Week 6: Quasi, N-of-1, Other Designs Flashcards
Quasi-experimental designs lack what
Random assignment, control group, or both
Time Series Design types
one-group pretest-posttest design
Repeated measures design
interrupted time series design
Single group designs have what as the independent variable
time
One-Group Pretest-posttest design
-all subjects have the same treatment
- no control group (limits in internal and external validity)
- IV is time
- IV has 2 levels
A One-Group Pretest-posttest design is defendable when
- the behavior of a control group has been documented
- ethical implications of withholding treatments
- time interval is very short
True experiments (RCTs) CAN have a pretest and a posttest BUT when change score are analyzed…
it removes time as an IV
Repeated Measures design is NOT a …
“true experiment”
Repeated Measures design has NO…
comparison (control) group
Interrupted Time Series Design (ITS)
- multiple DV measures
- interrupted by 1+ treatment occasions
- treatment does not vary
- IV is time
- no control group
Multiple measures within an ITS =
sudo-control
nonequivalent =
not formed by randomization
types of Nonequivalent Group Designs
- nonequivalent pretest-posttest control group design
- historical controls
- nonequivalent posttest-only control group design
Pretest-posttest design
- no random assignment
- intact groups
- subject preferences
(some control over threats to internal validity)
Historical Controls include
- received a different treatment during an earlier time period
- imbalance in characteristics of the groups
- hard to use information from other studies cause you can’t control the past study
Nonequivalent Posttest-only control group design
- no random assignment
- no pretest
- exploratory purposes NOT explanatory
Which design type has the most threats to validity
Nonequivalent posttest-only control group design
Downsides of Group Experimental Research
- require control groups and large numbers of subjects
- very time intensive
- too few measurements
- treatments are standardized
- usually not feasible for clinicians
- “averaging” results and losing individual variation
Single-Subject Design Studies
individual patients
within-subject design
Length of Phases
important to extend phase until stability is reached
Target behavior
needs to be quantitative
Single-subject A B Design
baseline phase f/b treatment introduction
a: baseline
b: treatment
Limitations of the A B Design
- no control comparison
- impossible to conclude causality
- strengthen by replication
Single-subject “B C” Design
B: control intervention
C: experimental intervention
A-B-A-B design is
control, intervention, control, intervention
Multiple baseline across subjects
same intervention is introduced to 3+ patients after varying baseline phases
Multiple baseline across settings
one individual monitored in multiple settings with same intervention
Multiple baseline across behaviors
one treatment; multiple clinically-related behaviors
N-of-1 is what type of research
experimental research
observational studies are a type of _________ research
exploratory
types of longitudinal studies
prospective and retrospective
prospective =
exposed/unexposed ID beforehand
Retrospective =
outcome is known, researchers look for exposure in records
cross-sectional studies
“snapshot”
tends to take less time and less money
Risk in PT can be expressed in terms of:
experiencing an adverse outcome
patients not improving with treatment
requiring more invasive or expensive interventions in spite of treatment
Factors to Consider for Causality
- time sequence
- strength of the association
- biological credibility
- consistency
- dose-response relationship
How are cohort study participants grouped together
exposure
Challenges for cohort studies
- time
- bias and attrition
- misclassification of exposure
- outcome may not occur in sufficient numbers for robust statistical analysis
Case-control studies
- participants are grouped by outcome
- cases vs controls
Challenges for case-control studies
- no randomization
- collecting data from medical records
- observation bias
- recall bias
Descriptive Research types
developmental research
normative research
descriptive surveys
case reports
historical research
Developmental Research
- document the natural history of the phenomenon of interest
- longitudinal
- cross-sectional
Descriptive Research
- does not aim to establish causation or differentiate between IV and DV
- case studies and descriptive surveys are most common
