Chapter 10 Flashcards
Single case experimental designs
*Popular in clinical psych
* Research studies carried out on only one
participant.
* Subject’s behaviour is measured over
time during a baseline control period
* Reversal or ABA design
ex. child has messy room
Offer reward for cleaning room
one week latter, room is clean
ex. baseline, treatment baseline
ABA Reversal Designs
How often a child complied with teacher’s request
behaviour is measured in baseline treatment, then behaviour is measured again
ex.
Instate a 5-minute play time reward
behaviour improves
Remove reward
Extensions of the ABA Reversal
Design
ABA reversal designs can be improved by
extending them
* ABAB or even ABABAB
* Rules out alternative hypotheses
* More ethical
- you can add another exposure to improve results
ex. baseline treatment then baseline treatment
ending with treatment: better
Multiple Baseline Designs
Sometimes it’s not possible to remove treatment from something.
In these cases you take multiple measures over time can be made before and after the manipulation
* Effectiveness demonstrated when a
behaviour changes only after the manipulation
* Several variations:
* across subjects:
* across behaviours:
* across situations:
ex. introduce intervention at different times
Replication
- key to strengthening (often confirm or deny prior study)
- Procedures with a single subject can, of course, be replicated with other subjects
- Data usually presented by showing just individuals rather than averaging across
all people - Difficult to do statistics (not great), but procedures developing
Program Evaluation
- high demand
- researcher evaluates program in an attempt to achieve positive improvement in the program
- Researchers evaluate proposed and implemented programs
Questions that they might ask:
- Needs assessment: Are there problems that need to be
addressed?
- How will the problems be addressed?
- Is the program addressing the needs?
- Are the intended outcomes being
realized?- does it work
- Is the cost justified by the outcomes?- did people improve enough to justify the cost
- Are the services reaching target populations, is it ethical?
Quasi-Experimental Designs
- Often conducted in field settings
random assigned participants is not possible
-Used when the control of true experiments can not be achieved - used in psychotherapy educational intervention.
- Cause-and-effect more difficult (because there’s no random assignment-causes lower internal validity)
- No random assignment
- Lower internal validity
- Are the intended outcomes being
realized? - Is the cost justified by the outcomes?
Quasi-Experimental Designs
One-group Posttest Only Designs (simplest)
- can’t really rule out alternative explanations
- fast simple evidence
- maybe pilot test
- not acceptable for something published
* Compare the responses of a number of individuals exposed to the same event
* Lacks a control or comparison group
* Internal validity compromised
One-group Pretest-Posttest Design
* Adds a baseline measure to provide basis for
comparison
* Cannot rule out intervening variables
ex.
how many cigarettes people smock
- try and relax them
- how much they smoke now?
Without random assignment, you don’t really know if the relaxation is causing them to stop smoking or if it’s something else.
Non-Equivalent Control Groups (specific kind of quasi experimental design)
Post test Only Designs
* Separate control group
* Not equivalent due to lack of random assignment
* Selection differences
ex. 1 group gets relaxation
2 other group does not
see which one smokes less
often used when control groups are necessary but random assignment isn’t possible
Non-Equivalent Control Groups
Pretest-Posttest Designs
* An improvement on the non-equivalent control group design
* Improves internal validity
* Can assess selection differences prior to testing
- Add a pretest
- more control you know result is from change
- look from pretest to post test
Interrupted Time Series Design
Looks at dependent variable over time
Measure before and after iv
Useful for evaluating some polices but has limitations
Control Series Design
Improves Interrupted Time Series by adding a
control group
* Involves finding a similar population that did not receive a particular manipulation
Developmental research design
Studying the ways that individuals
change as a function of age
* Cross-sectional
* Longitudinal
* Sequential
Longitudinal
Look at the same group at different points in time
ex. test one group every 5 years
attrition is an issue
takes a long time
most often seen in health research
Cross sectional
looks at different ages studied at one point in time
- less expensive
- faster
- vulnerable to cohort effects
