Table 1. Cox Regression Coefficients.
| Covariate | Coefficient | exp(coefficient) | Standard error | Z-statistic | P-value |
| Experiment 1 | |||||
| Squared complexity | 0.052 | 1.05 | 0.026 | 1.969 | 0.049 |
| Experiment 2 - Non-transitional model | |||||
| Complexity | −0.216 | 0.805 | 0.094 | −2.29 | 0.022 |
| Squared complexity | 0.269 | 1.308 | 0.109 | 2.47 | 0.013 |
| Trial number | 0.029 | 1.030 | 0.007 | 3.99 | 6.5.10–5 |
| Model uncertainty | 0.261 | 1.298 | 0.174 | 1.50 | 0.13 |
| Experiment 2 - Transitional model | |||||
| Squared complexity | 0.356 | 1.43 | 0.084 | 4.27 | 1.9.10–5 |
| Trial number | 0.027 | 1.03 | 0.007 | 3.65 | 2.7.10–4 |
| First appearance | 0.500 | 1.64 | 0.272 | 1.82 | 0.069 |
All variables found in Experiments 1 and 2 that were added by the stepwise procedure. Note that some non-significant variables are added because the stepwise comparison is based on the Akaike information criterion [40]. These results reveal significant quadratic effects of complexity in both experiments. Complexity and squared complexity were shifted and scaled to have mean of 
and standard deviation of 
before being entered into the regression.