The double bounded dichotomous choice format has been proven to improve efficiency in contingent valuation models. However, this format has been criticized due to lack of behavioral and statistical consistencies between the first and the second responses. In this study a split sampling methodology was used to determine whether allowing respondents to express uncertainty in the follow-up question would alleviate such inconsistencies. Results indicate that allowing respondents to express uncertainty in the follow-up question was effective at reducing both types of inconsistencies while efficiency gain is maintained.