Survey researchers must now decide which data collection device or mix of devices is optimal for their Web survey (e.g., whether they will permit computers only or permit smartphones only). Their choice has implications for who can be observed and who cannot be observed. Yet there has been little research about the population subgroups that might be absent from different of types of Web surveys. This chapter takes a step in the direction of such research by exploring demographic differences across various subgroups of Internet users using data from a national telephone survey conducted in the US. Four overlapping groups are considered: the general population; those who go online using a computer; those who go online using a phone; and those who go online using mostly a phone, as opposed to other devices. In a novel approach to simplify the study of these groups, the process of using the Internet is modeled as a series of three transitions from one group to the next. This analysis sheds light on whether the effects of demographic characteristics are the same for each transition to a different level of Internet use. I also explore differences between these Internet use subgroups with respect to non-demographic survey variables after controlling for demographic differences.