Search
Create
PSYC 3111: Research and Methods Midterm
STUDY
Flashcards
Learn
Write
Spell
Test
PLAY
Match
Gravity
Terms in this set (80)
psychology
a science that seeks to describe, explain, and predict behavior
empiricism
basing one's conclusions on systematic observations
producers (of research)
conduct empirical research, understanding research methods is key
ex. your professors, your TAs, research assistants, research scientists in the government and industry
consumers (of research)
use research when applied to work, hobbies, relationships, etc.
ex. ALL OF US
4 scientific cycles
1. the theory-data cycle
2. the basic-applied research cycle
3. the peer-review cycle
4. the journal-to-journalism cycle
theory
a convenient summarizing generalization that is helpful in guiding decisions about future observations, but which is discarded when the obtained observations do not fit the conceptual framework
hypothesis
a basic research question that constitutes a prediction about the outcome of a scientific investigation
a good theory is...
- falsifiable
- parsimonious
- supported by data
operationalization
ex. measuring drunkenness with BAL and mood with the POMs
evidence for beliefs
- experience: it worked for me
- intuition: I have a sense it'll work if I try it
- authority: my mom said it was true
3 goals of clinical research
1. understand how psychopathology develops
2. understand the underlying mechanisms (so that you can design investigations to target those mechanisms)
3. understand the factors that predict the effects of the treatments
3 steps to writing a literature review
1. select a research topic
2. collect and read the relevant articles/chapters
3. write the review
basic-applied
- basic: attempt to increase our body of knowledge
- applied: attempt to solve a practical problem
basic research example
Does synthetic THC and/or CBD slow the proliferation of cancer cells in vitro?
applied research example
Is a medication that combines THC and CBD more effective than standard treatment for brain cancer in terms of survival rate?
clinical research
applying research to specific issues, looking for specific answer
4 functions of research
1. discovery
2. demonstration
3. refutation
4. replication
discovery
research that seeks to uncover NEW relationships between constructs, and is actually used to generate hypotheses as opposed to testing them (also called inductive research)
demonstration
research where the purpose is to demonstrate or support a hypothesis (also called deductive research)
refutation
research which attempts to disconfirm or refute a competing hypothesis
replication
when other researchers in other settings with different samples attempt to reproduce research as closely as possible (failure to replicate suggests the theory is wrong)
Karl Popper (1902-1994)
- theories are the result of our innate tendency to want to impose order on the world
- as such, they should be conceptualized as creations of the mind, rather than "truth" about "reality"
- what keeps science rational, and where the value of research comes in, is the face that scientific theories can be shown to be FALSE
Popperian View
- the falsification of a theory is how science progresses
- argued that science and scientific method of empirical research leads to a linear, cumulative growth of knowledge
Thomas Kuhn (1922-1996)
- wrote "The Structure of Scientific Revolutions" (1962)
- disagreed with Popper, growth is not linear at all but occurs in radical breaks called "paradigm shifts" that occur as a result of "crisis"
Kuhnian View
- since theories aren't truth, and we can't really ever know "truth" then what we are doing with the scientific method is essentially testing theories against other theories
- science proceeds in the "dominant paradigm" during periods of scientific normalcy
- when evidence proves a theory wrong, the theory undergoes a "crisis" and paradigm shift occurs
types of claims made in reserch
1. frequency
2. association
3. causal
frequency claims
reports number or proportion of people who experience a particular outcome
ex. 2/3 of Americans are completely sedentary
ex. less than half of sexually active adolescents use condoms consistently
association claims
suggest that two variables are related; typically the results of passive observational studies; often used in more focused research
ex. exercise is associated with better cognitive function in older adults
ex. older people are more likely to stereotype
verbs: is linked with, goes with, predicts, prefers
causal claims
suggests that one variable causes another; must be based on strong experimental research design; be VERY skeptical
ex. watching Baby Einstein causes higher IQ in children
ex. using marijuana shrinks your hippocampus
verbs: causes, affects, increases, leads to, reduces
types of statistics
- inferential
- descriptive
- correlation
- effect size
inferential statistics
allows you to generalize beyond your sample; population vs. sample
descriptive statistics
measures of central tendency; standard deviation and variance, r (the correlation coefficient)
correlation
used with two or more variables to explore a relationship/association; describes the strength of the relationship between two variables at a time
effect size
tells us how strong a relationship is; the magnitude of difference between groups
regression
- used with 1 (correlation) or more (multiple regression) independent variables and one dependent variables
- very flexible analytic technique
- can use multiple predictor variables
- can test interaction hypotheses
- control for effects of other variables (covariates/confounds)
chi-square
test of independence; compares the observed frequencies that occur in each of the categories
paired samples t-test
repeated measures; one set of participants each person providing both sets of scores on two different occasions and under two different conditions
ANOVA (analysis of variance)
compares the variability in scores
two way ANOVA
two or more categorical IVs; allows us to test whether levels of one IV are influenced by levels of another IV
internal validity
the extent to which a valid CAUSAL statement can be made about the effects of an IV on the DV
covariance
as one changes, the other changes
temporal precedence
which variable came first?
to increase internal validity...
- randomly assign participants to levels of the IV
- virtually eliminate the possibility that extraneous variables influenced your DV
external validity
the degree to which results of a specific study can be generalized to other people, places, times, or empirical realizations of the IV (also known as generalizability)
to increase external validity with regard to PEOPLE...
- be explicit about the population
- randomly sample from that population
generalizability of operations
to what extend do the operations and measures in the experimental procedure reflect the theoretical concepts that gave rise to the study in the first place
study validity
- internal: can you safely say that one variable caused a change in another?
- external: do the results generalize?
- statistical conclusion: were the stats done in a way that gives you the correct answer?
measurement validities
- face validity
- content validity
- concurrent validity
- predictive validity
- convergent validity
- discriminant validity
face validity
the extent to which an instrument appears to measure what it purports to measure
content validity
the extent to which a measure captures all parts of the concept
concurrent validity
the extent to which a measure is related to a concrete, simultaneous outcome that it should be related to
predictive validity
the extent to which a measure is related to a concrete, future outcome that it should be related to
convergent validity
the overlap between alternative measures that are intended to tap the same construct, but have different sources of irrelevant information (meant to tap into the same construct but ideally they have very different biases)
discriminant validity
a measure should fail to correlate with measures it's not supposed to correlate with; the correlation with measures it is supposed to correlate with should not be completely overlapping
projective tests
- presentation of ambiguous stimuli
- projection of personality and the unconscious
- psychoanalytic roots
ex. Rorschach Inkblot Test
objective assessments
- direct answers to specific questions
- normed and standardized
- given to large numbers of people from all backgrounds
- can compare answers to similar individuals in standardization
- clinician looks for patterns of responses and compares to norms of similar subjects
ex. Minnesota Multi phasic Personality Inventory (MMPI)
4 types of scales
1. nominal
2. ordinal
3. interval/ratio
nominal scale
categories or values where no ordering of the values is implied (gender, race/ethnicity, true/false, yes/no)
ordinal scale
categories that can be placed in rank order, but the distance between the values is not known (year in school: F/SO/J/S)
interval/ratio scales
scales in which the spacing between the values is known, and the distance between each successive value is the same; no true zero point
reliability
the proportion of total variability in scores that is due to true score variability
test-retest reliability
the correlation between scores on the same measure administered on two different occasions
internal consistency reliability
the extent to which items on a scale "hang together" or measure the same construct (Cronbach's alpha - a measure that approximates the average correlation between all possible pairs of items on the scale, ranges from 0 to 1, where 0.0=perfectly unreliable and 1.0=perfectly reliable)
inter-rater reliability
two or more observers come up with the same, or very similar, findings (ideally match on individual level, but must match ordinally)
non-probability sampling
lower external validity; there is no way to know the probability of sampling an individual person; there is no certainty that each person has a known chance of being selected
- pros: you can recruit participants quickly and easily, you can recruit from difficult to reach or "hidden" populations
- cons: lower external validity, homogeneous samples (white, high SES college students), selection biases
sampling
- we all begin with some defined POPULATION (e.g., all college women between 18-25, all children from single parent homes)
random sampling
the BEST method of increasing external validity; not ALL sampling is random
convenience samples
- type of non-probability sampling
- most commonly used technique in psychology
- ex. college students in intro to psych sign up for experiments
- ex. post a link to your survey on your Facebook page
- solicit participation from Mechanical turk users (MTurkers)
quota samples
- type of non-probability sampling
- simply a special case of a convenience sample of particular subpopulations
- have all the biases of convenience samples
- ex. 25% of the population is Hispanic, so you take a convenience sample where 25% are Hispanic
purposive sampling
- I want to study alcoholics, so I recruit from: a bar or AA meeting
- and hope that my sample snowballs (not random)
probability sampling
higher external validity, you specify the probability that each element in the population will be selected for your sample (every person...)
- pros: ability to generalize the population; maximizes high external validity
- cons: time, money, and effort intensive; rely on TOTAL participation of the sampled units, or the sample is biased
passive observational design
- what if we can't manipulate an IV? if we are interested in looking at relationships that exist in the real world we seek to make an association claim
- any type of analysis is possible with any design (it depends on the type of measurement rather than design); BUT correlation, regression and structural equation modeling are commonly associated with associational claims
- this is the only viable design for cannabis research
prospective observational design
- phase I to IV
- dependent t-test
multivariate designs
- more than one IV
- pros: can rule our potential 3rd variables that might explain association
- cons: without an experimental design, still cannot make causal claim
multivariate passive observational designs and the 4 validities
- construct validity: how well were the variables measures? how high was the face, content, and convergent validity?
- statistical validity: how strong are the regression coefficients? are they significant?
- external validity: would the results generalize?
- internal validity: no temporal precedence
longitudinal designs (pros and cons)
pros:
- establishes temporal precedence (e.g., you can see changes in vegetable consumption in your 20s can affect your chances of being diagnosed with cancer in your 60s)
- allows for measurement of variables in the sample at multiple points in time (if same variables, allows for assessment of change) (if different variables, allows for prediction)
- allows you to measure variables that would be unethical or impractical to manipulate (e.g., influence of being in armed conflict on stress over time)
cons:
- time (measures may become outdated over time)
- expense
- ultimately still non-experimental
longitudinal designs
- autocorrelations: how a variable correlates with itself over time (how one variable at one point in time correlates with itself at a different point in time)
- cross-sectional: between two variables measured at the same time
longitudinal designs and the 4 validities
- construct: how well have you measured each variable? were your measures really measuring your construct of interest?
- statistical: what is the significance of the relationship? what is the effect size? (usually assess with multiple regression or structural equation modeling)
- external: by design, there is high external validity for longitudinal designs
- internal: directionality (yes), manipulated variable/random assignment (no and no), still cannot make causal claim
moderator
- a third variable that determines subgroups for whom an association is more or less true
- ex. sex, race, socioeconomic status, geographical location, stress, motivation
mediator
- a third variable that explains how or why there is an association between two variables
- ex. why does sex lead to happiness? IV (sex) - mediator (endorphin release) - DV (happiness)
;