When using online access panels for attitudinal research there are a number of reasons why we might expect highly divergent results from what we would obtain in a study using an RDD telephone survey. There are strong self-selection features in using an online panel and there are differences in how respondents rate items in a visually-based survey that is self-administered compared to an orally administered survey conducted by a human interviewer. A variety of authors have indicated that data obtained in different modes of interviewing are likely to be different, even if using the same sampling strategy from the same population. In a series of 4 studies, we compared results from numerous attitudinal measures in web-based studies using access panel respondents with results obtained from RDD phone interviews conducted in the U.S. The first 3 studies represented parallel administrations of the same surveys, while the fourth study presents the results of 9 years of measurement of presidential approval, and compared with the average ratings of telephone polls fielded at the same time. We found that results were highly comparable, most likely due to the large sample sizes and sufficient diversity within the samples. The major difference between modes appeared to be that the average ratings online were lower than those obtained by phone. We further explored the relationship between attitudinal measures and common criteria in both phone and online surveys and found slightly higher correspondence between predictors and criteria in web-based surveys.