Monday, April 06, 2009

SciFi’s Ghost Hunters

I have roommates who find Ghost Hunters amusing. I suppose it is, if you like watching a study of how not to conduct proper scientific research.

If you’re unfamiliar with the premise, The Atlantic Paranormal Society (TAPS), which consists of a couple of plumbers and their friends, go to “haunted” locations and attempt to verify whether paranormal activity is occurring. Needless to say, these people have little idea how to properly conduct such an investigation, and their methods are ripe for confirmation bias.

Normally, the TAPS crew visits a location in response to a request from the owners or residents. Their first action upon arriving is to question the “witnesses” of “paranormal” events, asking what they thought they saw and where they thought the saw it. With these “hot spots” identified, they set up cameras and microphones to record activities in those particular locations. The team then spends the night at the site, wandering through the hot spots, trying to provoke reactions from “spirits”, and recording any “significant” instrument readings or “personal experiences” consistent with the stories they were told. Need I mention that they do this in the dark, making them all the more susceptible to “chills” and other spooky feelings.

Like I said, their method is ripe for confirmation bias. Any unexplained noise or “chill” or draft is easy to attribute to paranormal activity. They do dismiss many events that can easily be traced to something mundane, but they seldom actually dismiss claims of a haunting (if ever - I've never seen it happen), because there are always a few “personal experiences” for which they found no specific mundane explanation.

The most obvious thing that their method lacks is any type of control, and I mean “control” in the scientific sense often associated with drug trials. In a drug trial, some test patients get a placebo instead of the drug being tested, so they have the same experience as the other patients, just no actual chemical. The patients don’t know whether their drug is the real thing or a placebo, so they can’t prejudice the results of the test with their knowledge (it's well known that receiving treatment -- even placebo treatment -- affects the expression of a patient's symptoms).

TAPS could do something similar in their investigations. First, they need some blinding. The investigators who spend the night in the site looking for evidence of paranormal activity can’t be the people who interview the original “witnesses”; knowing where events are supposed to happen taints the evidence. Second, they need to create some “placebo hot spots” – attribute some paranormal activity to locations on the site where the witnesses haven’t actually reported anything. Write out the full list of “hot spots” and “placebo hot spots” – with no distinction between them of course – for the investigators who will spend the night at the site. Finally, have a third team of investigators (more blinding) to analyze the evidence that the on-site team gathered without actually speaking to the on-site team first, again seeking to minimize confirmation bias.

If the experiences of the site's witnesses are connected to actual paranormal events, you would expect the true “hot spots” to generate significantly more instrument readings and “personal experiences” than the placebo “hot spots”. If there’s no significant difference between the two, then you can safely dismiss the activity in the “hot spots” as investigator error.

But Ghost Hunters would never institute such a rigorously scientific method of investigating a “haunted” site. That would inevitably result in an endless string of negative results, and they won’t be able to generate ratings with results like that, so they will undoubtedly continue to use the useless protocols they currently follow.


The Leaky Pen said...

And that, I suppose, is why Ghost Hunters is on the SciFi channel rather than, say, Discovery.

Lord Runolfr said...

Unfortunately, Ghost Hunters -- or actually a clone of it -- now is on Discovery.