Ch9 and Ch10 Flashcards
What are within-subjects designs?
designs in which each participant is exposed to all levels of the IV
Advantages of within-subjects designs
- some kinds of confounding variables are less likely (e.g., participant variables)
- fewer participants required
- lower variability means statistical advantage (since test statistics are a proportion of effect over error)
- no control group issues
time-related threats to internal validity in within-subjects designs
- MRs smITH
- Maturation
- Regression to the mean
- Instrumentation
- Testing*
- History
Order effects in within-subjects designs
- Testing/Practice effects: subjects perform better due to familiarity/practice with testing procedure
- Carry-over/contrast effects: After-effects of one treatment remain/influence perception when moving on to another treatment
- Fatigue effects: participants perform differently at the end of the experiment
Cons of within-subject designs:
- TIME-RELATED AND ORDER-RELATED THREATS TO INTERNAL VALIDITY:
- MRs smITH threats
- carry-over/contrast effects
- fatigue effects
- IMPOSSIBLE FOR SOME RESEARCH QUESTIONS
What can be done to reduce the order effects in within-subject designs?
COUNTERBALANCING: order of levels of IV is varied
Method used for partial counterbalancing?
latin square: each condition occurs in each position once
In a study with multiple variables of interest, some factors are within-subject and some are between-subjects. What is this called?
a MIXED-METHODS design
Which design is typically preferred in psychology research? between- or within-subjects design?
within-subjects, because it minimizes variability between conditions and it’s fewer participants required. Participant variables are likely not confounding and no control group issues exist
In an experiment in which handedness and memory for images vs. words is tested, what variable is tested in a within-subject design, and which in a between-subjects design?
- both groups were exposed to all levels of the images and words IV, therefore within-subjects
- one group was left-handed and the other was right-handed and they were compared, therefore between-subjects
- overall, this is a mixed-methods design
In what kind of experimental design are researchers NOT concerned with trying to maximize internal validity?
in a non-experimental design
When are nonexperimental or quasi-experimental designs necessary?
Often they are necessary when it would be too difficult or unethical to perform a true experiment.
Why can we not be certain of cause and effect relationships in nonexperimental/quasi-experimental designs?
- Because the levels of the IV are pre-determined, NOT MANIPULATED by experimenter
- True random assignment is not possible
- we cannot counterbalance to avoid order effects
Differential research design
- between-subjects
- nonexperimental
- groups are designed by pre-existing categories
There are two tables, both with two columns.
The first one has the test scores of people diagnosed with GAD and people not diagnosed with GAD.
The second table has the scores on an anxiety measure next to test scores. What kind of research is each table from?
- first one is differential. Pre-determined groups being compared.
- Second one is correlational
The design which involves one group being measured following treatment is called…
single group posttest only design
Is the posttest only design experimental/nonexperimental/quasi-experimental?
NONEXPERIMENTAL
A researcher is interested in whether nonconscious factors could disrupt test performance. He plays supersonic noise maker to a class to see if it disrupts performance on a spelling task. He concludes that is did, because performance was at 70% What is wrong with his conclusion? What could he do
-He has nothing to compare his findings to to see whether a true effect is observed.
-His test could have been too hard to begin with (floor-effect)
-He could add a non-equivalent control group by giving the spelling test to a different class without the sound maker.
-still nonexperimental due to pre-existing groups
-he could also employ a pretest-posttest design with added non equivalent group.
one class does spelling test, then noisemaker, then spelling test. The other class does spelling test, then no noisemaker, then spelling test. QUasi-experiment
What is a single group pretest-posttest design?
- nonexperimental
- differential
- pre-existing groups
- SINGLE GROUP MEASURED BEFORE AND AFTER TREATMENT
How can you turn a pretest-posttest nonexperiment into a quasi-experiment?
by adding a NON-EQUIVALENT CONTROL GROUP
Why is a pretest-posttest design with a non-equivalent control group added a quasi-experiment?
because there is an increased level of control.
Two groups are tested and compared. This helps rule out alternative explanations
What kind of design includes doing multiple measurements before and after treatment?
Time-series Designs
Are time-series designs experimental?
NO. They are quasi-experimental designs
time-related threats to internal validity which can be checked for in time-series designs
Regression to the mean, maturation, instrumentation, testing, history.
MRs smITH