FINAL - SSD Flashcards
RCT
randomized controlled trial
gold standard for experimental design and why?
RCT
incorporates large samples, randomization, and control groups
RCT inferential flaw
can affect decision making with individual people: the average result does not represent every person
RCT cannot track:
unique or disparate responses among individuals, nor determine who is a responder or non responder
RCT may lack:
replication
reduces confidence in the outcome
single subject design
uses one subject as the population of interest and the subject serves as her/his own control
measurements from ssd are:
sampled and analyzed over time
ssd: kind of case study
quantitative, empirical, and objective
ssd: internal validity
achieved by designs that control extraneous variables and by statistical tests that meet their underlying assumptions
ssd: external validity
achieved by replication
internal validity
cause effect
external validity
generalization
why are single subject designs needed?
- when one subject is the population of interest
- when few subjects with a target condition are available for a larger study
- to test a clinical hypothesis, identify the effectiveness of specific treatments, and identify responders and non responders
- in early phase (pre clinical) research
- to cast doubt on existing assumptions or theories
- when between subject heterogeneity is large, and group averaging threatens statistical inference validity in group designs
aggregate type proposition
makes an assertion about a group of individuals considered as a whole; lawful relationship about averages
general type proposition
makes an assertion about each member of the group; lawful relationship about individuals
for a law to be universal,
it must produce or predict the same result in every person
positive change
positive effect
zero change
no effect
negative change
negative effect
life saving or quality of life improving (responders)
positive effect
no benefit, no harm (non responders)
no effect
contraindicated due to potential harmful effects
negative effect
ssd are sometimes considered unscientific because they do not seek a general law
criticism of ssd
ssd are perceived as unable to generalize to the population
criticism of ssd
ssd are quasi experimental bc they do not use a control group or randomly assign subjects to treatments
criticism of ssd
observations from a subject may not be independent, which affects the statistical analysis options
criticism of ssd
ssd are perceived to provide a low level of evidence
criticism of ssd
common ssd
baseline treatment design (AB)
treatment withdrawal design (ABA)
treatment replication design (ABAB)
multiple baseline and multiple probe design across behaviors (ABAB w more than 1 behavior)
multiple baseline and multiple probe design across participants (ABAB with more than 1 participant)
ABAB w more than 1 behavior
multiple baseline and multiple probe design across behaviors
ABAB with more than 1 participant
multiple baseline and multiple probe design across participants
AB
baseline treatment design
ABA
treatment withdrawal design
ABAB
treatment replication design
single subject experimental designs come in different forms, but all use
- repeated measurements across time
- baseline phase
- treatment or intervention phase
some use: - post-treatment phase, to establish generalization or maintenance
AB designs show changes in
level, trend, variability
AB is the basis for
more complex designs, but ALONE is considered a pre experimental design bc it does not adequately control many sources of variation that threaten internal validity and does not demonstrate replication
ABA examines the effect of
a single treatment (independent variable) on an outcome measure (dependent variable)
why is ABA stronger than AB?
provides a second demonstration of the treatment effect, thus improving confidence that the treatment, and not some extraneous variable, caused the outcome measure’s change
ABAB is an extension of
ABA design
ABAB provides
a third demonstration of the treatment effect on the outcome measure
ABAB designs are straightforward and among the
strongest ssd
ABAB with 2 behaviors
probes the treatment effect across experimentally defined conditions: behaviors, people, stimuli, settings
strength of ABAB with 2 behaviors
multiple demonstrations of Tx effects by replication across behaviors
appropriate design for nonreversible behaviors
ABAB with 2 behaviors
ABAB with 2 participants
participants graphed separately, unlike behaviors
criteria to assess the quality of a ssd study;
- actively manipulate ind variables
- systematically measure meaningful (externally valid) dep variables over time and by more than one scorer
- use ind scorers who a blinded to the participants and phases to measure tx effect
- use 5 OR MORE observations per phase to evaluate behavior stability and improve representation to reduce small sample bias
- provide 3 OR MORE opportunities to evaluate the experimental effects using replication WITHIN (internal validity) or BETWEEN (external validity) subjects
- provide a detailed description of the tx (ind variable) so that it can be administered consistently and replicated across experimenters
ssd visual analysis
the basis for a ssd study, however it can lead to bias and type 1 statistical error
ssd statistical analysis
can support the visual outcomes with objective, statistical inference based data and improve confidence in the effect
statistical analysis options in a ssd include:
test designed for time series ss experimental designs
parametric and non parametric tests, like those used for group designs
analysis of variance acronym
ANOVA
ANOVA
a test used to determine differences between research results from 3 or more unrelated samples or groups
C statistic
an omnibus test for abrupt changes in level and gradual changes in the trend of time-series data
ratio of variances like a t- or F- ratio
metric that characterizes the strength of a tx and is often used to compliment p values
effect size
one measure of effect size in ssd
single subject designs
100% overlap =
no treatment effect
PND
calculated as the # of tx values that do not overlap with the baseline values
single subject research desings:
- provide evidence of treatment effects in individuals,
- are flexible, efficient, and can be implemented in clinical practice and
research to support EBP and discipline-based knowledge, - are an excellent option for studying small or very heterogeneous
populations, - provide strong internal validity when well-implemented,
- provide external validity (generalization) when replicated.
investigating the individual is important for:
practical, statistical, and philosophical reasons