Research to support the effectiveness of a minority comes from Moscovici et al (1969). Moscovici et al found in a laboratory experiment with female participants, that when there were two confederates being consistent when estimating the colour of slides, the real participants agreed with the minority on 8.42% of the trials (i.e.
they said the slides were green), whereas when the four confederates were inconsistent, only 1.25% of the participant’s answers were green.