Research Article
A Weighted Statistical Network Modeling Approach to Product Competition Analysis
Table 4
Estimated coefficients of the 2013 choice network.
| Input variables | Model 1 | Model 2 | Model 3 |
| Network configuration effect | Sum/intercept | 6.19 | 6.11 | 5.77 | Cyclical weights | | −0.06 | | Transitive weights | | | 0.16 |
| Main effect (inward node attributes) | Import | 0.03 | 0.03 | 0.03 | Price (log2) | −1.05 | −1.04 | −1.03 | Power (log2) | 0.30 | 0.30 | 0.30 | Fuel consumption | 0.58 | 0.58 | 0.57 | Brand origin (USA) | 0.82 | 0.81 | 0.78 | Brand origin (Europe) | 0.15 | 0.13 | 0.15 | Brand origin (Japan) | 0.80 | 0.80 | 0.75 | Brand origin (Korean) | 0.56 | 0.56 | 0.54 |
| Homophily effect (dyadic attributes) | Market segment matching | 0.68 | 0.69 | 0.67 | Brand origin matching | 0.99 | 1.00 | 0.98 | Akaike information criterion (AIC) | −68 209 | −68 211 | −68 252 | Bayesian information criterion (BIC) | −68 113 | −68 106 | −68 147 |
|
|
|