CHAPTER 12. MODEL VALIDATION
1. What are three reaearch uses (purposes) of the multi-sample approach? p. 262
2. What is the "basic multisample model"? What is "case c"? p. 262
3. What are the steps for the basic multisample model in Amos? pp. 263-264. Warning: Stop at p. 264 as pp. 265-267 are
replaced in the current Amos version, which has a menu option for Analyze, Multiple Group Analysis. This will be discussed in class.
4. In the output for a multiple group analysis on pp. 267-268, are the regression weights (structural or path weights) the same
for both samples? Is the correlation of the two latents, Math and Verbal, the same? Can the measurement model be considered
the same for both samples? Explain.
5. What is cross-validation? What coefficients are commonly used when doing cross-validation? pp. 274-277
6. What is the bootstrap procedure, where is it in Amos, and how do you interpret the Amos output at the bottom of p. 297? pp. 294-298?
7. What is the jacknife procedure? pp. 310-311
CHAPTER 13: SEM APPLICATIONS. PART I.
1. What are MIMIC models? p. 324
2. In MIMIC models, what is the difference in the arrows associated with the indicator variables compared to the
observed variables (causes)? p. 324
3. Are goodness of fit statistics interpreted the same as regular SEM models in the case of MIMIC models?
4. Is multigroup analysis the same as multisample analysis discussed in Chapter 12?
5. What is the "critical ratios for differences" test in multigroup analysis? pp. 328-329
6. What are multilevel models? p. 330
7. What are structured means models? p. 348
CHAPTER 14: SEM APPLICATIONS. PART II.
1. What is an MTMM model and what is the main problem with this type of model? pp. 355, 359
2. How can one assess whether one has a common method variance effect using a correlated uniqueness
model? pp. 359-363.
3. What is a second-order factor model? pp. 364-366.
4. What is an interaction model? Bonus: can you explain the steps in the continuous variable approach? pp. 369-376
5. What do you do in you want to test for interaction effects when the interaction variable is categorical rather
than continuous? pp. 376-379
6. What is the latent variable approach to interaction effects about? How does it work in general? pp. 379-386.
7. When would two-stage least squares estimation of parameters be used? pp. 386-388
8. What is latent growth curve analysis? How does it work in general? pp. 390-394
CHAPTER 12: MODEL VALIDATION (MULTIPLE GROUP ANALYSIS IS SECTION 12.1)
1. Load Fig12.1 from the textbook disk, Chapter12 folder.
a. Extra credit: draw it manually! Or at least follow steps 2-6 to verify what has been done for you.
2. From the menu select File, Data Files; then in the Data Files dialog and load in the two separate .sav covariance matrix files
as input (sample1,sav and sample2.sav). Make sure they are both loaded under File, Data files, and appear on the left as
Group Number 1 and Group Number 2. Optionally, click on View, Data, to actually see the data, which is a covariance matrix.
3. Draw the path diagram in Amos (Fig. 12.1 on p. 263).
4. Note the regression weights of the error variance terms are specified.as 1 (right click the arrow and enter 1 in the Regression
Weight box under the Parameters tab).
5. Note the factor variances of Verbal and Math have been set to 1. 1 (right click the latent variable ellipse and enter 1 in the
Variance box under the Parameters tab).
6. Note the factor loadings have been labeled (in Object Properties) which tells Amos to constrain them the same across groups.
The book just says to use View, Matrix Representation. But when the Matrix Representation box appears, you must drag the
indicator and latent variables into the matrix from the left-hand column of symbols (see p. 264, bottom). The book says to label
you could also double-click the path. Actually you have to right-click, then enter the label (ex., "loading1") under the Parameters
tab, in the Regression Weight box.
7. Choose Analyze, Multiple Group Analysis from the menu; accept the defaults. Then Analyze, Calculate Estimates.
8. From the menu, select View, Text Output. If needed, click the options icon (the checkbox symbol), select the View tab, and
check "View entire output file". Verify the output is the same as pp. 267-268.
9. Click the print icon and print the output.
10. Click on the "View the output path diagram" icon (right under "Tools" in the menu). Select File, Print; check you
want standardized estimates for the default model; print the path diagram.
11. Annotate the print output manually and be prepared to discuss it in class, section by section.
HOMEWORK CHAPTER 14: SEM APPLICATIONS PART II (LATENT GROWTH CURVE ANALYSIS IS SECTION 14.4)
1. First, change the name of the data file to "chap14log.sav" (not chap14_log.sav as supplied by the text) so it will match up with
the .amw file.
2. Load Fig14.10.amw
3. Follow the directions on pp. 391-393 and run the model. Note the current Amos version menu uses View, not View/Set.
Note the constraints are already set for you, but right-click on any arrow and select Object Properties just to see where
these regression weights get set.
4. Annotate the print output manually and be prepared to discuss it in class, section by section.