How the polls are being reported
April 12, 2010
The prevalence of polling data presents a dilemma for news organisations – how to present it in a way their readers can best understand.
The Guardian’s handling of election polling data is a model of clarity and concision. It even includes information on the margin of error. The only additional feature I would like to see is a tool to translate the predicted share of the vote into Parliamentary seats. [Update – I hadn’t spotted this interactive Swingometer, which effectively does the same job]
The Telegraph’s version does incorporate such a feature, although the decision to mark the y-axis in multiples of seven is slightly baffling (presumably this is formatted on the fly). The Telegraph could have added a “winning post” to the seat forecast, though, to indicate how many seats the party needs to win in order to have a parliamentary majority.
The Times has combined its polling data coverage with an election map. While it lists more parties (separating out the minor parties rather than lumping them all together as “Other”), the interface is unnecessarily convoluted. Also, the poll data only seems to come from Ladbrokes.com, whereas the Guardian and the Telegraph present aggregated results. [Update – according to the Times site, data from Populus is also being used]
There is a benefit to using Ladbrokes’ data, because the company is offering odds on every constituency – this enables a more detailed picture than that provided by traditional pollsters, who only give the national picture.
Couple of curious points when I looked at my constituency of Chester, though. Tories are favourite to win it, with Ladbrokes giving them an 84% chance of success. And it seems reasonable that Labour have a 13% chance of holding on. But is the chance of a Green, English Democrat, UKIP or even LibDem victory really as high as 1%? I don’t think so. (Oh, and that all adds up to 101%).