Final answer:
In statistics, kappa values cannot be used to test statistical significance directly. Instead of relying on subjective thresholds, it is recommended to estimate the strength of evidence using concepts like AIC or BIC weights. The likelihood of alternative models should also be evaluated.
Step-by-step explanation:
In statistics, when using kappa values, one cannot test statistical significance directly. The traditional approach in statistics involves setting up a null model and evaluating whether the data conforms to this model. To determine statistical significance, researchers often use a threshold probability to conclude significance. However, this subjective approach has been criticized, and many have called for the abandonment of arbitrary Type I error probability thresholds. Instead, they recommend estimating the strength of evidence using concepts like AIC or BIC weights and evaluating likelihood of alternative models.