The resulting similarity between an outlet and its audience can run in both directions. Congenial content may reinforce attitudes, while strong prior attitudes also make the outlet more attractive. Friends and platform ranking add exposure that is not fully voluntary.

A panel can show whether a source choice precedes later change under everyday conditions. An experiment isolates assigned exposure more cleanly but does not reproduce the same act of choosing.

The central causal problem runs in both directions: congenial media may reinforce attitudes, but established attitudes also guide source choice. A cross-sectional study-a comparison made at one point in time-cannot distinguish those paths. A panel, which reinterviews the same people, can test whether a source change precedes an attitude change. An experiment can assign exposure, though it may not reproduce everyday choice and conversation.

A one-time survey of strong partisans cannot solve the direction problem. They may watch more partisan news because of existing attitudes, or the source may intensify those attitudes. A panel showing that respondents changed their media diet before moving politically makes reinforcement more plausible; random assignment of sources isolates exposure more cleanly. Neither design implies that every viewer reacts alike.

Selective exposure is the individual's tendency to choose congenial information. Algorithmic amplification is the platform's ranking of material, and an echo chamber describes an information environment with little meaningful disagreement. A person can encounter congenial material accidentally, and a selective user can still see opposing views. A one-time correlation cannot reveal which direction causation runs. Ask whether selection occurred before the measured outcome and whether prior attitudes were controlled. A well-designed panel or experiment can separate the competing explanations more clearly, although each design answers a somewhat different question.

Video lesson: Media Institution: Crash Course Government and Politics #44