10/5/2015
Repeated Measures Designs
Outline
Repeated measures Counterbalancing designs Complete Why / Why not Balanced Latin Square Carry-over effects Latin Square Order vs sequence APA Discussion Counterbalancing Randomization Block randomization A BA / Reversal
Repeated Measures Designs
A repeated measures design is one in Condition which every 1 2 3 participant participates in every condition of the experiment Historically called a within-subjects design
1 10/5/2015
Why Use Repeated Measures Designs?
Uses fewer participants More efficient Study changes in participant’s behaviors across time Increased sensitivity
Repeated Measures Designs
Repeated measures designs increase sensitivity by reducing error variance that would be due to having different participants in the various conditions (i.e. it increases statistical power) Repeated measures designs should only be used when carry-over effects are unlikely Carry-over effect occurs when participating in a prior condition influences participation in the current condition Order effects Sequence effects
Order vs Sequence Effects
An order effect occurs when the absolute order of conditions influences the results Later performance is often better than early performance because of practice
2 10/5/2015
Order vs Sequence Effects
A sequence effect occurs when the relative order of conditions influences the results Receiving the context condition prior to the no-context condition Also know as differential transfer
Counterbalancing Techniques
Counterbalancing techniques determine the order of presentation of the conditions in a repeated measures design They attempt to control (not eliminate) order effects They neither control nor eliminate sequence effects
Counterbalancing Techniques
Two broad categories of counterbalancing techniques: Complete designs – counterbalancing occurs within each participant Randomization Block randomization Reverse counterbalancing (ABBA) Incomplete designs – counterbalancing occurs across the group of participants Complete counterbalancing Balanced Latin Square Latin Square
3 10/5/2015
Techniques
Randomization -- for each participant, present the conditions multiple times in a random order
Techniques
Block randomization -- for each participant, present the conditions in a random order with the restriction that every condition must be presented once before any condition can be repeated Each condition should be presented multiple times
Techniques
Reverse counterbalancing -- when the conditions are to be presented twice, the second presentations are in the reverse order of the first presentations Sometimes called the ABBA design Works best when the effects are linear over time; 1 + 4 = 5; 2 + 3 = 5
4 10/5/2015
Techniques
Complete counterbalancing occurs when all possible orderings of the condition are used in the experiment: Order 1: A B C Order 2: A C B Order 3: B A C Order 4: B C A Order 5: C A B Order 6: C B A Randomly assign participants to the various orders
Techniques
Complete counterbalancing has each condition appearing in each ordinal position of the experiment an equal number of times has each condition preceded by every other condition an equal number of times has each condition followed by every other condition an equal number of times should not be used when the number of conditions is larger than approximately 4 should use a number of participants that is a multiple of the number of orderings
Balanced Latin Square
The first sequence of conditions is given by the pattern: 1 2 n 3 n-1 4 n-2 5 n-3 … until you have as many numbers as conditions (even number of conditions only)
5 10/5/2015
Techniques
After creating the first row, count up as Order 1 1 2 4 3 you go down the Order 2 2 3 1 4 columns Order 3 3 4 2 1 Block randomly assign participants to Order 4 4 1 3 2 the orders The number of participants should be an exact multiple of the number of orders
Techniques
1 2 5 3 4 4 3 5 2 1 2 3 1 4 5 5 4 1 3 2 3 4 2 5 1 1 5 2 4 3 4 5 3 1 2 2 1 3 5 4 5 1 4 2 3 3 2 4 1 5 When there are an odd number of conditions, the table is created as before and then reflected Each condition is presented twice Block randomly assign participants to the various orderings of conditions
Techniques
A Latin square is a technique that determines the order Order 1 1 4 5 3 2 of presentation Order 2 2 5 1 4 3 List a random order of Order 3 3 1 2 5 4 the conditions (1, 2, 3, Order 4 4 2 3 1 5 4, 5) across the top Order 5 5 3 4 2 1 Sequentially count up as you go down the columns
6 10/5/2015
Techniques
The Latin square ensures that each condition appears in each ordinal position of the experiment an equal number of times However, not every condition is preceded by and followed by every other condition an equal number of times Block randomly assign participants to the various orderings of conditions
Problems with Counterbalancing
Counterbalancing does not eliminate carry-over effects It attempts to equally distribute order effects over all the conditions Can promote “order of conditions” to an IV and test to see if it has an effect
Discussion
Support or non-support of each hypothesis / prediction Compare / contrast your results to those in the literature Discuss! Why are the findings important? What do they tell us about the theory behind the research? How can the results be applied?
7 10/5/2015
Discussion
Discuss alternative explanations of the results, if any Discuss any non-trivial limitations, if any Trivial limitations (e.g. sample size) will result in an editor telling you to fix it and then re-submit Future research
Discussion
Starts immediately after the results For us, with a single study, use a level one heading with Discussion Include appropriate citations when relating your results to the literature
8