4: Experimental research: Between subjects Flashcards
between subject variable
-Different groups of subjects are exposed
to each level of the variable
Multigroup designs
-Multiple experimental groups to simultaneously test for
effects of multiple levels of the IV
-If different levels of the IV represent quantitative
differences, the experiment has a parametric design
-Multiple control group designs may include both a
positive control (known effect) and a negative control
(known to have no effect)
keeping EVs constant
-Hold EVs constant in a study to avoid design confounds
-Balance EVs out across conditions to avoid selection
effects or order effects, depending on experimental
design
• Random assignment
• counterbalancing
design confound
-you are confused about whether the effect was caused by one variable or another
-To avoid design confounds, hold extraneous
variables constant
between subjects… single factor experimental designs
-designs must reduce selection effects resulting from individual differences
• Different groups of subjects are exposed to each condition
• Random assignment to conditions balances out subject variables
(individual differences)
• May use a matching procedure to better equilibrate groups
strength of experimental research
weakness
-Can identify clear causal relationships among variables
(which is not possible with correlational research)
- Requires that you be able to manipulate IVs
- may limit generalizability to real-world contexts (ecological validity may suffer)
randomized group designs
-At least two conditions are needed to constitute an
experiment (control and experimental treatment groups)
- Subjects are randomly assigned to conditions
- The IV is manipulated
-The DV is measured and
means are compared
randomzied two group post-test
-the simplest experiment to conduct
-However, the amount of information yielded may be
limited with only two groups
-Additional levels of the independent variable can be added to form
matched group designs
-first form matched sets of subjects and then randomly assign subjects – within that matched set – to groups
• Matching groups on a characteristic will better equilibrate the groups and minimize selection effects
• The matched subject variable will thus contribute very little to the difference between groups
•Used when a particular characteristic is thought to exert a strong influence on the dependent variable
• Can match based on multiple characteristics
matched pairs deisgn
- Obtain a sample
- Measure subjects for a certain characteristic that
may relate to the DV - Match subjects according to that characteristic to
form matched pairs of similar subjects - Randomly assign within each pair – one to Group A
and the other to Group B - Conduct experiment
matched multi-group design
-extends the matched pairs design to experiments
with three or more groups
• Must find a matched subject for each group
• Matched-multigroup designs can become unwieldy,
especially if matching on multiple characteristics
advantages matched group
disadvantages
advatages:
• Controls matched subject variable(s) that may otherwise be confounds
• Increases sensitivity to effects of your independent variable
disadvantages:
•Requires large pool of potential subjects and valid pretesting measure
• If matched characteristic has small/no effect, nothing is gained for all this effort and the power of statistical test is reduced
> Only use a matched-groups design when you are confident that the matched characteristic has a strong effect on the dependent measure
pretest/posttest
-may be measured on the DV both before and
after the IV manipulation in a pretest/posttest design
- Evaluate whether random assignment equilibrated the groups
- Track individual changes over time in response to IV manipulation
- But must be careful about the pretest itself affecting the posttest
