Who conducted the poll? Are they reputable and independent?
If the organization that carried out the poll is not independent of all political parties, then its findings may be biased. It is important to know who commissioned the poll - was it a political party, a media organization or some other body?
How many people were interviewed?
The size of the sample is obviously important. If it is too small then the result is less likely to be accurate. But if the sample is well selected, a relatively small number may still yield important results.
How were they chosen?
A reputable pollster will publish the sampling method. Clearly who is chosen is going to determine the result. If the poll is conducted by telephone it only reflects the views of telephone users. In some countries that will be a fair proportion of the population, while in others it will be a tiny minority. A proper sample will reflect the variety of geographical or population groups in the country.
Are the published results based upon the answers of all those interviewed?
An easy way of manipulating poll results is to leave out the answers that you do not like. That is unprofessional, for both a pollster and a journalist.
When was the poll conducted?
People's views change, especially during an election campaign. A poll that is a month old may not be a reliable guide to how people will vote tomorrow.
What is the sampling error?
This means simply, how far out are the poll results likely to be? No poll can predict the result with total accuracy, but a professional one should get within a few percentage points.
What questions were asked - and how were they worded? In what order were they asked?
Professional media reporting will reflect exactly what the questions were, to avoid the danger of 'interpreting' the results in the most politically convenient way. It is important to know what all the questions were, because answers may have been contradictory. For example, respondents may say that they feel that Candidate X has been doing a good job - but then express a view that is in disagreement with his major campaign platform. Most importantly of all, the public must know that the questions were asked in a balanced way, not leading respondents to give a particular expected answer.
How do the results of this poll compare with other findings?
If there are several polls carried out by reputable organizations, then averaging out the findings might help to reduce the overall margin of error. Alternatively, if the results of one poll are way out of line with others - a 'rogue' poll - it would be interesting to know why. Is it just a badly conducted poll? Or does it show some important change in public opinion?
