Advanced Multilevel Analysis - Dr. Gilad Chen

What is a key reason to use Random Coefficient Modeling (RCM) instead of OLS regression when analyzing nested data?

OLS yields biased and inefficient estimates in nested data
RCM provides larger effect sizes
RCM is easier to interpret than OLS
OLS can only be used with categorical data

Which of the following is not one of the four types of multilevel effects?

Fixed Random Effect
Level 1 Direct Effect
Level 2 Direct Effect
Cross-Level Interaction Effect

Which centering strategy helps distinguish between cross-level and between-group interaction effects?

Group-mean centering
No centering
Raw scores
Z-scoring the Level 1 variable

According to the slides, what is a major limitation of using slope variability tests (e.g., χ²) in small samples?

They often lack statistical power
They lead to spurious interactions
They inflate the ICC
They assume equal group sizes

Which of the following lists the correct sequence of steps when analyzing multilevel data using Random Coefficient Modeling (RCM)?

Testing intercept variability → Predicting intercept variability → Testing slope variability → Predicting slope variability
Testing intercept variability → Testing slope variability → Predicting intercept variability → Predicting slope variability
Predicting slope variability → Testing intercept variability → Testing slope variability → Predicting intercept variability
Testing slope variability → Predicting slope variability → Testing intercept variability → Predicting intercept variability

What are the 2 main types of time-varying RCM studies?

ESM focuses on repeated measurement of both IV and DV, while growth modeling focuses on repeated DV only
ESM focuses on between-person differences, while growth modeling focuses on within-person fluctuations.
Growth modeling uses only Level 1 variables, while ESM uses Level 2 predictors
ESM cannot model cross-level effects.

Which of the following is NOT a core statistical assumption should you examine when analyzing time-varying data?

Normality of predictors
Non-independence of observations
Heteroscedasticity of error variance
Autocorrelation of errors

Which of the following is the correct first step when analyzing growth in a Random Coefficient Model (RCM)?

Estimate intercept variability
Add predictors of growth trajectories
Test for autocorrelation in residuals
Assess slope variability

What is the proper level of analysis for examining individual growth as predictor of individual outcomes?

Level 2
Level 1
Both
Neither

Which of the following is NOT one of the five main steps in analyzing growth in RCM?

Run a confirmatory factor analysis
Test error structure
Estimate intercept variability
Assess slope (growth) variability

What are two key benefits of examining 3-level models in Random Coefficient Modeling (RCM)?

To control for higher-level effects and to incorporate influences from a third level of analysis
To reduce the number of predictors needed for model convergence
To eliminate multicollinearity among predictors across levels
To ensure Level 1 predictors have fixed effects only

In a 2-1-1 moderated mediation model, the moderator (Z) is typically located at which level?

Level 1 or Level 2
Only Level 1
Only Level 2
Z must be at the same level as Y

What is a key assumption when testing multilevel mediation models?

The mediator is at the same or lower level than the predictor
The moderator must be controlled using a covariate
The outcome must be at the highest level
All variables must be grand-mean centered

What is the purpose of group-mean centering Level 1 predictors in Mplus models?

To remove between-group variance from the Level 1 predictor
To simplify syntax in multilevel modeling
To combine within and between effects
To transform predictors into categorical variables

According to Mathieu & Chen (2011), what is one of the key future challenges for multilevel research?

Addressing the nesting assumption
Increasing degrees of freedom
Replacing random intercepts with fixed effects
Removing cross-level interactions

In moderated mediation, the indirect effect of X on Y via M is said to be conditional because:

It depends on the value of the moderator Z
It always varies randomly across individuals
It eliminates all direct effects
It uses bootstrapping for estimation

When employees work in multiple teams, researchers should account for:

Complex nesting and partial membership
Only the team with the most hours worked
Random assignment to a single team
Aggregated empowerment scores only

In Chen et al. (2007), the 3-level model included random intercepts for:

Level 2 (teams) and Level 3 (stores)
Only Level 1 (individuals)
Only Level 2 (dyads)
Cross-level mediation pathways

In multilevel moderated-mediation models, at which level(s) can the predictor (X), mediator (M), moderator (Z), and outcome (Y) be located?

X, M, and Z can be at Level 1 or 2, but Y must be at Level 1
All variables must be at Level 2 for moderated mediation
Only M and Z can be at Level 1; X and Y must be at Level 2
X and M must be at the same level, while Z and Y must be at Level 3

According to Mathieu and Chen (2011), what is “the unit problem” in multilevel research?

Difficulty in determining the appropriate level of analysis for a construct
Inconsistent units of measurement across datasets
Using too many units of analysis within one model
Lack of agreement on naming conventions for hierarchical levels

Name	You must fill out this field.
Email	You must fill out this field.

Advanced Multilevel Analysis – Dr. Gilad Chen

Quiz Summary

Information

Results

Results

Categories

1. Question

2. Question

3. Question

4. Question

5. Question

6. Question

7. Question

8. Question

9. Question

10. Question

11. Question

12. Question

13. Question

14. Question

15. Question

16. Question

17. Question

18. Question

19. Question

20. Question