"Only God Knows Who Will Win"

DW-WORLD.DE: Mr. Krämer, do meteorologists or pollsters provide more accurate forecasts?

Walter Krämer: I would say when taken in the short-term, both are equally good, and in the long-term both are equally bad.

It's polling institutes' busiest season, with all the projections pointing to a close outcome. How seriously should voters take the forecasts?

The forecasts should be taken seriously. But it's easy to forget that they are snapshots that can change from day to day. Only God knows what will happen.

Polling institutes continue to make some big mistakes. In the state elections in Saxony and Schleswig-Holstein, for example, the results were very different from the forecasts. What can explain that?

That's because each polling institute allocates the undecided voters to the various parties differently. It's an art. It's also the pollsters' secret trick and sometimes you can make some mistakes doing it.

Often people polled are listed as one of the sources of mistakes.

Sure, there are people who will lie through their teeth. That's the reason why it's impossible to use polls to figure out how many people will vote for the NPD (editor's note: the neo-Nazi National Democratic Party) because the people who like it just don't say it, and in these cases you need other methods to decide how many people it could be. Then there are other people who change their minds. There are people who wanted to vote for the CDU (Christian Democratic party) for the last six months and today decide they don't want to anymore. Things can change from one day to the next.

Will there be any last-minute changes this time around.

It could happen. It's happened enough in the past.

Just days before the election, the Allensbach polling institute was saying that about 30 percent of voters were still undecided. Do undecided voters make the pollsters' lives difficult?

The proportion of undecided voters is especially high. That's not surprising this time around since the current as well as the possible new government are demanding sacrifices from the voters, and many of them think they are choosing between the lesser of two evils. So people take their time in deciding.

Some critics accuse polling institutes of influencing their results by manipulating the raw data. What do you think of such an accusation?

The raw data has to be processed -- it doesn't work any other way. The many undecided voters have to be divided up among the parties, as I've already said. The institutes don't like anyone seeing all the cards and that's where the art of polling comes in. If people wanted, they would be able to manipulate the data. But these are all institutes that live from their reputations and because people trust them. They definitely would not make mistakes on purpose.

What secret tricks do the institutes use?

You can extrapolate from past elections what kind of voter movements there could possibly be, how the trends are moving and which parties undecided voters went for in the end.

So you would say the polling institutes do not attempt to manipulate the results?

There is room for a small accusation. When I look at what the Forsa Institute has done for years, the SPD is always a few points higher than at Allensbach. You can see where their money comes from, but the institutes cannot afford large discrepancies. Once an institute becomes known for conducting slanted surveys its customers run away.