WithinSubjects Design

1
Within-Subjects Design

• Psych 7
• February 22nd

2
Outline

• Assigning Participants to Conditions
• Between-Subjects Design (yesterday)
• Within-Subjects Design (today)
• Characteristics
• Advantages disadvantages
• Order effects
• Threats to internal validity
• Dealing with threats

3
Within-Subjects Design Example

• IV (2 levels)
• Talking on cell phone while driving
• Driving while under the influence
• DV
• Performance on driving task
• Results

4
Characteristics of Within-Subjects Designs

• Uses a single group of Ps and tests or observes
  each individual in all of the different groups
  being compared
• Often called a repeated-measures design

5
Advantages of Within-Subjects Designs

• Dont need as many Ps
• e.g., consider a study that has 3 conditions and
  the E wants to run 30 Ps per condition
• Eliminates individual differences/person
  confounds!!

6
Disadvantages of Within-Subjects Designs

• Two Categories
• Order Effects
• Carryover Effects
• Progressive Error
• Time-Related Threats

7
Disadvantages of Within-Subjects Designs Order
Effects

• Carryover Effects
• More specific kinds of carryover effects
• Interference Effects
• Contrast Effects

8
Disadvantages of Within-Subjects Designs Order
Effects, cont.

• Carryover Effects
• Interference Effects
• Contrast Effects

9
Disadvantages of Within-Subjects Designs, cont.

• Please provide a brief, one word answer to each
  of the following questions.
• Try to answer each question as quickly as
  possible

10
Disadvantages of Within-Subjects Designs Order
Effects, cont.

• Carryover Effects
• Interference Effects
• Contrast Effects

11
Contrast Effects (Strack, Martin Schwarz, 1988)

• Order A
• What is your general life satisfaction?
• What is your dating frequency?
• Order B
• What is your dating frequency?
• What is your general life satisfaction?

12
Disadvantages of Within-Subjects Designs Order
Effects, cont.

• Progressive Error
• Not related to specific treatment or treatment
  conditions
• More specific kinds of Progressive Error
• Practice Effects
• Progressive improvement in performance over time
• Fatigue Effects
• Progressive decline in performance over time

13
Time-Related Threats

• History
• Environmental events other than the treatment
  that occur between the first and last tx
  condition
• Maturation
• Any systematic changes in an individuals
  physiology or psychology occurring over time
• Instrumentation
• Changes in the measuring instrument over time
• Regression toward the mean
• The tendency for extreme scores to revert to the
  mean

14
Dealing with Threats and Effects

• Consider an experiment with 4 treatment
  conditions
• Ps go through condition 1 first followed by
  condition 2, then condition 3, then condition 4
• Condition 4 may be in trouble
• May be influenced by order effects and
  time-related threats!

15
Dealing with Threats and Effects, cont.

• Counterbalancing to the rescue!
• Goal is to use every possible order
  of treatments with an equal of Ps
  taking part in each sequence
• Many different kinds

16
Counterbalancing

• Types of Counterbalancing
• Complete counterbalancing
• Incomplete counterbalancing
• Reverse counterbalancing
• Partial counterbalancing
• Latin square

17
Counterbalancing, cont.

• Complete Counterbalancing
• 2 conditions
• AB BA
• 3 conditions
• ABC ACB BAC CAB CBA BCA

18
Counterbalancing, cont.

• Reverse Counterbalancing
• ABCDE EDCBA

19
Counterbalancing, cont.

• Partial Counterbalancing
• ABCDE CEDAB
• ACBDE CBDAE
• BCDAE EBCDA
• BDAEC EDBAC

20
Counterbalancing, cont.

• Latin Square
• 4 conditions ABCD
• Start with a 4x4 matrix
• List the letters ABCD in order. This line of
  letters forms the first row of the matrix. To
  create the next row, simply move the last letter
  in line to the beginning. This creates DABC for
  the second row. Continue moving the last letter
  to the beginning of the line to create each new
  row. The result is the following Latin Square

21
Latin Square

• A B C D
• D A B C
• C D A B
• B C D A

• By definition, a Latin square is a matrix of n
  elements (letters) where each element appears
  exactly once in each column and in each row

22
Dealing with Threats and Effects, cont.

• Counterbalancing doesnt erase the threats and
  effects, it just unconfounds them with particular
  treatment conditions
• One last problem
• Ps may have figured out your hypothesis
• Can you figure out what Ps are thinking?
• Structured debriefing

23
Structured Debriefing

• Determine what Ps thought the study was about
• Just askor provide some options

24
Structured Debriefing, cont.

• Did the experiment seem fun? Boring?
• Did anything seem confusing?
• What did you think of the feedback that you
  received?
• Did anything seem odd or unusual to you? If so,
  at what point did that part appear odd?
• How believable did this experiment seem to you?
• What, if anything, seemed unbelievable?

25
Within or Between? That is the question

• Within-subjects designs are great
• Some experiments are impossible to conduct using
  a within-subjects design
• Use between-subjects design to eliminate any
  sequence or carryover effects, and time-related
  effects

26
Tuesday

• Complex Designs
• After this lecture, can start on Homework 2
• Due Tuesday, March 6th
• Cozby, Chapter 10