Key Points:

Statistics get a pretty bad reputation in this country, maybe even worse than the politicians who wield them. Dodgy facts are easy to produce, but hard to refute. They're usually not outright wrong, just misinterpreted or taken out of context.

It makes politics an unfair arena.

Good policies based on sound facts have to compete with the noise from oversimplified spin and outright misinformation. Even worse, in an environment driven by soundbites for the 6 o'clock news, good policies are often defended with spin, just because there isn't enough room to lay out the facts.

So here they are. We take the claims and find the facts.

"Violent youth crime is at an all-time high".
JOHN KEY (STATE OF THE NATION SPEECH, JAN 29)

That's true, but violent old-people crime is at an all-time high, too. Violent crime for every age group over 13 is, technically, "at an all-time high", and the fastest growing group of violent offenders is in the 51-99 category. Boot camp for old people, anyone? Young people commit more violent crimes than people over 30, but the increase is happening across the board.

Youth crime as a whole is actually decreasing: 2006 saw the lowest number of arrests for youth offending since 1995. When the change in population is taken into account, that's a 17 per cent decrease over 10 years.

Dr Fiona Beals, Education Studies lecturer at Victoria University, says the way young people are stereotyped in crime debates has a negative impact, especially on those already vulnerable.

"We often get [crimes] happen in this [summer] month that we wouldn't have happening in winter".
ANNETTE KING (NZPA, JAN 31)

More violent and sexual crimes do occur during warmer weather, says Pat Mayhew, Director of Victoria University's Crime and Justice Research Centre.

While this can be partly attributed to tempers rising with the temperature, there are practical reasons, too. People stay out later at night, when crime is more likely to occur, and drink more, which also contribute to violent crimes.

In the colder months, property crimes are more likely. While seasonal differences can explain the spate of murders in January, it doesn't really need explaining at all.

The number of murders in a month is very small, so even tiny, random change can appear to make a huge difference, when it actually means very little over the long term.

This statistical volatility means that we can't draw conclusions based on a month or two, we have to look at the data over a much longer period.

Analysts generally work out the long-term trend by looking at five year blocks. Between July 2002 and July 2007, there were 489 cases of homicide (including attempted murder, manslaughter, etc.).

In the five years before that, there were 523.

"Why is violent crime against innocent New Zealanders continuing to soar?"
JOHN KEY (STATE OF THE NATION SPEECH, JAN 29)

It might be because the police are getting better at filing. The biggest changes in violent youth crime rates occurred in mid-2005 when the police changed to the new National Intelligence Application database.

An independent report in 2006 found that the level of recorded crime shot up dramatically after July 1 2005.

Violent crime, which rose 3 per cent on the previous year, suddenly jumped to being 10 per cent. But if there was a real surge in violent crime, it didn't show up in 111 calls or ACC claims.

The report concludes that "the increase in recorded crime is not primarily driven by an increase in actual criminal incidents, but by different recording practices associated with a new computer system".

While there is evidence that the actual incidence of violent crimes is on the rise, it's probably a smaller increase than the raw figures suggest.