When using computer software to analyse data, it is easy to over-process images or sounds. So, if you are enhancing an image, if you go too far you can end up producing artifacts - 'things' that were never really there in the first place. The picture above is an original. The one below is a heavily sharpened version. The colours have altered and spots have appeared that were not there in the original.
Job done ...
After weeks of planning, you've finally staged your vigil in a haunted building. Your second priority, after sleep, is analysing all the data you've collected. Compared to staying up all night staring at nothing much, it should be easy.
Once upon a time, analysing vigil data meant sitting for hours watching video tapes of nothing much happening. There were also indecipherable investigator notes to cross check. While some of that still has to be done but a lot of data is now gathered automatically from instruments. Indeed, you may have CDs or DVDs full of megabytes of data streamed directly from computer-connected meters and sensors.
Computer data can look comfortingly accurate. However, it is still only as good as the instruments that collected it. If a digital thermometer has an accuracy of one degree, you can't start produce average to three decimal places. Data processing cannot increase the accuracy of an instrument.
It is tempting to buy specialised processing software. But just because it is used by the military or the space programme, it doesn't automatically mean it is the best for your data. There is always a danger, particularly with unknown software, of over-processing (left) or simply mangling your data. If you don't understand what some software is doing, how can you be confident of its results? If the output from processing is a wildly different from what you're expecting, consider the possibility it is wrong!
For examples of over-processed data see some an example of a photograph and sound recordings. See also here!
Talk to experts
If you think you've got some really exciting data, don't be afraid to consult experts. Often they will be intrigued by what you've done and happy to help.
So, if you're getting unexplained readings on your EMF meter, talk to a physicist. If there are things on your photographs you can't explain, consult a keen amateur photographer (they often know more about the technical side than professionals). They may be able to suggest things you should check before you decide your readings are really paranormal or unexplained.
EVP and photos
Two of the commonest instruments taken on vigils are sound recorders (often used to look for EVP) and cameras. We have suggesyions for analysing paranormal recordings (including EVP) and there is a whole section on anomalous photographs.
Playing the odds
With lots of instrumental data coming straight onto computers, statistical methods are called for. There is, unfortunately, no substitute for taking a course in stats or, at the very least, reading a book about them. Just producing pretty graphs with a spreadsheet doesn't make your results statistically valid.
It is easy to use statistics inappropriately. You will always get some sort of answer when using statistics. It doesn't mean it's valid or appropriate.
Consider an example. Suppose you want some people to judge an audio sample you've collected. You've decided that, in your opinion, it contains a voice, possibly of paranormal origin, saying a particular word quite distinctly. You give some judges a choice of five similar sounding words to choose from (including yours). What are the chances that they will agree with your verdict? One in 5? And if they ALL agree with you, will it be massively against chance?
The answer is, as you've no doubt guessed, no. It is a noise on a tape that you have interpreted as a voice. It might not even be a voice. The fact that several other people agree with your interpretation may only show that different people's brains interpret in a similar way.
In statistics, you are always comparing a result with the odds of it occurring by random chance. It is very difficult to say what the chance odds are of a sound being recorded that resembles a voice saying a particular word. Unless there is a sound mathematical way of calculating the chances of something occuring by random chance then you can't apply statistics to it (at least, not conventional stats).
© Maurice Townsend 2008,2012