This basis the winning party in an election can

this basis, the winning party in an election can reasonably claim that its policies most
closely correspond to the interests of the largest group of voters.
On the other hand, it can be argued that, rather than 'buying' policies, voters are
typically poorly informed about political issues and are influenced by a range of
'irrational' factors such as habit, social conditioning, the image of the parties, and the
personalities of their leaders. Moreover, the ability of parties to attract votes may

VOTING BEHAVIOUR 223
have less to do with the 'goods' they put up for purchase than with the way those
goods are 'sold' through advertising, political campaigning, propaganda and so on. To
the extent that this is true, election results may reflect not so much the interests of the
mass of voters as the resources and finances available to the competing parties.
A further, and some would argue more intractable, problem is that no elective
mechanism may be able reliably to give expression to the multifarious preferences of
voters. This is a problem that the US economist Kenneth Arrow described in terms of
his 'impossibility theorem'. In Social Choice and Individual Values (1951), Arrow
drew attention to the problem of' transitivity' that occurs when voters are allowed to
express a range of preferences for candidates or policy options rather than merely
cast a single vote. The drawback of casting but a single vote is not only that it is a
crude all-or-nothing device, but also that no single candidate or option may gain
majority support. For instance, candidate A may gain 40 per cent of the vote, candi-
date В 34 per cent, and candidate С 26 per cent. The situation could nevertheless
become more confused if second preferences were taken into account.
Let us assume, for the sake of argument, that the second preference of all candi-
date A's supporters go to candidate C, the second preferences of candidate В favour
candidate A, and the second preferences of candidate С go to candidate B. This
creates a situation in which each candidate can claim to be preferred by a majority of
voters. The first and second preferences for candidate A add up to 74 per cent (40 per
cent plus B's 34 per cent). Candidate В can claim 60 per cent support (34 per cent
plus C's 26 per cent), and candidate С can claim 66 per cent support (26 per cent plus
A's 40 per cent). This problem of'cyclical majorities' draws attention to the fact that it
may not be possible to establish a reliable link between individual preferences and
collective choices. In other words, election results cannot speak for themselves, and
politicians and political commentators who claim to find meaning in them are, to
some extent, acting arbitrarily. Nevertheless, the latitude that this allows politicians is
not unlimited, because they know that they will be called to account at the next
election. In this light, perhaps the most significant function of elections is to set
limits to arbitrary government by ensuring that politicians who claim to speak for the
public must ultimately be judged by the public.