There's always someone somewhere with a big nose who knows

Compare and contrast the following:

Patrick Wintour, Guardian political editor, Thursday May 22.

Analysis undertaken by political betting websites suggests that since the war there has been a regular swing back to governments in general elections in comparison with byelections.

The swing-back is defined as the difference between the average swing to the opposition in by-elections and the swing to the opposition at the subsequent general election.

So, for example, the average swing from Labour to the Tories in byelections from 2001 to 2005 was 7.9%. The swing the Tories got in the general election was 3.1%. Thus the swing-back to Labour in 2005 was 4.8%. For John Major's last term the equivalent figures were 13.6% and 10.2% - a swing-back of 3.4% to the Tories.

Remarkably, this swing-back to government has been highly consistent since at least 1974, irrespective of party in power, government term, parliament length, number of byelections, turnout, and the relative strength of the Liberals. It has ranged between 3% and 5%. It thus has all the appearance of an "iron law".

With this:

From Political Betting back in 2006.

So, for example the average Butler swing from Labour to the Tories in by-elections 2001-2005 was 7.9%. The swing the Tories obtained in the General Election was 3.1%. Thus the Swing-Back to Labour in 2005 was 4.8%. For John Major’s last term the equivalent figures were 13.6% and 10.2% - a Swing-Back of 3.4% to the Tories.

The remarkable feature of this graph is that the Swing-Back to government has been highly consistent since at least 1974, irrespective of party in power, government term, parliament length, number of by-elections, turnout, and the relative strength of the Liberals in either by-elections or General Elections.

It has ranged between 3% and 5% with a very stable average of almost exactly 4%. It thus has all the appearance of an “Iron Law.”

Now, the language in a newspaper piece will often be tweaked by the editors and subs for various reasons, but the essence is pretty much the same.

Via (where else?) politicalbetting.com