According to the latest survey conducted by the Guardian/ICM via a phone poll says that the Conservatives are ahead by 6 points. While, the YouGov poll for the Sun via an internet survey says the Labour leads by 3 points.
Internet polls conducted last week says that the Labour’s lead ranges from 3 to 6 points. However, phone polls that were published several hours’ later say that the Tories are ahead by 1 point.
The differences of the poll results can be attributed and explained by the standard margin of error by random sampling of positive or negative 3 points that all polls have to face. But, the trends in the polls from internet and phone surveys tell us significant differences.
Which Method is Better?
The problem is that we won’t be sure enough until May 8th this year.
The phone polls conducted in 2010 were generally closer to the final results; however, it was mainly driven by the surveys conducted by the internet that overestimated the support for the Lib Dems.
Majority of the internet polls were generally correct in terms of figures for the 2 largest parties during the election. YouGov’s final results in 2010 had the Tories on 35%, while the Labour on 28% — which are just 1 point off from their final figures.
History tells us that phone polls are generally more accurate; however there is no evidence that suggest that this will always be the situation. The fact that only fewer people nowadays use the telephone will prove this historical basis to be a challenge to phone polls. Internet polling, on the other hand, is cheaper; thus, more data can be collected from various domains, and panels have become increasingly representative of a certain population as it grows in size.
The Difference between the 2 Approaches
Phone polls utilize random sampling; while those who take internet polls usually do it voluntarily as they have to sign up to a panel, but they cannot choose which poll to respond to.
There are also differences in methods used by phone and internet polls. This includes past voting behaviors, undecided voters, and levels of uncertainty.
Note that all types of polls have a margin of error and a level of confidence. For instance, for a poll of 100 people, there will be a marginal error of +/- 3 points and a confidence level of 95%. Theory suggests that 95% of the results of the poll are within the 3 percentage of points of what the results would be for the whole British population that participated in the survey.
The truth is there is always that element of uncertainty. That is, there is no way of knowing whether a certain poll is right or wrong until the election takes place. Because of that, it is highly recommended that you evaluate the trends of the polls first.
The latest average of the polls conducted by the Guardian has 2 main parties that are tied up; the conservatives who are on 33.7% which is just 0.1 points ahead of the party of Ed MIlibands.