The news media also joined the bandwagon, with an increasing number of newspapers and television sources regularly using polls as a means of reporting on campaigns. From 1984 to 1988 alone, the number of news media polls on campaigns tripled (Taylor, 1990a, p. 261). Some argue that the release of such polls helps to construct public opinion, rather than just measure it. Generally, though, research indicates that polls can affect the perception of public opinion, but do not create a bandwagon effect that induces undecided voters to side with the probable winner (Daschmann, 2000). For that reason, political campaigns seldom release their data, even when they would report positive information for them. Instead, they prefer to use their polling resources to validate campaign messages and to discover the movement of public opinion on key campaign issues, information they would prefer that the opposition not have. When they do release information, it is highly selective, with information favorable only to their side typically released. As a result, those news outlets that wish to have polling data often gather their own data. The effect is that the news media regularly release their polling data showing who is leading and by how much, and political campaigns seldom offer theirs unless there is a vast and potentially destructive difference between their numbers and the media’s numbers.