[Prev][Next][Index][Thread]
(meteorobs) Re: SPO enhancement
Sirko & Rainer,
Your point is well taken, but I disagree with the assertion that my
observation have "no statistical significance" and thus inconclusive.
There is no such thing as absolutes in statistics. All statistical
conclusions are stated to a certain degree of confidence, be it 65%
or 95%, or whatever is the norm for the study. Yes, my observation
of sporadic enhancement between periods 2 and 3 is IN-conclusive
at 65% confidence, but IS conclusive at 55% confidence ! (Which
is 0.75 sigma, so HR becomes 12.8-3.6=9.2 vs 5.8+3.1=8.9)
Granted 55% confidence is not what I would "bet the farm" on, but
it could suggest something may be going on, however subtley.
(Note the Ondrejov radar results at the same time showed high
rates of faint meteors)
A couple of other observers have reported high SPO enhancements
in the past. Seems like I am the only one to notice it this time for
the Leonids. But I am not surprised, because one would need to
have observed under very good LM (implying moon and radiant
below the horizon) at 8 UT. That means Oceania observers, and
there's very few meteorobs out here (~.o)
Best regards,
Mike.
On 24 Nov 2000, at 10:30, Sirko Molau wrote:
> > It is POSSIBLE that due to statistical fluctuation, my 2nd period
> > count was elevated by +4.8 and simultaneously my 3rd period
> > count was low by -4.1, yielding a crosover as you say. However it
> > is EQUALLY POSSIBLE my 2nd period was low by -4.8 and my
> > 3rd period was high by +4.1, yielding a rate of 17.6 and 1.7 for
> > periods 2 vs. 3 ! So isn't it most PROBABLE the actual mean value
> > observed represents the truth? Of course, due to the small
> > statistics and crossover at one limit of 65% confidence, any
> > conclusion from these results are weak.
>
> they are not weak, but not conclusive at all! That's the point Rainer
> tries to make.
>
> Let's have a look at the following example: There are 5 observers watching
> the sky independently at the same time. They all observe the same average
> sporadic rate of, say, 10. Then you would not expect that every observer
> will have 10 SPOs in every intervall. In fact, there will be statistical
> fluctuations: One observer may have almost constant rates, the next one
> may have enhanced rates in the beginning, and another one in the end.
> This is simply what we expect from Poisson distribution of random events.
>
> If now one of these observers claims that he observes increasing/decreasing
> activity from his individual counts alone it's obviously not conclusive.
> By chance he was the one with the observed trend, not the one with
> constant rates.
>
> If an observation has no statistical significance, then you should always
> stick to the most probable explanation. In this case that's a more or less
> constant sporadic rate with statistical fluctuations. All you can do is to
> collect data from more observers, improve statistics, decrease the error
> bars and check whether you activity increase/decrease still holds.
>
> Cheers,
> Sirko
>
To UNSUBSCRIBE from the 'meteorobs' email list, use the Web form at:
http://www.tiacdot net/users/lewkaren/meteorobs/subscribe.html
References: