Significant challenges exist to obtaining the same answers to survey questions when asked by telephone vs. the Internet. In this presentation I will discuss results from a number of experiments aimed at obtaining equivalent data across these survey modes. One of the issues examined is the conversion of structures that is often made when going from telephone to web, ostensibly for the purpose of making them questions easier to ask on the web, e.g. changing forced-choice telephone questions to check-all web formats. I also report the results of multiple experiments done in a search for a format for asking scalar questions that will produce the same distributions for both modes. The difficulty of obtaining equivalent answers to such questions is traced to differences in the type of communication being used to present questions to respondents, aural for telephone and visual for the web. I argue that understanding effects of these two kinds of communication is essential for finding ways to achieve equivalent answers for a variety of questions commonly asked in telephone and web surveys.