|
We have several clearly different types of events in the prediction
registry, although it is difficult to categorize them exactly. In order
to explore the question whether the type of event matters, we can,
however, make some reasonably obvious distinctions. Dean Radin made a
first order attempt at studying the effect of event categories in the GCP
registry as of June 2002, when there was a total of 105 formal events.
The following table shows his results, as a composite Z-score, the
number of events, and an effect size, calculated as Z/sqrt(N), for seven arbitrary but
reasonably distinct subsets.
| Category | Z-score | N events | Effect Size |
Natural disaster | 0.17 | 11 | 0.05 |
| Political | 0.23 | 8 | 0.08 |
| Meditation | 2.16 | 27 | 0.42 |
| Accident | 1.69 | 10 | 0.53 |
| War | 2.30 | 12 | 0.66 |
| Celebration | 3.92 | 29 | 0.73 |
| Funeral | 1.43 | 3 | 0.83 |
From this rough sketch, it appears that natural disasters and big political
events don't have much effect on the GCP, whereas war, terrorism and
violent acts do, as to the gentler occasions of celebrations,
meditations, and funerals.
Please note that this a preliminary result, that should be interpreted
with great care. We will do more assessments of this sort, and as the
database grows larger, we will be able to ask this kind of question with
more confidence.
|