CHAPTER 11: WITHIN-SUBJECTS DESIGN 1.

It requires each subject to spend more time in

the experiment.
Within-Subjects Design- a design in which each subject 2. Fatigue Effects
take part in more than one condition of the experiment;
also called a repeated-measures design • Interference between Conditions
1. Participants understand the aims and goals of
the study.
A WITHIN-SUBJECTS EXPERIMENT: HOMOPHONE
PRIMING OF PROPER NAMES

 We can set up a variety of within-subjects CONTROLLING WITHIN SUBJECTS DESIGN

design. The basic principle remains the same: Controlling for Order Effects: Counterbalancing
Each subject takes more than one condition of
the experiment Counterbalancing- A technique for controlling order
 These designs are advantageous because they effects by distributing progressive error across the
enable us to compare the behavior of the same different treatment conditions of the experiment
subjects under different treatment condition.
 Therefore, if responses change across
conditions, the changes are more likely to be Kind of changes that occur when subjects are run in
caused by the independent variable. Because it more than one condition
controls for individual differences among
1. Fatigue effects- changes in performance caused
subjects, the greater power of a within-subjects
by fatigue, boredom, or irritation as the
design also allowed the researchers to use
experiment goes on
fewer subjects.
2. Practice effects- change in subjects’
performance resulting from practice
WITHIN-SUBJECTS FACTORIAL DESIGN

A factorial design in which subjects receives all the Progressive error- changes in subjects’ responses that
conditions in the experiment. are caused by testing in multiple treatment conditions;
includes order effects, such as the effects of practice or
fatigue
MIXED DESIGN

A design that combines within- and between-subjects

Subject-by-Subject Counterbalancing
variables in a single experiment.
A technique for controlling progressive error for each
individual subject by presenting all treatment conditions
ADVANTAGES OF WITHIN SUBJECT DESIGN more than once.

• We use the same subjects in different Two common techniques used to create subject-by-
treatment conditions subject counterbalancing
• We have the best chance of detecting the effect
1. Reverse counterbalancing- a technique for
of independent variable.
controlling progressive error for each individual
• We can get an ongoing record of subjects'
subject by presenting all treatment conditions
behavior overtime.
twice, first in one order, then in the reverse
order.
2. Block randomization- often use when
DISADVANTAGES OF WITHIN SUBJECT DESIGN
progressive error is nonlinear. A process of
• Practical Limitations randomization that first creates treatment
 blocks containing one random order of the • Order effects - emerge as a result of the
conditions in the experiment; subjects are then position of a treatment of a sequence (1st, 2nd,
assigned to fill each successive treatment block 3rd, etc.) it doesn't matter what the specific
treatment is, if it occurs first in a sequence,
subject will handle it differently that if it occurs
Across-Subjects Counterbalancing last.

A technique for controlling progressive error that • Carryover effect - a function of the treatment
pools all subjects’ data together to equalize the itself.
effects of progressive error for each condition

Across-subjects techniques We can control carryover effect to some extent by

1. Complete counterbalancing- a technique for using some of the same counterbalancing procedure
controlling progressive error using all possible that control for order effects.
sequences that can be formed out of the
• Subject-by-subject counterbalancing
treatment conditions and using each sequence
the same number of times • Compete counterbalancing
2. Partial counterbalancing- a technique for Avoid using randomized counterbalancing and Latin
controlling progressive error by using some square counterbalancing because control is less certain.
subset of the available sequences of treatment But it is possible to construct a special Latin square
conditions called Balanced Latin Square that can control both
order and carryover effects.

Carryover effect
Balance Latin Square
 The effect of some treatments will persist, or
carryover, after the treatments are removed. Each treatment condition
 Carryover effect can sabotage any experiments.
1. Appears only once in each position in the other
sequence and
Note: 2. Precedes and follows every other condition an
equal number of times.
 We do not want to give subjects treatment that
will give them clues to what they should do in For example, consider the following balanced Latin
later conditions. square for an experimental design involving four
 We do not want the effects of early conditions treatments (A, B, C, and D) and four people:
to contaminate later conditions

Example: Studying Emotions

Choosing Among Counterbalancing Procedures

Difference between Order and Carryover effect
Every experiment with a within-subjects condition will
need some form of counterbalancing.
 • In within-subject design with one independent the order treatment produced significant
variable or a within-subjects factorial design, effects, you will know your experiment is
you must counterbalance all conditions. If the confounded by order.
design is a within-subject factorial, remember
to multiply the level of each factor together to  When one condition carries over more than
get the total number of conditions. the others, control is extremely difficult, if not
• In a mixed design, only within subject factors impossible. In such situations, an experimenter
needs to be counterbalance. should reconsider the design of the experiment
• We need to counterbalance for each subject and switch to a between-subject design if
when we expect large difference in the pattern possible.
of progressive error from subject to subject.
• When we know the effects will be about the
same for everyone, and no progressive error HOW CAN YOU CHOOSE A DESIGN?
within each subject then we can counterbalance How do you decide whether to use a within-subjects or
across subjects. a between-subjects design?
• Avoid using randomized and Latin square
 Always think about the hypothesis of the
counterbalancing if you expect carryover effect.
experiment.
When in doubt counterbalance subject by
 How many treatment conditions you need to
subject if you can.
test the hypothesis?
• Counterbalance can be use in between subject  Would it be possible to have each subject in
design experiment. more than one of this condition?

If so, you might be able to use a within-subject design.

Order as a Design Factor
 Does your treatment condition interfere with
If you are concerned that a partial counterbalancing one another? Yes? Then you might want to use
technique might now be controlling adequately for a between-subject design.
progressive error or carryover effects, there is a way to
You must consider the practical advantage of each
test whether treatments are producing similar effects
approach.
in all order sequence.
 is it simpler to run the experiment one way or
the other?
Treatment order as an additional factor in the design.  Which will be more time-consuming?
 If you can get only few subjects, the within
Example: subject design might be better.
Suppose you want to conduct the happy (H) versus sad  You can control subject variables better in a
(S) film clip experiment. You might be concerned that within-subject design.
one type of film will produce more carryover than  If there are likely to be large individual
another. It would be good idea to include order as a differences in the way subjects respond to the
factor in your design. With only two sequences: half experiment the within-subject approach is
the subjects receive the sequence HS and the other will better.
receive SH. Your experiment would be statistically Remember to view the research literature if the
analyzed as 2x2 (order x Film) mixed factorial design. experimenter used within subject design then it is
probably because within-design is the better option.

 If the order factor produced no significant

effects, you can feel more confident that your
counterbalance procedure work. However, if