Is there value in opinion polls?

Opinion polls published in recent months are giving results which are not consistent with each other. This makes people wonder whether there is any value in opinion polling. One needs to keep in mind that some of the polls published before the 2024 MEP elections did not give a correct prediction of the eventual result and, as such, the doubts that there may be about such polls may seem to be justified.

I would like to state at the outset that opinion polls have great value for a country’s democratic process. The purpose of an opinion poll is to capture the sentiment of the electorate at a moment in time. It cannot be seen as a prediction of the actual result of an election. If that were the case, we should not bother to hold elections but should just have an opinion poll.

In other countries, exit polls are conducted to provide an indication of the actual result. In Malta, it is still uncertain whether conducting an exit poll while voting is still going on would breach the law or not. Exit polls also have an inherent margin of error, like all surveys which make use of random probability sampling, and that margin of error can be quantified.

Another consideration to make is the methodology used by those conducting opinion polls. Has the poll been conducted through face-to-face interviews, through the telephone or online. Each of these three methodologies present their own pitfalls.

If an online poll is based on self-selection of the respondents, a bias is immediately evident. Moreover, various demographic segments may have different behaviour and attitudes towards the use of technology, which could result in biased results. Likewise, if a telephone poll were conducted using a high percentage of landlines as opposed to mobile numbers, younger voters are likely not to be adequately represented. Face-to-face interviews can have their own bias as well.

If an online poll is based on self-selection of the respondents, a bias is immediately evident- Lawrence Zammit

This is why the guidelines of ESOMAR, the World Association for Social, Opinion and Market Research, and WAPOR, the World Association for Public Opinion Research, are very specific on the type of information which needs to be provided when the results of opinion polls are published. These guidelines promote transparency to enable the readers of such polls to interpret them correctly.

These guidelines state that the information which needs to be made available should include the names of the organisation which conducted the poll and its sponsor (who paid for it); the universe effectively represented (who was interviewed) and whether certain groups were excluded from the design (those without landline telephones or internet access, for example); the actual sample size (number of completed interviews included in the reported findings) and the geographical coverage; the dates of fieldwork; and the sampling method used (probability or non-probability sampling).

The information should also include the response rate; the method by which the poll was conducted (face-to-face, telephone interview, internet access panel, mixed mode etc.); whether weighting was used to adjust the results and the general demographic or behavioural characteristics used for the weights; the percentages of respondents who give ‘don’t know’ answers, are undecided, refuse to answer the voting intentions question, or say they will not vote (the interpretation of particular findings will be quite different if the level of ‘don’t know’ answers is five per cent or 30 per cent); and the relevant questions asked.

It is within such a context that opinion polls need to be read and interpreted. The more transparency there is in the publication of the results, the more value opinion polls have. Opinion polls should not be viewed as a way of playing god by trying to predict the future but as a way of measuring public sentiment at a moment in time.

Lawrence Zammit is a founding partner and a director of Misco, a research and consultancy company.