Chapter 13 Quasi-Experiments Flashcards
how are quasi-experiments similar to experiments?
has IVs and DVs
main difference between quasi experiments vs experiments
quasi experiments don’t have full experimental control and don’t use random assignment
when are quasi experiments typically used?
in situations where random assignment is unethical (real world situations that can’t be manipulated)
what does the term “nonequivalent” refer to in quasi designs?
no use of random assignment
types of quasi designs
nonequivalent control group posttest only
nonequivalent control group pretest/posttest
nonequivalent control group interrupted time series
2 main characteristics of a nonequivalent control group design
- no use of random assignment; instead participants were either all born with something, exposed to something naturally occurring, etc.
- at least 1 treatment group and 1 comparison group
what comparisons do we want to make in nonequiv posttest only designs?
between-subjects
which threat is heavy in nonequiv posttest only design?
selection threats
nonequivalent pre/post design controls most threats to internal validity except for?
selection effects
when is the DV measured in interrupted time-series design?
repeatedly (before, during (could be several times) and after)
3 pros to interrupted times series design
-results are interpretable
-cam see normal fluctuation and trends
-can see how long lasting an effect is
why is an interrupted time series design better than a 1 group pre/post design?
there’s more data points which makes it easier to interpret
how many groups does an interrupted time-series design have?
2 or more
does the timing of the interruption differ b/w groups in interrupted-time series designs?
yes
which effects are found in quasi-experiments
selection effects; we don’t know if the change in DV is due to the IV or a shared participant characteristic
solutions to eliminating selection fx in quasi experiments
- create matched groups
- waitlist: upgrade the quasi to a true experiment
threats found in quasi experiments
design confounds
maturation threat
history threats
regression to the mean
attrition
testing and instrumentation threats
observer bias
demand characteristics
placebo fx
maturation threats are typical in which designs
pretest-posttest
which threats can be addressed by adding a comparison group in quasi experiments
maturation
history
testing
how to control for observer bias in quasi experiments
use a masked or double blind design
pros of carefully designed quasi experiments
-provides real-world data (not simulated experiments)
-typically high external validity
-allows us to study issues that cannot be ethically studied in experiments
-high construct validity of the IV
similarities b/w quasi and correlational designs
-no random assignment
-both prone to internal validity threats
key differences b/w quasi and correlational designs
-in quasi: we actively seek out naturally occurring comparison groups
-quasi is more meaningful/closer to establishing causation
small N vs large N designs
small N: limited number of participants, high amount of data for each participant
large N: high number of participants, their data gets averaged out and we learn less about each person typically
adv and disadv for small N designs
adv: studies special cases, can have some experimental control if compared to matched control groups
disadv: external validity, internal validity sometimes
small N designs are commonly used in which field of psychology and why?
therapy/behavior analysis, because it allows for modifying the behaviors of certain individuals to produce a desired result
3 small N designs
stable-baseline
multiple-baseline
reversal
stable-baseline design
an extended observation of baseline before intervention is introduced
when do we know a stable-baseline design was effective
if the target behavior changes only after intervention
what threats can stable-baseline designs rule out
maturation
regression to the mean
multiple baseline design
staggers interventions across times, settings, or situations
reversal designs
target behavior improves during treatment but reverses to baseline when treatment is removed
when are reversal designs used
when treatment does not have a long term effect
discuss 4 big validities for small N designs
internal: can be high if the study was carefully designed
external: can be problematic depending on the goals of the study
construct: can also be high if definitions and observations are precise
statistical: not always relevant to small N studies bc we’re looking at a limited number of participants