Stats Flashcards
what is a theory
an organized set of beliefs about a phenomenon
what is internal validity
list the three criteria required:
the extent to which the association between x and y is causal in nature
- statistical association between variables
- temporal precedence
- nonspuriousness
what is statistical conclusion validity
examines whether there is a statistical association between x and y, and what the magnitude of the relationship is.
some issues can include statistical significance, effect size, and statistical power
what is construct validity
the extent to which conclusions can be made from a test regarding the higher order construct.
e.g. did you capture all aspects of the construct?
what does random assignment to groups do in the context of a group-based randomized experiment?
it precludes systematic pre-test group differences because the groups are probabilistically equated on all measured and unmeasured constructs
what is an efficacy trial
conducted in ideal circumstances to see fi an intervention has an effet
what is an effectiveness trial
conducted in real-life setting to see if an intervention can actually lead to changes as it might be implemented in the community
what is the purpose of intent-to-treat analyses?
to provide a conservative (and real-world) estimate of the treatment effect, which includes cases exposed to varying levels of treatment
what is an ABAB design?
a type of single-case experiment that examines:
baseline (A)
intervention (B)
baseline (A
intervention (B)
what is a multiple-baseline design
capture baseline data of several behaviors, then monitors how these behaviors change after implementation of some intervention
what are things you look at when evaluating a single-case experiment?
mean changes
level shifts
slope (or function) changes
what is a quasi-experimental study and what are two common examples of these?
when you can’t do random assignment
- when patients self-select to an intervention
- when intact groups (e.g. patients at different community health centers) are differentially exposed to treatment conditions
researcher has to figure out how to rule out these threats to internal validity
what is an interrupted time series design?
you collect data consistently before and after an intervention
what is a regression discontinuity design?
you compare groups who are just above vs. those just below a cutoff score
e.g., you look at people who score 79 and dont get a scholarship with those who score 80 and do
what is a case-control design?
comparing a group of participants with a certain characteristic (e.g., those with ADHD) to a group without that characteristic
what is a cohort design?
you follow an intact cohort over time to examine emergence or changes in the variable of interest.
this is a LONGITUDINAL design.
you can use multiple cohorts of different ages to get a cross-sequential design (AKA accelerated longitudinal design) (and shorten the time you have to gather data)
how are multiple cohort designs different from case-control designs?
multiple cohort: groups differ on exposure to something (e.g., poverty)
case control: groups differ on central characteristic (e.g., ADHD)
how to evaluate the potential impact of internal validity threats to your study
in single group studies: threat to validity has to be capable of producing the pattern of results observed
in multiple-group studies: threat needs to be capable and vary systematically by group
what is external validity?
how well the results can generalize, or under what conditions the intervention can work.
threats include:
what are interactions/limits of the intervention? (e.g., does this work for different genders, SES, race/ethnicity)
list the three basic principles of the Belmont report
beneficence
respect for individual’s autonomy
justice (equitable distribution of potential risks/benefits)
what is a nomological network
the relations among observed measures, relations between observed measure and latent constructs, and relations among latent constructs
what is the reliability of a measure from the perspective of CTT
ratio of true score variance to total variance
what is tau equivalence
in the population, all the items in our test have the same relationship with the underlying construct,
5 domains to assess construct validity
1) test content, 2) response processes, 3) internal structure, 4) relations with other variables, 5) consequences of use
what is structural validity
the extent the measure’s structure aligns with the theorized factor structure
Rasch model
psychometric model for analyzing categorical data - a function of the responder’s abilities and the item difficulty
3 decisions to make in EFA
1) method of factor extraction, 2) method of factor rotation, 3) how many factors to retain
orthogonal vs oblique rotation
oblique: factors assumed to be correlated
how do you determine how many factors to retain in EFA?
examine scree plot, eigenvalues >1
what are some fit indices in CFA?
chi-square test, root mean square error of approxiamtion (RMSEA), standardized root mean square residual (SRMR). ALL quantify how well the model-implied variance reproduces the estimated population covariance matrix
four types of data sources
LOTS: Life events, observational, testing, self-report
what is measurement invariance
ensuring that the same measure in different groups actually measures the same construct in the same way
what are the four types of data (levels of measurement)
nominal, ordinal, ratio, interval
nominal scale
categories of qualitative variables
ordinal scale
ordered/ranked on some construct. E.g., likert scales
interval scale
allows for ordering and examining magnitude differences among responses. NO real zero point. (e.g., temperature)
ratio scale
variables have equal intervals and a true zero (e.g., distance)
when to use the mean as a measure of central tendency
for interval or ratio scales, when data is not highly skewed
when to use the median as a measure of central tendency
for ordinal data or when data are interval or ratio but highly skewed
what is the interquartile range?
captures the middle 50%: subtract 25th percentile from 75th percentile
how do you calculate variance
Numerator: subtract mean from each value and square. Sum all
Denominator: n-1
standard deviation vs. standard error
The standard deviation reflects variability within a sample, while the standard error estimates the variability across samples of a population
what is the standard error of the mean?
measures the precision of the sample mean to the population mean. Decreases as sample size increases
what is the formula to calculate z
(x-M)/SD
what is skewness, what is + vs -
departure from symmetry. Negative skew: more cases on higher end. Positive skew: more cases on lower end
what is kurtosis
relative peakedness of the distribution. Lepto (+) = more peaked. Platy (-) = flatter
long-range probability of type II errors
Beta
statistical power
probability of correctly rejecting a false null hypothesis
what factors affect power
sample size, alpha, directional vs nondirectional hypotheses, magnitude of the effect, reliabilty of measures
what does Cohen’s d measure
effect size. Reflects the standardized mean difference between two groups divided by the pooled SD.