SEMs In Experimental Research

1
SEMs In Experimental Research

• James G. Anderson, Ph.D.
• Purdue University

2
Rationale for the SEM Approach

• Means of discovering the existence of a causal
  relationship using a chi square test
• Estimation of the magnitude of the experimental
  effect
• Estimation of the measurement error of constructs
• Analysis of problems in the application of an
  experimental design

3
Experimental Method

• ANOVA
• The IVs are categorical
• Multiple regression analysis
• DV is continuous
• IVs and DV are both continuous and categorical
  (i.e., ANCOVA)
• Cell frequencies in a factorial design are
  unequal and disproportionate
• Studying trends in the data

4
ß x ----------------------------
-gt y Figure 7.1 A simple two-variable model
one independent variable x and one dependent
variable y
5
Problems With Traditional Experimental Assumptions

• Assumption that the IVs and DV are measured
  without error
• Variation in the DV may be due to other factors
  than the experimental manipulation

6
(No Transcript)
7
(No Transcript)
8
Marketing Example

• IV two levels of communication credibility
• Two manipulation checks (i.e., Likert Scale
  questions)
• DV two measures of attitude towards a product

9
(No Transcript)
10
(No Transcript)
11
(No Transcript)
12
(No Transcript)
13
Results

• ?2 0.30, df4, plt.99
• Residuals are small and nonsignificant
• The magnitude of the effect of the IV on the DV
  is 0.55

14
Diagnosing Problems in Experimental Research

• Used in cases where the hypothesis that the
  experimental manipulation produced the predicted
  effect on the DV is NOT confirmed
• Examine the effect of the experimental
  manipulation
• Examine the measurement error in the manipulation
  checks

15
Diagnosing Problems in Experimental Research (2)

• Examine the measurement error in the DV
• Test rival hypothesis

16
(No Transcript)
17
SEMs for More Complicated Experimental Designs
18
(No Transcript)
19
(No Transcript)
20
(No Transcript)
21
Quasi-Experimental Study of Physician Use of a HIS

• Setting Teaching hospital
• System HIS
• Experimental Design
• Experimental group 4 hospital services
• Control group 10 hospital services
• Experimental Manipulation Influential physicians

22
Analysis

• Post-test only design
• Pre-Post test design
• MANOVA

23
(No Transcript)
24
(No Transcript)
25
(No Transcript)
26
(No Transcript)
27
(No Transcript)
28
(No Transcript)
29
(No Transcript)
30
(No Transcript)
31
Results

• The experimental manipulation had only a small
  effect on physicians use of the HIS
• At the same time, the study results suggest that
  assisting physicians to develop personal order
  sets would result in a significant increase in
  the use of the HIS

32
(No Transcript)