SLIDE 1
Network Models I
- Latent factors reflect disease model
– Single underlying cause
- Requires conditional independence
– Symptoms/behaviors (likely) causally-related
- Network models provide an alternative
Applications: Network Models Network Models I Latent factors - - PowerPoint PPT Presentation
Applications: Network Models Network Models I Latent factors - - PowerPoint PPT Presentation
Applications: Network Models Network Models I Latent factors reflect disease model Single underlying cause Requires conditional independence Symptoms/behaviors (likely) causally-related Network models provide an alternative
SLIDE 2
SLIDE 3
Network Models II
- Contemporaneous and directed models are
estimated separately
- Directed networks are estimated by running
k univariate VARs for k variables.
- Model imprecision is not taken into account
- No model fit
- Almost no idiographic work
SLIDE 4
Integrated Network Model
- Utilizes SEM framework
– Automatic search procedure via Lagrange multiplier test (modification indices) – Bottom-up, data-driven model construction
- Provides
– Unconditional correlations (time t) – Directional, time-lagged relationships – Residual correlations (time t+1)
SLIDE 5
To what degree have you:
- 1. Felt down or depressed
- 2. Felt hopeless
- 3. Experienced loss of
interest or pleasure
- 4. Felt worthless or guilty
- 5. Felt worried
- 6. Felt restless
- 7. Felt irritable
- 8. Felt angry
- 9. Felt afraid
- 10. Had difficulty
concentrating
- 11. Experienced muscle
tension
- 12. Felt fatigued
- 13. Avoided people
- 14. Avoided activities
- 15. Procrastinated
- 16. Sought reassurance
- 17. Dwelled on the past
- 18. Felt positive
- 19. Felt content
- 20. Felt enthusiastic
- 21. Felt energetic
SLIDE 6
P025 Unconditional Model
SLIDE 7
P025 Directed Model
SLIDE 8
P025 Residual Model
SLIDE 9
P025 Integrated Model
SLIDE 10
Expected Force
- Centrality measures are poorly equipped for
flow characteristics of weighted networks
(Borgatti, 2005)
– Not designed to quantify spreading power – Underestimate influence of non-hub nodes
- Expected Force quantifies the spreading power
- f each node in a network (Lawyer, 2015, Nature)
– Spreading power is determined by the influence
- f the node and the influence of its neighbors
– Provides a normal, continuous metric
SLIDE 11
Expected Force
Uncondi4onal ¡ Directed ¡ Integrated ¡
SLIDE 12
InStrength
0.4 ¡ 0.5 ¡ 0.6 ¡ 0.7 ¡ 0.8 ¡ 0.9 ¡ 1 ¡
SLIDE 13
P068 Directed Model
SLIDE 14
P111 Directed Model
SLIDE 15
P072 Directed Model
SLIDE 16
P048 Directed Model
SLIDE 17
P014 Directed Model
SLIDE 18
P023 Directed Model
SLIDE 19
P075 Directed Model
SLIDE 20