Table 5. Best-fitting predicted models.
∆i = difference in AICc from best-fitting model; ωi = Akaike weight. See text and Table 3 for model specifications. All models with ∆i<4 are listed. ∆i for best-fitting models is always 0 and so not shown. For four measures (self-enhancement, categorisation, horizon ratio and additional objects), the global model did not fit the data so model comparison was not possible. See S1 Table for full model comparison statistics.
| Measure | Best fitting models (where ∆i < 4) |
|---|---|
| Individualism | HORIZONTAL (ωi = 0.80) |
| Collectivism | CULTURALGROUP (ωi = 0.58), GLOBAL (ωi = 0.25, ∆i = 1.70), VERTICAL (ωi = 0.16, ∆i = 2.57) |
| Closeness | PARENTSBIRTH (ωi = 0.47), VERTICAL (ωi = 0.27, ∆i = 1.08), CULTURALGROUP (ωi = 0.18, ∆i = 1.90) |
| Self-enhancement | Poor global model fit |
| Categorisation | Poor global model fit |
| Dispositional attribution | COUNTRYBIRTH (ωi = 0.41), CULTURALGROUP (ωi = 0.23, ∆i = 1.20), VERTICAL (ωi = 0.16, ∆i = 1.90), PARENTSBIRTH (ωi = 0.12, ∆i = 2.51), HORIZONTAL (ωi = 0.06, ∆i = 3.86) |
| Situational attribution | CULTURALGROUP (ωi = 0.53), PARENTSBIRTH (ωi = 0.43, ∆i = 0.40) |
| Drawing task: Horizon ratio | Poor global model fit |
| Drawing task: Objects | Poor global model fit |