[PVRCNC] Re: [PVRC] Packet and SS, some stats. Read #s and weep.

Wed, 6 Nov 2002 10:05:51 -0500

Interspersed.

Remember that the point of my post was whether recommending a general
use of packet to improve club SS scores is likely to actually improve
club scores.

----- Original Message -----
From: "alsopb" <[email protected]>
To: <[email protected]>
Sent: Wednesday, November 06, 2002 8:40 AM
Subject: [PVRCNC] Re: [PVRC] Packet and SS, some stats. Read #s and
weep.

> Guy,
>
> Here are some remarks of a statistical nature on your analysis.
>
> 1) Why was median rather than mean used for score.

Not against the law. My choice for the reason I listed. The use of
medians is basic statistics, even if a little scarce in popular
statistics.

> Statistical tests
> of hypothesis can't be done on medians but they can be done on
means.
> 2) What was the standard deviation of the distributions?  Is is
quite
> possible that deviations are so great that one can't show anything
> but:
>     "The hypothesis that SOHP scores are the same as U scores is not
> contradicted by the data"

This text in quotes is not in my post anywhere.

> This conclusion in statistical mumbo jumbo
> terms is contrary to yours.
> 3) Where is the control group?  Without a control group of people of
> the same skills to compare to, conclusions are generally invalid.

The data is naturally bifurcated. It is not necessary to artificially
create a control group.

> You
> have assumed that either the U or the SOHP group constitutes a
control
> group with equal skills.  I question that.

I think you have to make vastly larger assumptions to NOT assume that,
e.g. that all packet operators are XXXXX, whatever XXXXX is. I know a
lot of the ops in BOTH groups. Even talking to myself privately, where
I'm not worried by political correctness, I can't figure a discernable
XXXXX.

>
> Now let's assume that the truth is that assisted isn't a benefit.

I only listed numbers for the SS, and made no statements about
extension to other contests. I made no calls for elimination of the
category. I made no statements about SO2R.

I personally surmise that having only 80 multipliers in SS caps any
possible packet advantage at a point reacheable by unassisted
operators (80.00 vs. 79.68 for all scores above median). Unassisted
ops can get into the high seventies only operating 8 hours. I only
operated 7 hours, used an amp, mostly called CQ, never went looking
for mults, and still got 72.

The question then becomes whether packet helps RATE. At this point I
have to surmise that the thing that helps rate the most is high power
and calling CQ. In SS the cost of a "miss" on a packet pileup is
greater because of the length of the SS exchange. With the growing
popularity of packet, the probability of multiple callers on any
packet spot increases. Another way of stating that would be that
situations where you have a lock on getting the first reply increase
AWAY from the packet spots.

If I'm calling CQ, and I'm the station SPOTTED, it's an advantage to
me, and a disadvantage to everyone calling me except the one who gets
called first. I can only work one right away, and everyone else waits
while their rate goes down. This by itself could be the ten or fifteen
percent.

>How
> do we get the ARRL and others to fess up and realize that it is the
> operator, not what technology he uses, that determines the outcome.
> SOA categories should be deleted and entered as SO.  Right now
> assisted operation puts one in the multi class in some ARRL
contests.
>
> Where is your analysis for SO2R?

Am I supposed to have an analysis for SO2R? Somebody paying me for all
these hours? Unless I've missed it, there is no ready-for-use
naturally bifurcated data for SO2R.

My only point is that assisted doesn't appear to cut it as a general
method to improve PVRC SS club score.

73.

>
> 73 de Brian/K3KO
>
>
>
> "Guy Olinger, K2AV" wrote:
> >
> > I hate to disagree with all the stuff about running "U" to run up
our
> > SS scores. The numbers tell a different and completely obvious
(IMHO)
> > story.
> >
> > Following gleaned from Dink's latest offering of 3830 listserve SS
> > claimed scores.
> >
> > An infrequently used term: Median, point where as many scores are
> > above as below, Gives a more meaningful concept of center when
there
> > are very large or very small scores that can skew an average
> > (arithmetic mean).
> >
> > Examining the characteristics of the "top half" of scores may
expose
> > trends among the most serious and/or proficient operators
> >
> > Median score
> >   SOHP 168.3k
> >   "U"  141.0k
> >
> > Average number of sections for stations above median score for
class:
> >   SOHP 79.68
> >   "U"  80.00
> >
> > Average number of hours operation for stations above median score
for
> > class:
> >   SOHP 23.90
> >   "U"  22.60
> >
> > Conclusions:
> >
> > Top Half scores are almost entirely full time efforts, with
non-packet
> > stations almost a perfect 24 hours.
> >
> > As could be expected, top half packet scores are all clean sweeps,
but
> > that is a microscopic advantage over an average of 79.68 among the
> > SOHP. No top half SOHP missed more than one section, and MOST were
> > clean sweeps.
> >
> > The variation in median score between the groups is fairly
constant
> > over the ENTIRE RANGE of scores. This indicates that on whole,
even
> > accounting for the difference in operating time, operating "U" is
a
> > DISADVANTAGE.
> >
> > There are a scattering of stations in "U" that bucked that trend.
They
> > are, to the man, all operators that have WON contests, divisions,
> > zones, etc, before.
> >
> > This leads me to suspect that operating "U" has some difficult
aspects
> > that take top rank skills to subdue. One obvious disadvantage of
> > packet sniping is that following spots there is always a pileup
when
> > you get there.
> >
> > The maxim still holds: Run an amp, call CQ, don't worry about
> > sections, and put in the hours.
> >
> > This year I tuned up on a clear spot on 10 above the crowd, and
when I
> > lifted the key I was called by the only station I worked in VI. He
was
> > 40 over S9, by far the loudest signal I heard the entire contest.
SS
> > has a lot of serendipity to it.
> >
> > An hour later I heard him again and he was only S5, and subject of
a
> > pileup. Probably couldn't have worked him. If I follow a packet
spot
> > and find him at the S5 level, I have to have the sense to keep on
> > going and keep up my rate.
> >
> > 73, Guy.
> >
> > _______________________________________________
> > PVRC mailing list
> > [email protected]
> > http://mailman.qth.net/mailman/listinfo/pvrc
> _______________________________________________
> PVRCNC mailing list
> [email protected]
> http://mailman.qth.net/mailman/listinfo/pvrcnc
>