A Purpose For Frt Car Distribution Studies. Was: Re: Re


Malcolm Laughlin <mlaughlinnyc@...>
 

From: Bruce Smith
So, here's a philosophical question... For my foreign boxcar fleet, I need roughly 5 B&O boxcars based on the national fleet. So what cars? Well the WWII fleet consisted of the M15, M26 and rebuilt M15 and M53 wagontops. The knee jerk response of many modelers might be to use the iconic wagontops. After all, you can identify them as B&O without even reading the reporting marks! But I can't resist the temptation to educate and the wagontops turn out to be a cliche. Their combined made up just over 10% (or 1 in 10) of the B&Os boxcar fleet..

.===========

Let's not get too hung up on averages. After all, the average day rarely happenned. If 10 % of the cars were wagon tops, then there were probably days when out of ten cars five would be wagons. So you're modeling that one day in 20, but it wouldn't be reasoanble for anyone to suggest that you are unprototypical.


Malcolm Laughlin, Editor 617-489-4383
New England Rail Shipper Directories
19 Holden Road, Belmont, MA 02478


Anthony Thompson <thompson@...>
 

Malcolm Laughlin wrote:
Let's not get too hung up on averages. After all, the average day rarely happenned. If 10 % of the cars were wagon tops, then there were probably days when out of ten cars five would be wagons . . .
Of course, but the average is a place to start. Otherwise you are just winging it. I don't understand the excitement of many posters, to point out that there are substantial swings from the average. That doesn't eliminate the importance of the average as the core datum. Moreover, my instinct is that what "looks right" is probably the average, not the extremes of the data. Without more specific data, and we have precious little, I believe that trying to look typical is an excellent beginning.

Tony Thompson Editor, Signature Press, Berkeley, CA
2906 Forest Ave., Berkeley, CA 94705 www.signaturepress.com
(510) 540-6538; fax, (510) 540-1937; e-mail, thompson@signaturepress.com
Publishers of books on railroad history


Malcolm Laughlin <mlaughlinnyc@...>
 

Tony said
> That doesn't eliminate the importance of the average as the core datum. Moreover, my instinct is that what "looks right" is probably the average, not the extremes of the data. Without more specific data, and we have precious little, I believe that trying to look typical is an excellent beginning
=========================

I'm OK with the average as the core datum, but recognizing that the average in the car distribution case may be at one side of a skewed distribution orm otherwise not a true average. Any average is better than zero or infinity as the starting point.


Malcolm Laughlin, Editor 617-489-4383
New England Rail Shipper Directories
19 Holden Road, Belmont, MA 02478


Steve Lucas <stevelucas3@...>
 

Tony--

After reading this thread for some time, I have to agree with your
basic premise. Most major through transcon routes such as the UP in
Wyoming would very likely have over a given period, percentages of
each road's boxcars in keeping with the national averages. The law
of averages supports this.

The danger lies in having a model car fleet that models many
exceptional cars, rather than a good overall representation of the
national fleet. But it's too easy for the modeller to, absent
knowledge of each road's freight car fleet and its relative
percentage to the national fleet, to model those exceptions. For
example, as much as I want a model of a B&O M-53, there are other B&O
cars that I have come to learn to be more common to that road. I
have to have many CN boxcars (fortunately, I do) to justify modelling
just ONE of the three aluminum-sided cars that they had. If you model
20 B&O boxcars, you can justify having one M53 in your fleet far
sooner than I can, where I will only need one B&O boxcar for my
layout.

Likewise there is some weighting that one has to be factored in by
the modeller to represent the location that they model. Your road
may move a lot of grain in boxcars, and the ratio of cars (home road
vs. national fleet) will change whether or not the grain is running.
Likewise, boxcar precentages from the ORER are not a good way to
determine the car population for a grain-hauling layout. Preferred
will be 40' steel or steel-frame boxcars with 6' doors. Far less
common will be cars with larger doors, as they were practically
useless for grain loading. And open coal hoppers, though not part of
this discussion, will have interesting distribution variations of
their own!

Steve Lucas.



When translated --- In STMFC@yahoogroups.com, Anthony Thompson
<thompson@...> wrote:

Malcolm Laughlin wrote:
Let's not get too hung up on averages. After all, the average
day
rarely happenned. If 10 % of the cars were wagon tops, then
there
were probably days when out of ten cars five would be wagons . . .
Of course, but the average is a place to start. Otherwise
you
are just winging it. I don't understand the excitement of many
posters,
to point out that there are substantial swings from the average.
That
doesn't eliminate the importance of the average as the core datum.
Moreover, my instinct is that what "looks right" is probably the
average, not the extremes of the data. Without more specific data,
and
we have precious little, I believe that trying to look typical is
an
excellent beginning.

Tony Thompson Editor, Signature Press, Berkeley, CA
2906 Forest Ave., Berkeley, CA 94705 www.signaturepress.com
(510) 540-6538; fax, (510) 540-1937; e-mail, thompson@...
Publishers of books on railroad history


Mike Brock <brockm@...>
 

Steve Lucas writes:

"Tony--

After reading this thread for some time, I have to agree with your
basic premise. Most major through transcon routes such as the UP in
Wyoming would very likely have over a given period, percentages of
each road's boxcars in keeping with the national averages. The law
of averages supports this."

Well...perhaps someone's law of averages supports it but the data does not. Can you explain that?

Mike Brock


Steve Lucas <stevelucas3@...>
 

Wrong phrase, from an admitted statistical neophyte. I should have
referred to the "law of large numbers".

http://en.wikipedia.org/wiki/Law_of_large_numbers

Steve Lucas.

.--- In STMFC@yahoogroups.com, "Mike Brock" <brockm@...> wrote:

Steve Lucas writes:

"Tony--

After reading this thread for some time, I have to agree with your
basic premise. Most major through transcon routes such as the UP in
Wyoming would very likely have over a given period, percentages of
each road's boxcars in keeping with the national averages. The law
of averages supports this."

Well...perhaps someone's law of averages supports it but the data
does not.
Can you explain that?

Mike Brock


Malcolm Laughlin <mlaughlinnyc@...>
 

Steve Lucas writes:

"Tony--

After reading this thread for some time, I have to agree with your basic premise. Most major through transcon routes such as the UP in Wyoming would very likely have over a given period, percentages of each road's boxcars in keeping with the national averages. The law of averages supports this."
------------

I say that statement is dead wrong.
================


Well...perhaps someone's law of averages supports it but the data does not. Can you explain that?
--------------------------------------

The law of averages, or the law of large numbers, works only for homogeneous data sets. The national box car fleet did not meet that qualification. It could be reasonably be said to apply to the percentages of NYC, PRR and B&O amrks on the UP, but not to the number of thsoe marks releative to UP direct connsections.



Malcolm Laughlin, Editor 617-489-4383
New England Rail Shipper Directories
19 Holden Road, Belmont, MA 02478