Posted on

PETER WICKHAM: Vote buying and the margin of error

Peter W. Wickham

PETER WICKHAM: Vote buying and the margin of error

Social Share

This issue is an awkward one about which I have said very little for reasons that are worth discussing. 

As a pollster, there is an expectation that my opinions would be grounded in evidence and the issue of vote buying is one where evidence is notoriously difficult to collect since the only incontrovertible evidence would come from someone who either purchased or sold a vote. Such persons would presumably not want to confess since they would also effectively be agreeing to subject themselves to the criminal sanction associated with their act.

There is, however, another perspective with which I am more comfortable and this relates to the concept of the “margin of error” associated with any survey. Since political polls are designed to predict elections, then the extent to which the result varies from poll is to some extent a reflection of such a margin. This margin is a declaration which all pollsters make but is little understood by those who often believe that the statement is an admission that our methodology is imperfect. I have, therefore, often been told by clients and readers to “reduce” my margin and give them a perfect poll.

A perfect poll is impossible for reasons which are both intrinsic to polling methodology and external to our methods which are located within the broader reality of behavioural research. Intrinsically there are methodological flaws that are unavoidable with any sample since in simple terms we might interview the wrong person and get a skewed result. If one were to avoid this, then the research exercise would not be a survey but a census which is a suggestion that is not worthy of consideration. Sampling techniques attempt to interview a representative quantity but often pollsters do not reach the correct representatives and hence the margin explains the extent to which a poll result could deviate from the “truth”.


On the external side, behavioural research needs to navigate a set of considerations which are peculiar to our discipline, such as the possibility that people will change their mind after having spoken to a pollster. There are several reasons why this happens and one of them is obviously an inducement to vote. It is clear that such inducements are more effective in some areas of Barbados than others and it is also obvious that these inducements were considerably more effective in 2013 than in the past. Regarding specific locations, anecdotal evidence suggests that housing areas are more susceptible to such inducements largely because these are large groups of people who can easily be “corralled” by vote buyers or their agents.

There is also statistical evidence available by reference to the electoral swing comparison which demonstrates that residents of St Michael North West were comparatively more happy with the DLP’s representative there than any other DLP representative in the island since his personal swing was +11 per cent while the DLP’s was -1.5 per cent. This comparative happiness was, of course, not related to vote buying but was perhaps associated with the superior quality of the representation and the comparative quantity of residents who live in housing areas.

These arguments which attempt to “prove” vote buying are therefore futile and it makes more sense to focus on the solution which is actually quite simple if one attempts to narrow the problem to the most troublesome practices. This approach targets the practice on election day and ignores inducements during the election season since these are virtually impossible to isolate from the traditional campaign activity which deliberately associates candidates with the good deeds they do for voters. If the desire is to prevent or discourage the buying of votes on election day, it makes little sense increasing the penalty as proposed by three members of corporate Barbados, since problems with reporting remains. Instead we should exploit the role of the informant who should be incentivised and insulated from prosecution.

Raising the bar

Therefore if we think that people are offered $300 for a vote and if they report this are liable to be prosecuted, we need only to offer that person $600 to inform on an attempt to buy his/her vote with the promise that they would not be prosecuted. This effectively raises the bar for vote buying to above $600 since that is what either a candidate or agent would need to offer to ensure the informant wouldn’t be motivated to act against them. Once this logic is accepted then the reward need only be adjusted upward based on the market rate for a vote.

The fact that I would need to speak to this issue is a poor reflection on the place we are at in our development and this is sad. There is also a slightly more veiled assumption which underlies vote buying and this speaks to a level of electoral ignorance on the part of this population. It is clear that some people believe a politician who pays them can actually know who they voted for in a supposedly secret ballot, demonstrating that we have some serious educational work to do.

Peter W. Wickham is a political consultant and a director of Caribbean Development Research Services (CADRES). Email: [email protected]