Individual stability and aggregate stability are different. A population percentage can stay unchanged while individuals switch in opposite directions, and an aggregate shift can occur because the population's composition changes even if most individuals do not.

Repeated cross-sections estimate the population at several times. Panels reinterview the same people and reveal individual movement, but attrition and repeated-interview effects can threaten the result. Equivalent wording and consistent sampling are essential in either design.

Suppose support for a bond measure is 52 percent in spring and 52 percent in autumn. Two fresh samples show aggregate stability, not necessarily stable citizens. A panel might reveal that some earlier supporters became opponents after learning the cost while an equal number of opponents became supporters after seeing the projects. The same headline percentage can conceal considerable movement.

Video lesson: Public Opinion: Crash Course Government and Politics #33