I see. Seems small longitudinal study are easier to done. As having it go on for 20 years could mean a lot of patient drop out for whatever reasons.
This usually only becomes an issue if the study is being done on elderly people, as they may not live long enough to finish the entire study. Longitudinal studies are nice because they allow the researcher to keep watch and collect data over an extended period of time, which can lead to more accurate results. For example, if they are doing a study of children starting at the age of 10, their views/identity/interactions with the world may be quite different after they go through puberty. So, if a longitudinal study were to be done over the course of 10 years for example, the researchers can watch their behavior develop and grow, to see which changes actually occur and can take those into account when making a conclusion for their study.
There are also cases where they think a child is too young to understand something (such as being adopted) and once they are older they will be able to grasp this concept. The problem with these studies though is that they are usually expensive and hard to do, one of the reasons being that the people may never finish the study (as you hinted at), but they are quite useful when used.