The growing difficulties in telephone surveys should lead researchers to examine different ways to overcome these problems. Researchers could find alternate modes of surveys to employ, or to alter techniques in telephone surveys. Adding cell phones to survey samples does some of both. However, the prevalence and potential for the coverage error incurred by not including these depends on estimates that likely to come from surveys themselves. If the measure of prevalence also contains error, then estimates of coverage error become difficult to interpret and inaccurate. The Hui-Walter method for estimating the true prevalence and error rates (false positives and negatives) when two indicators but no gold standard is available provides a possible tool to assess measurement error in surveys. There are a number of assumptions placed on the data, which possibly may be violated in survey data. Still, the tests may provide an insight into which questions may be most fl awed. It may still be necessary, however, to examine additional variables that are measured in the same survey context that may give clues which questions were the most fl awed. This study continues initial research in the potential coverage error of cell phones and measurement error generally in surveys. More research should focus on accurate measurements of the cell only population, and the potential it has to bias results by not including it in survey samples. More work on the efficacy of methods such as the Hui-Walter and other latent class models to estimate measurement error in surveys is also worthwhile. It is possible that such assumptions made the models are untenable in a survey setting, and thus are inappropriate for this application. The potential benefits seem clear, however, and if barriers do exist, researchers should develop techniques to overcome them.