It’s been a long battle between Barack Obama and Hillary Clinton to become the Democratic nominee for the next US Presidential race.

The process for choosing the nominee is complicated. The final decision will be reached by delegates at Democratic National Convention in August. Most of these delegates will have been sent by their States with instructions on who to vote for, and different States will send different numbers of delegates. How the States decide who their delegates will vote for varies; some States let people vote, others hold caucuses.

At the end of the selection process, though, it’s simple: whoever gets the backing of a majority of delegates at the National Convention wins.

For most of the selection process, Obama has been the clear favourite. In fact, for months there have been calls for Clinton to concede defeat, but she’s hung in there. Today, as Obama closed in on number of delegates needed to win the race, Clinton came out with a strangely positive statement: “We’re winning the popular vote.”

Winning the popular vote? She’s as good as lost!

This is one of those sneaky stats that dresses up a defeat as a victory. Here’s how it works:

Clinton isn’t claiming to have the most delegates (she doesn’t, Obama does); she’s claiming have the most votes, and she’s right.

By focusing on votes, Clinton gets to ignore several States that are sending delegates to vote for Obama. In States that held caucuses no votes were cast. In Colorado and Minnesota, for example, zero votes were cast for Obama and zero votes were cast for Clinton, but at their caucuses there was strong backing for Obama and the majority of their delegates will vote for him at the conference. In the caucus states, Obama has done much better than Clinton but without getting any more votes than her.

By focusing on votes, Clinton also gets to include some States that aren’t sending any delegates to the party conference at all. In an attempt to grab centre-stage, Michigan and Florida brought their votes forward so that they would be among the first to choose a nominee. In doing so, they broke party rules, and as a punishment were stripped of their delegates. They held their votes anyway, and both States backed Clinton. In California Clinton beat Obama by over 420,000 votes, and in Michigan (where Obama’s name wasn’t even on the ballot paper) by 328,000 votes, but none of these extra votes translates into extra delegates at the National Convention.

So that’s how Clinton was able to say that she’s winning the vote even as she loses the nomination. It just goes to show that you can put a positive spin on any set of data if you have enough ingenuity.

The method used to collect data can have a significant affect on what the results seem to show. A good illustration of this is in British crime statistics.

There are two rival sets of statistics representing crime in Britain: the British Crime Survey and the police’s own figures. Neither is without difficulties.

The British Crime Survey has some clear limitations. Because it is based on asking members of the public about crimes in which they have been the victim, it has blind-spots. Victims of murders, for example, cannot respond to surveys, and so do not show up in the figures. Crimes committed against institutions (e.g. businesses), rather than individuals, are also omitted. Victimless crimes, such as drug use are overlooked by the survey too. The method in which the data is collected thus distorts the statistics, restricting their value.

The police figures are, in many ways, just as unsatisfactory. They only reflect reported crime, meaning that only a fraction of the more minor crimes committed make it in. This also means that any change in people’s willingness to report crimes will be indistinguishable from a change in the number of crimes committed; how are we to tell whether crime is falling, or confidence in the police is falling? Add to this the fact that the way in which crimes are classified has changed over time, and it becomes clear that the numbers can be misleading.

For these reasons and more, crime statistics cannot be taken at face value. Interpreting the figures is a complicated matter.