Académique Documents
Professionnel Documents
Culture Documents
Categorization of SEM
Major components include:
Path analysis Confirmatory factor analysis
Categorization of SEM
When SEM is useful:
When you need to deal with latent (unobserved) constructs. When you have a strong theoretical background to your data (a priori hypothesis). When you are examining complex relationships. When you have access to a large sample.
SEM, on the other hand, is able to estimate the multiple and interrelated dependence relationships simultaneously. Because it tests the model as a whole, rather than in a piecemeal fashion, statistics can be calculated to show the goodness-of-fit of the data to the hypothesized model.
For example, a researcher wants to investigate the pattern of relationships between three psychological constructs: aggression, authoritarianism, and intelligence All three constructs cannot be directly observed and, therefore, cannot be directly measured.
They are measured indirectly using various types of scale items, from questionnaires and inventories.
Business/ Commerce
Application of structural equation modeling to evaluate the Intention of shippers to use Internet services in liner shipping. Human Capital and SME Internationalization: A Structural Equation Modeling Study
Health/Medicine
Application of Structural Equation Modeling to Health Outcomes Research Structural equation modeling of sleep apnea, inflammation, and metabolic dysfunction in children
Like with path analysis, it can be helpful to draw hypothesized relations in a diagram.
Most commonly used computer programs, such as LISREL (SSI, Lincolnwood, IL), AMOS (SPSS, Chicago, IL), EQS (Multivariate Software, Encino, CA), and Mplus (Muthn & Muthn, Los Angeles, CA), accept these diagrams as input.
CFA diagram
Error
Path
Path
Constructing diagrams
3 types of diagram symbols used in SEM:
Rectangles: observed variables (endogenous AND exogenous); Circles : disturbance, or error terms; Ovals : latent variables.
Constructing diagrams
Linking the symbols Direction of arrows between symbols are important:
for the analyses as a reflection of the underlying theory of latent variables, CFA, and SEM in general.
Path
Path
Error variances
Path coefficients
Path coefficient is equivalent to the factor loadings in FA. Therefore, this is a regression value.
Standardized coefficients range: -1 to 1 > value = stronger association
JobSat
USE
IT-efficacy
experience
review of related literature may support you to propose the following RQs
1. Does instructors IT-efficacy influence their ICTuse? 2. Does experience determine their use of ICT? 3. Does IT-use affect instructors job satisfaction? 4. Do instructors IT-efficacy and experience indirectly influence their job satisfaction?
e1
1
jobsat
Q3
USE
Q2 1 Q1
e2 EFFICACY EXPERIEN
Q5
If you have (i) the data (SPSS); (ii) AMOS software, then . . .
e1 jobsat
.38
USE
.43 -.28
e2 EFFICACY EXPERIEN
.01
Be Familiar With:
Path diagram Observed var Exogenous var Endogenous var Para estimates Path coefficients Covariances Residual/error Path analysis
Chi-Square
qol1
.18
e1 e2
.17
qol2 qol3
.64
e3 e4
.62
QLO
.79 .79
e5 e6 e7
.14
Measurement Invariance
.34 .37
qol7 qol8
e8
RESULTS
HOW TO DO IT
Model specification Model estimation Assessment of model adequacy
NEXT
Model revision, if necessary Comparing competing models
EXAMPLE
The purpose was to advance this theory development by determining the model fit of the Student Online Academic Persistence model.
53
RESEARCH QUESTIONS
1.
What paths provide the best fit for the model: Student Online Academic Persistence? What are the structural effects of the model: Student Online Academic Persistence? How well does the model fit the theory set forth: Student Online Academic Persistence?
2.
3.
54
Persistence persist leave Context Satisfaction with institution Satisfaction with faculty
55
METHOD
A structural equation modeling (SEM) analysis of student persistence was undertaken using the Mplus statistical program, Version 5.21 SEM was selected as a statistical methodology due to several advantages over regression modeling
56
PROCEDURE
Structural Equation Modeling Three Step Process 1. Path Analysis- analysis of variables 2. Confirmatory Factor Analysis- number of factors and corresponding indicators are specified 3. Structural Regression Model- Synthesis of #1 and #2.
57
Findings
58
GOODNESS OF FIT
Recommendations are to use at least 4 Goodness of Fit tests: Chi-Square Test of Model Fit Chi-Square (df= 31) Significance level: p<= .001 CFI- Comparative fit index RMSEA- Root mean square error of approximation SRMR- Standardized root mean square residual
59
0.74***
0.88***
0.35*** 0.26***
Motivation
0.88***
R2=0.76 0.54***
0.24**
R2=0.35
0.65***
0.93***
Decision making
-0.01 0.81*** 0.90*** 0.89***
0.59***
Persistence
0.72*** 0.83***
0.11
0.94***
Context
0.71*** 0.54***
cba1
0.34***
cba2
0.20***
cba3
0.21***
pers1
0.49***
pers2
0.32***
Model fit: Chi-square=99.20 (31);p<=.0001 CFI=.967 RMSEA=.056 90% CI=(.044,.068) SRMR=.035 *p<.05; **p<.01; ***p<.001
60
RESEARCH QUESTION #1
What paths provide the best fit for the model: Student Online Academic Persistence? The best fit resulted in the model when Extrinsic Goal Motivation and Learned Resourcefulness were removed
61
RESEARCH QUESTION #2
What are the structural effects of the model: Student Online Academic Persistence? Motivation explains most of the variance Motivation predicts Decision Making Decision Making predicts Persistence Context does not predict Decision Making
62
RESEARCH QUESTION #3
How well does the model fit the theory set forth: Student Online Academic Persistence? There was a good fit of the resultant model.
Chi-Square= 99.2 (df= 31) Significance level: p<= .001 (Close to 1.0) (Comparative fit index) CFI= .967 RMSEA= .056 (< .06) (Root mean square error of approximation) SRMR= .035 (< than .05) (Standardized root mean square
63
residual)
Implications
64
MOTIVATION IS IMPORTANT
Motivation (latent variable) explains most of the variance Factors: Goal Commitment Technology Self Efficacy Intrinsic Goal Orientation
65
MOTIVATION PREDICTS DECISION MAKING Students who report high levels of motivation also report the benefits in continuing education out-weigh the costs.
66
Students who report the benefits of continuing education out-weigh the costs are more likely to intend to persist in educational goals.
67
There was no relationship between reported satisfaction with the institution or the faculty and the students report of decision making.
68
RECOMMENDATIONS
Understanding and supporting the adult student in commitment to educational goal, technology self efficacy and intrinsic goal orientation could assist the student in program retention.
69
RECOMMENDATIONS
Providing counseling to adult students along the educational process by formalizing the continual process of cost-benefit appraisal may help to support the student and assist in clarification of the impact of this thought process on goal attainment.
70
71
72