Slides 5 Flashcards
what is the gold standard in psychology experimental designs
random selection
- every individual I that population has the same chance to be selected to participate in that research study
- improves external validity
why isn’t true random selection performed often if its the gold standard?
rarely feasible unless the population is very narrowly defined
when using convenience samples, what should the investigator do?
the investigator should evaluate and at least discuss whether unique features of the sample contributed to the results
what is equivalence of groups in a study dependant on? why?
N
- even if participants are randomly assigned groups, if your sample is very small, say 10, it could be that by chance one group is different than the other
- also: in small groups, there isn’t enough power there to determine if your groups are significantly different, so really, you’re in the dark
what to do to make sure groups are equivalent if sample of participants for study is small
match participants
- guarantees equivalence on key variables
- of course have to be cautious with this bc sometimes you may think you’re matching participants when in fact you aren’t
ex: multiple ways to get same IQ score
If the midterm includes an experimental design that says its nonrandom, what is it likely to be ?
quasi-experimental design
what does R mean in experimental design notation
random assignment
what does O mean in experimental design notation
observation/assessment
what does X mean in experimental design notation
treatment or manioulation
what is this design called
R O1 X O2
R O3 O4
What are it’s strengths? Weaknesses?
Pretest-posttest control group design
Strengths:
is period between pretest and posttest are equivalent than many threats to interval validity are controlled for
- allows for matching using pretest scores
- increased power, better examination across groups
- allows for predictors and descriptors of attrition
Weaknesses:
pretest sensitization
- did exposure to the pretest contribute to the response of the manipulation?
what is this design called
R X O1
R O2
What are it’s strengths? Weaknesses?
posttest only control group design
strengths:
pretest sensitization
- did exposure to the pretest contribute to the response of the manipulation?
weaknesses
is period between pretest and posttest are equivalent than many threats to interval validity are controlled for
- allows for matching using pretest scores
- increased power, better examination across groups
- allows for predictors and descriptors of attrition
what is this design called
R O1 X O2
R O3 O4
R X O5
R O6
what does it do?
Soloman Four-group design
evaluates the effect of the pretest
- typically only really used when we are sure that pretesting will produce problems
what is a factorial design
allows for simultaneous investigation of the effect of two or more variables
allows for testing of interactions
what is this design called
nonR O1 X O2
nonR O3 X O4
what does it do?
Quasi-experimental design
- includes pretest, posttest, and other variations
- doesn’t have random assignment
Why might an experiment have nonrandomly assigned subjects
subjects who are already in separate clinics, schools, or classrooms
- there is a reason they cannot be compared
what is this design called
R O1 XA O2 XB O3
R O4 XB O5 XA O6
what does it do?
crossover design
- subjects differ only in the order they receive tx
- design is optimal for 2 randomly assigned groups
what happens for multiple treatments counterbalanced designs
in crossover deisgns, the order of treatments are counterbalanced
increased number of treatments becomes more difficult
what is it called when an experimental design has sequences arranged where each treatment occurs in a positive only once (kind of like suduko)
a latin square
A B C D
B A D C
C D A B
D C B A
what is not good about latin quare designs
not every treatment is proceeded and followed by every other treatmet so its possible there is an interaction between a treatments affectiveness and its order to other treatments
What happens in a true counterbalanced design
there is random assignment to every possible sequence of treatments
what are sequence effects
arrangement of treatments contributes to effects - like carry over effects
what does a no treatment control group accomplish
helps rule of confounds
-
how do you assign your no treatment control groups
randomly assign potential subjects
- cannot use drop outs as controls (they are self selected as opposed to randomly selected
what often happens with no treatment control groups
may seek treatment elsewhere
they often drop out
what are waitlist control groups
use subjects who are waiting for treatment as a control group
treatment is given after treatment is over for group 1
- must only use participants who in advance agree to remain if placed in this group, otherwise could introduce serious confounds
waitlist group completes 2 rounds of pre-assessment and one of post
what are no contact control groups
they don’t even know they are in the study
- pre-test info from another sutyd or a large screening
- ethical issues
- subsequent test info obtained through a ruse so they don’t know its related to the study
what are nonspecific-treatment or “attention-placebo” control groups
nonspecific factors are controlled so the specific mechanisms of change can be somewhat isolated and evaluated
- control for common factors
- many ethical issues
- reduced faith in therapy
- leads individuals to never seek real treatment
- problem with withholding treatment
what are routine/standard treatment control groups
all receive an acceptable treatment
limited attrition
controls for nonspecific factors
more likely to have therapist compliance
what are yoked control control groups
control subjects are matched with subjects to complete aspects of the protocol
- issues with construct validity should drive this decision (rule out demand characteristics)
