Studies of new political parties often attempt to explain their electoral s
uccess. Methodological problems linked to the notion of selection bias, how
ever, considerably hamper the results of such studies. This note offers a d
iscussion of these biases, which are due to the fact that new political par
ties are a 'self-selected' sample. An empirical example studying the impact
of rules regulating the access to ballots in national elections in 21 west
ern democracies illustrates this problem. Taking into account the formation
process when studying the success of new parties, however, allows for a co
rrection. Similar solutions exist for other research designs employed to st
udy the success of new parties.