Scholars have argued that framing effects are caused by assumptions about unstated information, such as assuming that the sure option is lower-bounded (e.g., assuming "at least" 200 will be saved; Mandel, 2013). We disambiguated missing information before presenting framing problems with either the traditional risky option (e.g., 1/3 chance 600 saved and 2/3 chance none saved) or a truncated risky option (e.g., 1/3 chance 600 saved). A manipulation check indicated that, following presentation of detailed instructions with worked examples, over 90% of subjects correctly interpreted missing information while completing framing problems. Contrary to the ambiguity hypothesis, subjects who did not mentally add "at least" to the sure option nonetheless showed a robust framing bias. Among the few subjects who showed ambiguity effects, framing and truncation effects still replicated. Although adding words can change interpretations of framing information, this manipulation does not account for the core mechanism of framing bias.