Crunching numbers.

I'm not the only one analysing paranormal data with a serious approach. There are others out there. A lot.

The way you see an investigation portrayed on TV isn't real. We don't just wave EMF meters at random, we don't run to places where there might have been a 'clunk', we definitely don't run at the hint of something paranormal.

The actual investigation process, if TV were to show it, would be very dull indeed. It's methodical and slow, deliberate and progressive. There are several cameras involved, most on tripods, which have to be checked for battery and recording time and batteries/tapes/memory cards replaced as necessary. Voice recorders too - although these new digital ones have so much recording time there's no need to worry about that. Just their batteries.

No rewinding during the investigation, no matter what you think you saw. All analysis comes afterwards. The investigation is data collection. Times of potentially interesting events are noted for later, but nothing--nothing--is rewound and looked at at the time. Rewinding is the best way to miss something important or to accidentally overwrite whatever you found.

A discontinuous recording also smacks of tampering. It's impossible to prove the recording wasn't tampered with if it's in short, broken chunks. It might have been assembled later. The original must be in one continuous stream.

So the actual investigation won't make good TV. Analysis of the data won't make good TV either - watching someone sit, eyes closed, with headphones on while they track through sound for EVP's, or staring at a screen waiting for movement is hardly riveting.

It's therefore not surprising that the newly started beginner, even when they find something, is shrugged off by the sceptics. The TV shows give entirely the wrong impression of how to run an investigation. But then, that's TV. It's entertainment. If it doesn't entertain, it doesn't get the ratings.

It does, at least, raise people's interest in the subject which is good. On the plus side, it means that paranormal investigators are less likely to be labelled 'nut'. On the negative side, it means that anyone who thinks they have a haunting expects to cash in. I don't charge for an investigation - but I won't pay to go to work.

A good example is on the Paranormal Database, where they have analysed their database of UK paranormal reports by region. This is data collection and analysis put to good use.

The regions are, or course, arbitrary because they are based on counties/countries. The borders aren't defined by any kind of paranormal parameter so they aren't a perfect basis for the distribution of paranormal events. Nonetheless, since we have no other parameter at present, they will do.

London tops the list, because it has the highest population density and the least 'blank' space (space where nobody lives).

Wales is down at no. 43 - but then Wales is taken as a whole and not split into counties. Much of Wales is occupied only by sheep and shepherds, and large areas don't even have those. Most of the population is concentrated in towns in the easily accessible parts. I suspect, if it was split into counties, the same effect would show as in the overall database. There are more reports in areas of high population density.

I'd also like to see the data collated in terms of how long each settlement has existed. New towns, built on green space, I'd expect to have fewer incidences than those built on previous Roman or even Pictish sites, simply because many more people have lived (and died) there.

That sounds like it's worth doing because if there is a correlation between length of occupation/size of town and paranormal reports, surely that's useful evidence?

It is. It's evidence. It's still not proof because there are alternative explanations for such a correlation. For example, the more people live in a place, and the longer they live there, the greater the chance of stories being invented and handed down until they are eventually passed off as fact. So it's not proof.

Dividing the paranormal events by arbitrary regions doesn't allow any kind of statistical analysis. A better (though much more labour-intensive) method would be to draw a grid over the UK, determine the paranormal events/current population density/average age of settlements for the area under each square of the grid. The smaller the squares, the better.

That's a lot of squares. Statistically, you don't need to examine every single one. You could take, say, ten percent of squares and pick them at random by throwing a dart at the grid. No cheating and aiming for London all the time!

You'd then determine the parameters for each square and test for a correlation between paranormal activity and one or more of the variables listed. If you get a significant result it would then be worth doing all the rest of the squares.

It sounds dull. It is dull. It's slogging away at numbers for hours upon hours.

That's what science is really like. Playing with the gadgets, hanging around locations, travelling, yes it's all part of the deal but the biggest part involves sitting with a calculator and working out standard deviations.

If I find the time I'll have a go at the Paranormal Database myself.