Midterm 1 - Lecture 4 (Ch8) Flashcards
Random Assignment of people to condition (Experimental Design)
- Balances influence of random effects across the conditions
- Especially as sample size increases
What are the potential Experimental Design options (3)
- Independent Groups
- Repeated Measures
- Matched Pairs
Independent Groups - Between Groups (Experimental Designs Options)
Between-subjects design: in which different people experience the levels of the independent variable
- The experimental and control groups are distributed between different people
3 broad steps to designing a between-subjects design:
- Obtaining two approximately equivalent groups of participants
- Introducing different levels of the IV to the participants
- Measuring the IV
First step to planning a between subjects design is…
To decide how to assign participants to the levels of the IV
- Must create equivalent groups and eliminate any potential selection differences: the people selected to be in experimental condition should not differ in any systemic way from those selected to for the control condition
- Prevented through random assignment
Second step (to any experiment, really) is to…
Operationalize the IV, creating at least 2 levels
- This can look like an exp. group that receives treatment, and a control that doesn’t
Third step to planning a between-subjects design:
- Third step: planning an experiment to operationalize the depended variable, which will allow us to measure the effect of the IV on the DV
- Same measurement procedure is used for both conditions, so that they can be compared
When the results can be confidently attributed to the IV, the experiment is said to have…
high internal validity
If the scores of the dependent variable differ between groups, researchers can conclude that…
the independent variable is based on the logic that if the only difference between the groups is the manipulated independent variable, then this difference must be the cause of any difference in the measured depended variable
In random assignment for Between-Subjects, participant characteristics cannot be…
an alternative explanation to an experiment’s results
Repeated Measures - Within Groups (Experimental Designs Options)
In which all the same people experience all levels of the IV
-All participants experience all conditions; ensures that the groups for each condition are absolutely identical
- Removes all possibility of selection bias
Benefits of Repeated Measures:
□ Need fewer participants (because everyone participates in all conditions)
□ Less cost (in times when participants are scarce - can also maximize data collection)
□ More power to see the effect of IV on DV if there is one (When the exact same people are in both conditions; By accounting for individual differences in a within-subjects design, we are better able to detect an effect of the IV on the dependent variable, if one exists)
Problems with Repeated Measures Designs:
□ “Mortality”
□ Length of time interval
□ Dealing with counterbalancing orders
□ Order effects
Carry-Over Effects/Order Effects (Repeated Measures)
Whenever a participant’s response in a given condition is affected by having previously participated in other conditions in that experiment.
3 types of order effects:
◊ Practice effects
◊ Fatigue effects
◊ Contrast effects
How can we address order effects?
- Filler tasks (for contrast effects); meditating
- Passage of time (for fatigue & contrast effects); Can be minutes, hours, days, weeks…
How can we deal with counterbalancing orders?
Balanced Latin Squares:
- Each treatment must occur once with each participant
- Each treatment must occur the same number of times for each period or trial
- Each treatment must precede and follow every other treatment an equal number of times
Pretest-Posttest Design:
Pretest-posttest design: an experimental design in which the dependent variable is measured both before and after manipulation of the independent variable
- To ensure equivalent random assignments, researchers can choose to add a pretest to measure that variable before any experimental manipulation
- Scores are then compared to ensure before any experimental manipulation
3 Advantages of a Pretest-Posttest Design
- To counter problems associated with a small sample size
- To select appropriate participants
- When participants might drop out of the study
“To counter problems associated with a small sample size” (Pretest-Posttest Design)
Although RA is likely to produce equivalent groups, as sample size decreases, it becomes less likely that the groups will be approximately equal
Thus, a pretest enables the researcher to asses whether the groups were already roughly equivalent on some critical variable before the manipulation began
“To select appropriate participants”(Pretest-Posttest Design)
A researcher might use a pretest to find the lowest/highest scorers on a measure of smoking, math anxiety, etc.
Once targets are identified, they’re randomly assigned to the experimental or control group
However, this method may introduce another issue: regression toward the mean
“When participants might drop out of the study” (Pretest-Posttest Design)
To ensure that those who remain do not differ across conditions
Participants choosing to leave a study can produce a problem known as selective attrition, with dropouts rates from long-term studies
If people are dropping out for some reason related to the manipulation, this selective attrition may cause a difference between groups
Can become an alternative explanation for results
Thus, pretests can enable researchers to examine whether selective attrition is a plausible alternative
Disadvantage of Pre-test/Post-test Design
Can sensitize participants to what you’re studying, enabling them to figure out your hypothesis
May react differently to the manipulation than they would have without the pretest
Difficult to generalize these results to people who have not received a pretest
How can we prevent sensitizing participants to what you’re studying?
How can we prevent this?
Pretests can be disguised using deception
Embed the pretest in a set of irrelevant measures, so that it’s not obvious that the researcher is interested in a particular topic
® Examine the influence of a pretest directly, by including the presence or absence of a pretest as another condition, embedding the levels of the IV within those two groups
Matched Pairs Design:
Matches people on a crucial participant characteristic (EX: pairs of people with similar age, need for control, typing speed, etc.)
One member assigned control group, other member assigned experimental
Can be either the DV itself or another variable
When are matched pairs designs most likely to be used?
When not possible to obtain a large sample size, making RA less likely to be successful in created equivalent groups
Large numbers can be too pricey
