A test person was shown different lines on a card: besides a reference line there were three other lines. The test person had the exercise of guessing which of the three lines had the same length as the reference line. It was quite clear which line was just as long as the reference line (see the picture below). There was a person leading the experiment, a test person and other people, who'd already initiated the experiment.
In the control group, the initiated group (all except the test person) should guess (truely) which line is the same length as the reference line. As expected, the test person hardly makes any mistakes in his own assessment (less than 1%).
In the experiment group, the initiators should unanimously guess wrongly. The test persons adapted their guess to the majority in about one third of the cases, despite it being an obviously wrong decision. Only a quarter of the test persons remained uninfluenced.