Within Subjects Designs AKA Repeated Measures Design Remember matched groups design? –Allowed more power if you matched on correct variable –Within Subjects.

Within Subjects Designs

• AKA Repeated Measures Design

• Remember matched groups design?– Allowed more power if you matched on

correct variable– Within Subjects Designs can be considered

the ultimate way of matching.

Reasons to use Within Subjects Designs

• 1. Requires fewer subjects

• 2. Often more convenient or efficient• ludwig and DeWitt (1993)

• 3. More sensitive (has more power)• Infant smiles

• 4. Sometimes the question requires a within Ss Design

• Learning studies – acquisition

Issue to worry about with within Ss designs

• Progressive Error– Test person over and over again– Might get better

• Practice effects

– Might get worse• Fatigue effects

• Which threat to internal validity are we talking about?– Maturation

Progressive Error

• Kahneman, Fredrickson, Schreiber, and Redelmeier (1993)– Pain Perception– Hypothesis

• It is the final moments of a painful episode that we remember the most.

• Duration of pain plays a smaller role

Potential Design• Long trial

– Keep hand in 14 degree Celsius water for 60s• Then leave hand in for an additional 30 seconds while

temperature is raised 1 degree to 15 degrees• 7 minute break• Short trial

– Keep hand in 14 degree Celsius water for 60s• Asked participants which trial they would prefer to repeat

– 69% chose Long trial.• Supports notion that final moments are more memorable. (14

degrees vs 15 degrees)– IV? DV?– Any issues?

Actual Design

• Randomly assigned participant to one of two balancing conditions

• Half of Ss– Long/break/Short 72% 28%

• Other half of Ss– Short/break/Long 33% 67%

• Collapse across balancing condition– 69% prefer long trial

Balancing Progressive Error

• Subject x Subject Balancing– progressive error is balanced for each

participant– each participant experiences the conditions of

the experiment more than once, and in a different order each time

– Each person experiences a balanced presentation of conditions

Balancing Progressive Error

• Across Subjects Balancing– progressive error is balanced across different

participants– Each participant only experiences each condition of

the experiment one time.– different participants experience different orders of

conditions– Thus, an individual participant would not be balanced,

but after we average across participants the design would be balanced.

• Which balancing technique was used in the cold water (pain) experiment?– Subject x Subject or Across Subjects?

Balancing Progressive Error

• Subject x Subject– Block Randomization– ABBA counterbalancing

• Across Subjects– Complete (All Possible Orders)– Partial Counterbalancing

• Latin Square• Systematic Sequential Rotation

example of subject x subject balancing

• Sackheim, Gur, and Saucy (1978)– does one side of our face express more emotion?

• brain lateralization – left brain – right brain

» contralateral

– composite pictures• left composite• right composite• normal

– six emotions• happy, sad, surprise, fear, anger, and disgust

Latin Square

• 1) Randomly order the conditions of the experiment – we will do a study with 6 conditions

• Let’s say 6 perfumes or colognes

– Perfume and cologne A, B, C, D, E, F– F=1– C=2– B=3– E=4– A=5– D=6

Latin Square• To generate the first order use the following rule

– 1, 2, N, 3, N-1, 4, N-2, 5, N-3, 6.........• N – refers to the number of conditions in the study

• With 6 conditions our rule would be– 1, 2, N, 3, N-1, 4

• So our first row (first order of conditions would be).– 1, 2, 6, 3, 5, 4 or F, C, D, B, A, E

• Now to generate the other conditions we simply add 1 to each column– 1, 2, 6, 3, 5, 4 or F, C, D, B, A, E– 2, 3, 1, 4, 6, 5 C, B, F, E, D, A– 3, 4, 2, 5, 1, 6 B, E, C, A, F, D– 4, 5, 3, 6, 2, 1 E, A, B, D, C, F– 5, 6, 4, 1, 3, 2 A, D, E, F, B, C– 6, 1, 5, 2, 4, 3 D, F, A, C, E, B

Latin Square• Notice each condition occurs in each of the six possible positions (columns)

– F, C, D, B, A, E– C, B, F, E, D, A– B, E, C, A, F, D– E, A, B, D, C, F– A, D, E, F, B, C– D, F, A, C, E, B

• Also notice that each condition occurs after each of all of the other conditions

– F, C, D, B, A, E– C, B, F, E, D, A– B, E, C, A, F, D– E, A, B, D, C, F– A, D, E, F, B, C– D, F, A, C, E, B

Systematic Sequential Rotation

• Randomly order the conditions– 1, 2, 3, 4, 5, 6 or F, C, B, A, E, D

• Now simply add 1 to each column until you have 6 rows– 1, 2, 3, 4, 5, 6 or F, C, B, A, E, D– 2, 3, 4, 5, 6, 1 or C, B, A, E, D, F– 3, 4, 5, 6, 1, 2 or B, A, E, D, F, C – 4, 5, 6, 1, 2, 3 or A, E, D, F, C, B– 5, 6, 1, 2, 3, 4 or E, D, F, C, B, A– 6, 1, 2, 3, 4, 5 or D, F, C, B, A, E

Systematic Sequential Rotation• Notice each condition still occurs in each of the six possible conditions

– F, C, B, A, E, D– C, B, A, E, D, F– B, A, E, D, F, C – A, E, D, F, C, B– E, D, F, C, B, A– D, F, C, B, A, E

• However each condition is no longer preceded by each of the other conditions. It is always the same condition that precedes.

– F, C, B, A, E, D– C, B, A, E, D, F– B, A, E, D, F, C – A, E, D, F, C, B– E, D, F, C, B, A– D, F, C, B, A, E