Re: Freight car with ICC report

Aley, Jeff A


I was thinking more along the lines of Dave's message of May 4, 2009 (STMFC #81532):


The wheel reports did, IMO, confirm the basic hypothesis and allowed us to
further specify when it applied: Post WWII, on mainline trunk routes,
excluding home road boxcars, the percentage of boxcars marked for foreign
roads will closely match the percentages of boxcars contributed to the US
fleet by each railroad. By boxcar, I mean one that can be put into general
purpose use. The wheel reports do show a bit of a bias towards nearby
connections, but here I believe the sample size of locations that we have is
too small to make a good hypothesis on that point. Further, as almost all
railroads contributed less than 5% of the national fleet (most less than
1.5%), even a large bias towards local connections would compute to a very
small number per 100 foreign road boxcars.

Last, the hypothesis makes no predicition on what one might observe in
individual trains.

[End Quote]

Somewhere, I think there's a msg stating that locations of direct interchange would have additional bias (e.g. Ogden on the SP). But I haven't yet found such a msg today.



From: STMFC@... [mailto:STMFC@...] On Behalf Of Bruce Smith
Sent: Tuesday, April 13, 2010 10:35 AM
To: STMFC@...
Subject: Re: [STMFC] Re: Freight car with ICC report

On Apr 13, 2010, at 11:22 AM, Aley, Jeff A wrote:


My recollection of the G-N model is that it DOES
account for regional bias. It will be difficult for us to educate
each other if we are each talking about a different "G-N model"!

If you have the time, please point out to us which
STMFC message contains the G-N model. I will also try to do so,
but I won't be able to get to it until late in the day.

It is important to recognize that both emails I cite below are early
examples of our discussions on this subject, and that the discussion
has matured over the years so that these may not represent the
ultimate iteration of the NG hypothesis.

I would start with Dave Nelson's message 18550, May 5, 2003. This is
perhaps the earliest and most basic statement of the hypothesis. In
this email, Dave does not mention anything about regional biases, but
does allow for 20% of the fleet being "other cars". Then, I would go
to Tim Gilbert's message 18757 of May 8, 2003, which is a
comprehensive statistical discussion. Interestingly, in that email,
Tim did say "the modeler should use a fudge factor to account for the
bias towards nearby railroads which might interchange directly with
his modeled road in the 1945-1955 era." However, his DATA, says that
this is not true. When looking at the 1946 Southern RR wheel
reports, the Southern region actually had slightly fewer home region
cars than predicted (10.5% vs. 11.7%). The 1947 UP wheel reports
have slightly more Central West region cars than expected (17.9% vs
15.7%). Statistically, both of these are likely to be judged to be
the same (you cannot reject the null hypothesis) and therefore, there
is no real regional bias present, or if there it, it is so small as
to be statistically insignificant (which is the same as saying it
ain't there <G>).


Bruce F. Smith
Auburn, AL

"Some days you are the bug, some days you are the windshield."
/ &#92;
__<+--+>________________&#92;__/___ ________________________________
|- ______/ O O &#92;_______ -| | __ __ __ __ __ __ __ __ |
| / 4999 PENNSYLVANIA 4999 &#92; | ||__||__||__||__||__||__||__||__||
| O--O &#92;0 0 0 0/ O--O | 0-0-0 0-0-0

Join to automatically receive all group messages.