In order to correctly gauge the opinions of the GMU community on the issue of whether or not the University should have an NCAA football program, the GMU Football Task Force decided to survey a representative sample of each population that would be affected by such a decision. These populations included students, faculty, staff, alumni, and Patriot Club members. Because of limitations of time and funding, it was initially decided to survey the first three groups using the WEB and the last two by telephone. However, because of problems with the WEB survey that will be discussed, students were also surveyed by telephone. In any survey of individuals there will be people who either cannot be contacted or refuse to participate. Such individuals are called nonrespondents and they account for a great deal of error in surveys. In particular, individuals who refuse to participate when contacted most certainly feel differently about issues than respondents. In the football survey, refusals are likely to be much less interested in football and/or George Mason University than respondents. Therefore, inference to the full population must account for refusals being less interested in football than respondents. This nonresponse error can be much larger than sampling error, causing confidence intervals to become unusually wide. The best way to reduce these ranges is to minimize the nonresponse rate. The next best course is to conduct follow-up interviews with a sample of nonrespondents. The vast majority of surveys do not attempt follow-ups, and even so, the basic theory assumes full response among the follow-up sample. This leaves the analyst with little choice but to model nonresponse behavior. These models lead to weighting the data to adjust for nonresponse. This weighting is common practice among the major US Government Statistical Agencies and top level survey research firms, but is neglected in the vast majority of surveys conducted by other firms in the private sector. In discussing analysis of the football data, we exhibit results with and without nonresponse adjustment to show the effect. An answer to the question of whether or not GMU should have a football team is suggested by contrasting desires by willingness to provide financial support. This is important because such support must be fully provided by student fees, private contributions, and ticket revenue. Another problem common in surveys is the difference in what people have done and what they say they will do. This is illustrated with the football data by comparing current attendance at athletic events with planned occurrence at future football games.