral genetics studies)
Neumonic: index cards with a band on it thumbs up we all have schizophrenia
twin method (in behavioral genetics)
monozygotic (MZ) or identical twins (100% shared genes)
dizygotic (DZ) or fraternal twins (50% shared genes)
adoptees method/ adoption studies vs cross-fostering (in behavioral genetics)
adoptees method- study of adoptees who have BIOLOGICAL parents with psychopathology
cross fostering- study of adoptees who have ADOPTIVE parents w/ psychopathology
Neumonic : kid going to a sad foster family
two types of molecular genetic studies
- association studies
- genome-wide association studies (GWAS)
association studies (in molecular genetics)
examine the relationship btwn specific allele and a trait or behavior in the population
genome-wide association studies (GWAS)
(in molecular genetics)
examines the entire genome of a large group of people to identify variations btwn people
the experiment
- provides info about causal relationships
- can evaluate treatment effectiveness
what do experiments involve
involves
- random assignment
- independent variable (manipulated)
- dependent variable (measured)
experimental effect
differences btwn conditions on the dependent variable
internal validity
refers to extent which experimental effect can be attributed to the independent variable
need a control group to have internal validity
also need random assignment
control group
level of the independent variable that doesn’t receive the experimental treatment
is needed to claim that the effects of an experiment are due to the independent variable (internal validity)
random assignment
Helps ensure that groups are similar on variables other than the independent variable
done by randomly assigning participants to different levels of the independent variable
external validity
extent to which results generalize beyond study
not needed (as much) if internal validity is prioritized
analogue experiment
- experiments not always possible in psychopathology (due to ethical or practical constraints
- analogue experiments are when we examine related or similar behavior (usually less severe) in lab
- induce temporary symptoms
- recruit participants w similarities to diagnosable disorders
- animal research
single-case experimental research
examines how individual participants respond to changes in ind. variable
single-case experimental research vs case studies
single-case experimental research can have high internal validity
reversal (ABAB) design
type of single-case experimental research
baseline (A)
treatment introduced (B)
reinstate baseline (A)
reintroduction (B)
reversal technique not always possible
biggest drawback of single-case design
lack of external validity- if it works for one person, doesn’t mean it’ll work for others
can combat this by using it to test if research w larger groups is warranted, or replicating in a series of single-case experiments on participants who differ-
- if so can provide strong test of hypothesis
treatment outcome research
research designed to answer the q: Does treatment work?
treatment study should include:
- clear definition of sample being studied (diagnosis description)
- clear description of treatment (as in treatment manual)
- experimental design involving random assignment of clients to treatment or comparison conditions
- reliable and valid outcome measures
- large enough sample size
Randomized controlled trials (RCTs)
studies in which clients are randomly assigned to receive active treatment or comparison
- ind variable is treatment, dep. variable is clients outcome
Empirically supported treatments
treatments that follow certain standards, have been studied and considered to be supported by evidence
empirically supported treatments include
- treatment manuals
- research that used:
- control groups
- placebo
- double-blind procedure
- sample composition
- efficacy and effectiveness
- need for dissemination
- control groups
treatment manuals
books that provide details on how to conduct a particular psychological treatment
includes specific procedures for therapist to follow at each stage of treatment
placebo
in a study- someone thinks they’re getting the treatment but they’re not
double-blind
the patient and the researcher don’t know who is getting the real treatment
sample composition (in empirically supported treatments)
recruit people who mirror the characteristics of real-world people seeking help
specific and well-defined populations
exclusion of diverse populations
efficacy (in RCTS)
whether a treatment works under the purest conditions
effectiveness (In RCTs)
how well the treatment works in the real world
dissemination
process of facilitating adoption of efficacious treatments in the community
typically by developing guidelines about best available treatments along with training for clinicians
cultural competence
a therapist that has an appreciation for and understanding of cultural differences and similarities, appreciates sociopolitical events that may be salient for some cultural groups, and has awareness of their own culturally based assumptions and biases
replication
repeating a research study to see if you get the same results- findings considered “reproducible”
publication bias
tendency of journals to only publish positive results,
may keep us from seeing replication failures
reasons studies fail to replicate
- small samples (more likely to yield unusual patterns of results)
- unreliable measures used
- methods of orginal study not described well
- differences in methods
p-hacking
tweaking analyses or combing through data until arriving at a significant result
ways to address issues in replication
- registering hypotheses, measures, analysis (before study)
- making data publicly available (after study)
- paying more attention to quality of study than findings (reviewers in journals)
- large sample sizes
- consider whether findings have been replicated
meta-analysis
way to look at findings from multiple related studies together
first identify relevant studies, then compute effect size
effect size
transforms results from different studies to one scale
steps of meta-analysis
-identify relevant studies
- calculate effect size of each study
- evidence of publication bias considered
- average effect size calculated
smith et al. (1980)
first meta-analysis
meta-analyzed 475 outcome studies (25,000 subjects)
results: psychotherapy is effective
critisms: gave equal weight to both poor and well-controlled studies
meta-analysis criticisms and how to combat them
researchers sometimes include studies of poor quality
a good meta-analysis will be clear about criteria for including or excluding studies
