|
Some explorations have been done on the relative effect size
for various categories of events. Deciding on the groupings
and on which category a given event fits is a largely
subjective matter, but it is an important issue, so we have
made some effort to look carefully at this.
Dean Radin and Roger Nelson made independent lists of
categories in 2004, and plotted the effect size for each of
8 or 10 categories. The two following plots show their
results, which are generally compatible.
More recently (2007), Peter Bancel made another
foray into this territory, assigning events to fewer
categories and looking at two independent measures as well
as the relative weight of the available data in each
grouping. Here is his description:
I calc the netvar/covar for a breakdown of event categories.
I exclude early events (through event 25) where the network
is sparse.
I also exclude 9 astro events as these are small in number
and don't
fit well into any other category.
(the astro events are weird, too, since they have a strong
netvar...)
Blue is netvar and red is covar. The brown bar shows the
relative
proportion of events in each category.
The calcs are time-weighted and the brown bar is the
relative weight
of data-seconds per category, not the number of events.
The Disaster group includes natural disasters and accidents.
Politics includes national/political events as well as acts
of war.
Rest includes spectacle/media events and celebrations.
You can see the netvar is strong for Terror, Politics and
Disasters.
The covar is strong for Disasters and Meditations.
Is netvar more about horror/polarization and the covar more
about
openness/outreach ??
The questions raised by this study suggested looking at
another, more compact categorization. The intent is to
simplify the conceptual structure, to see whether deeper
insight might be possible.
In the following figure, Peter combines
Terror & Politics and Disaster & Meditations, leaving just
two characterized categories plus a third, catch-all group
(with about a third of the data).
The resulting comparisons allow us to ask whether there is
a difference between Terrorizing and/or Partisan
events and events that evoke openness or concern?
The 3 groups have roughly equal weight. Blue is netvar and red is covar.
|