Comparison Procedures

1
Comparison Procedures

• Two-Factor,Completely Between-Subjects Design

2
Exercise

• 8.1 A researcher tests the influence of
  different dosages of a drug on the detection of
  targets flashed briefly in the visual field.
  Target location is also manipulated For some
  participants targets flash on the periphery of
  the visual field whereas, for other participants,
  targets flash toward the center of the visual
  field. Scores represent the number of targets
  identified in 10 trials. 18 participants are
  assigned randomly to conditions.

In our last class, we tested the interaction of
dosage and location and rejected the null
hypothesis at ? .05. (a) If warranted, perform
appropriate tests of simple effects. (?
.05) (b) If further warranted, perform
appropriate post hoc pairwise comparisons within
simple effects (? .05).
3
Mean Alertness Scores as a Function of Target
Location and Drug Dosage
Collapsed
2.33
4.00
4
Summary Table
Reject H0.
5
Simple Effects

• A simple main effect is a treatment effect
  examined within a particular context. Analyzing
  simple main effects is a way of further assessing
  significant interaction effects.

Factor B
B1
B2
B3
A1
Simple effect of B at A1
Factor A
A2
Simple effect of B at A2
Simple effect of A at B1
Simple effect of A at B2
Simple effect of A at B3
6
Simple Main Effect Hypotheses
Examining the influence of dosage within the
peripheral-target condition.
Examining the influence of dosage within the
central-target condition.
7
Preliminary Calculations
8
Hypothesis Tests
Simple main effect of dosage within the
peripheral-target condition.
Simple main effect of dosage within the
central-target condition.
9
Tukeys HSD post hoc test
Preliminary Calculations
Hypothesis Tests
Reject H0. When targets are presented on the
periphery of vision, the high dose impairs
detection relative to the low dose.
10
Conclusions

• We have used a post hoc strategy to analyze data
  from the drug study.
• Perform the overall ANOVA first.
• If main effects are significant, perform Tukeys
  procedure, if the research problem warrants.
• If interactions are significant, analyze simple
  main effects. Where warranted, conduct Tukeys
  procedure within simple main effects.
• Which simple main effects do you test? B at
  levels of A? A at levels of B? Both? This is
  determined by the research problem and alpha.
• If the post hoc strategy is to be used to analyze
  main effects, the approach is straightforward
• Treat the main effect as a one-way ANOVA and
  perform the Tukey test comparisons accordingly
  (see worked out example on web site)
• The a priori testing strategy in the two-factor,
  between-subjects design is similar to the one
  used in the one-factor, between subjects design.

11
Exercise

• 8.2 A study is conducted in which 24
  participants consume a soft drink and then take
  part in a simulated car driving task. A
  performance score out of 10 is awarded, with
  higher scores representing better driving. Soft
  drink contents vary. Presence vs. absence of
  caffeine is combined with presence vs. absence of
  sugar in a factorial design. Participants are
  randomly assigned to the four conditions. Data
  from the study are shown here

Test the prediction that mean performance in the
condition that contains both sugar and caffeine
are present differs from mean performance in the
other three conditions combined. (? .05)
12
Driving Performance of a Function of Caffeine and
Sugar Levels
13
Preliminary Calculations
14
Hypothesis Test
15
Conclusions

• We have just used the a priori comparison
  strategy.
• If the strategy must be performed on a main
  effect, the analysis is straightforward.
• Treat the main effect as a one-way ANOVA and
  perform the comparisons accordingly.
• See web site for a worked out example