Concept
experiment (research method)
A method in which the researcher changes a condition and randomly assigns participants to groups to test the cause of a difference; strong internal validity, external validity to be questioned separately.
In an experiment the researcher changes a condition (for example the form of a news item), splits participants into groups by random assignment, and compares the outcome. Because the lottery makes the groups similar at the start, it is reasonable to attribute the difference to the condition: this is internal validity. The price is the problem of external validity, which depends on who was taken into the study and on whether the setting resembles real life.
A laboratory experiment increases control but makes the setting artificial; a field experiment is run in real life (such as the randomly distributed newspaper subscriptions of Gerber and colleagues). In communication, experiments run from the Payne Fund’s before-and-after measurements to Iyengar, Peters and Kinder’s television-news experiments, and on to the 689,003-person study on Facebook. See validity, correlation and causation.
Example
Giving households a newspaper subscription or nothing at random, then comparing political knowledge and turnout after the election.