Comparison of a new expert elicitation model with the Classical Model, equal weights and single experts, using a cross-validation technique