PITY the pollsters. In the spam age of junk emails, unwanted flyers and chirpy street interviewers, the serious job of finding out how people rate their political masters is becoming much more difficult.
The Pew Research Center in Washington DC, which studies people鈥檚 attitudes to the media and politics, reckons response rates to US opinion polls have fallen from 36 per cent of households in 1997 to 27 per cent in 2003. And the more people refuse to reply, the harder it is to put together a representative sample of voters. Some pollsters have turned to the internet to solve this difficulty, but although internet polling has had some success, it has its problems too.
The irony is that voters worldwide are relying on opinion polls more than ever. In the UK, the percentage of voters who wait until the election campaign itself to decide how to cast their ballot has more than doubled since 1964, and the same trend holds for Australia and the US (see 鈥淲ait-and-see鈥� graphic). The increase in 鈥渓ate deciders鈥� goes hand in hand with a rise in tactical voting, in which voters switch their support to oust their least favourite candidate. But without accurate opinion polls, tactical voting can be haphazard.
Advertisement
In a bid to get around the flagging response rates faced by traditional face-to-face and telephone surveys, pollsters are turning to the internet. The trick is to pay people to take part, says Peter Kellner of YouGov, an opinion research company founded in 2000 that does all its polling online. YouGov has assembled an army of 90,000 people in the UK who get 拢50 per 50 to 100 online polls they take part in. 鈥淭his gives us response rates of up to 60 per cent,鈥� says Kellner.
鈥淰oters worldwide are relying on opinion polls more than ever to vote tactically鈥�
But response isn鈥檛 everything, say critics. 鈥淭here is no way of drawing a scientifically representative sample of the population from internet users,鈥� says Roger Mortimore, senior political analyst with traditional research company MORI. Kellner concedes that the 鈥渆xtremely old and the extremely poor鈥� are difficult to reach via the net, but says that phone pollsters have the same problem.
Internet polling has one unique advantage though: it removes the human interviewer. This might avoid the so-called 鈥渟hy Tory鈥� problem of some Conservative voters in the UK not wanting to own up to their ballot preference, says David Sanders, a political scientist at Essex University in Colchester. Shy Tories were partly blamed for the pollsters鈥� spectacular failure to predict the slender Conservative win in the 1992 general election. A similar phenomenon exists in Australia with supporters of John Howard鈥檚 Liberal Party, which also has right-of-centre policies.
YouGov鈥檚 internet approach got within 1 or 2 percentage points of each party鈥檚 actual result in the UK鈥檚 2001 election, but it did not work as well for the US presidential election last November, when it predicted that John Kerry would win by 3 points. The problem, according to Kellner, was that too many net-savvy Kerry supporters didn鈥檛 actually turn out to vote. Despite such slip-ups, though, pollsters as a whole have generally been very successful. In the UK for example, they have successfully predicted the winner of the popular vote in 13 out of 16 general elections since the second world war (see 鈥淥pinion poll accuracy鈥� graphic).
However, the job of telephone pollsters is getting harder 鈥� because of cellphones. More and more people are dropping their landline in favour of a mobile-only existence, and the mobile generation is skewed towards twenty-something professionals. Ofcom, the UK telecoms regulator, says that 7 per cent of UK homes rely solely on a cellphone. And according to the Consumer Expenditure Survey of the US Department of Labor, cellphone-only users there rose from 0.8 to 4.3 per cent between 2001 and 2003.
What鈥檚 more, because UK cellphones don鈥檛 have area codes like US cellphones 鈥� they use a code that identifies the network operator 鈥� pollsters cannot ensure that their sample is spread around the country.
Whatever the answer, British voters sick of being bombarded with statistics on the parties鈥� fortunes may wish to emulate the French. They ban opinion polls during national elections.
