Chapter 15, Structural Equation Models Video Solutions, Statistics for Marketing and Consumer Research

Problem 1

Using AMOS, estimate the following path model using EFS data, where beer consumed at home (c21311) is the dependent variable and purchase of sausages (c11251) and hot take-away meal eaten at home (cb1127) are the explanatory variables:

(Figure can't copy)
a. Compare parameter estimates with those of standard regression models
b. What is the correlation between the explanatory variables?
c. Check model identification. How many degrees of freedom are available?
d. Fix correlation to zero to create an extra degree of freedom and re-estimate the model
e. Doing the goodness-of-fit diagnostics suggest that the model fits the data well?

Lainey Roebuck

Numerade Educator

07:53

Problem 2

Open the Trust data-set and run the following confirmatory factor analysis using AMOS, where it is assumed that variables q27a to q27g are manifest variables which measure the same latent construct, risk aversion:
a. Estimate the above model (HINT: do not forget to fix the variance of the latent factors at 1 for identification)
b. Can the model be regarded as satisfactory according to goodness-of-fit indicators?
c. Maintain only three manifest indicators related to daily behaviors, driving, eating chicken and eating beef - do results improve? (HINT: set the regression weight 'eating chicken' to be one in order to have an overidentified model)
d. Try changing the fixed regression weights (HINT: fix 'eating beef' or 'driving' instead of 'eating chicken,' with appropriate values as they emerged from the initial analysis)

(Figure can't copy)

e. Compute the scores of the latent factor (HINT: see AMOS help on 'how to estimate a latent variable score'). How do they correlate with directly measured risk aversion (q28)?

Heather Duong

Numerade Educator

04:35

Problem 3

3. Consider the following theory - Chicken consumption (q4kilos) is the outcome of price (PRICE), income (INCLEVEL) and un-measurable attitudes as measured by $b b a, b b b, b b d$ and $b b e$.
a. Draw the structural equation model, considering correlations across variables
b. Using constraints and deleting non-significant elements, find an appropriate model
c. Compare the model with an alternative theory which explains chicken consumption (CONS) as the outcome price (PRICE) and habits, intended as a latent construct measured by $\mathrm{q} 2 \mathrm{~b}, \mathrm{q} 2 \mathrm{c}$ and $\mathrm{q} 2 \mathrm{~d}$. Which theory works best? Can either of the two theories be retained as valid?

Sophie Knight

Numerade Educator