Beware of bias

[Guest guest]

In jyotish research unless one is using a properly designed database,

with sufficient samples (horoscopes) which are uniformly distributed

and balanced in other ways, it is very easy to go astray. Any finding

which seems promising must then be questioned by the researcher. The

biggest question would be: is there a sampling bias? Sampling bias

arises easily in most databases (even commercial) because not all

categories are equally represented. An example that comes to mind and

which I have written about a few months ago was conducted to examine a

claim that was made about a certain trait being present in individuals

who had a certain planetary configuration. Two groups were used, one in

which the attribute was likely to be highly present and a comparison

group which were unlikely to have the trait. Sure enough, the

astrological factor was about four times higher in the first group than

the second group. The percent incidence (always must do that!) however

was similar. It so happened that the number of charts in the first

group was four times that in the second group! In other words the

astrological factor claimed to be associated with the trait was useless.

 

When we examine something like a mahapurusha yoga, surely we will find

it in charts of celebrities. We must then immediately see if the yoga

also is present in the charts of ordinary people. The second step is

important in the empirical research that we all conduct.

 

More later ...

 

RR