Home
»
Trains Magazine
»
Forums
»
Passenger
»
Connecting sleepers to the Pennsylvanian

Edit post

Edit your reply below.

Post Body
Enter your post below.
[quote user="John WR"] [quote user="Sam1"] Once thing is reasonably certain.  Their numbers are based on statistical samples[/quote] Why do you say this, Sam?   Amtrak doesn't have to take a statistical sample.  It has a census of the whole population of Amtrak riders.   I don't know how Amtrak makes its projections.   I have heard making projections is like driving your car by blacking out the windshield and looking through the rear view mirror.  But projections are widely used. John [/quote] Amtrak knows how many tickets it lifts and, therefore, how many passengers connect to or from the Pennsylvanian and Capitol Limited in Pittsburgh. This is descriptive statistics. Projecting how many people might ride a through coach and sleeping car from the Pennsylvania to the Capitol Limited and vice versa requires the use of inferential statistics. Amtrak does not tell us how it derived its estimates for this proposed enhancement. The most effective way to determine the likely ridership would be to take a statistical sample of the population of potential riders. The population must be defined carefully, e.g. anyone over 18 and under 80 with sufficient income to be able to travel commercially. The size of the sample would be a function of the confidence level, confidence interval, and sampling error. Whether the sample would be with or without replication would also impact its size. The sample must be random. That is to say, everyone in the population must have an equal probability of being selected. The researchers must determine the best way to get the information, i.e. telephone interviews, questionnaires, personal interviews, focus groups, etc. They must build the queries so as not to bias the outcomes. These are challenging tasks.  Most market research does not go beyond telephone interviews and follow-up questionnaires. It is too expensive.  However, our company made extensive use of focus groups as to prepare for competition in the deregulated Texas electricity market. The researchers must project the results of the sample to the population as a whole as a range of potential outcomes. The range will be a function of the tightness of the sampling construct, i.e. we are 95 per confident with a margin of error of plus or minus three per cent that the potential number of riders for a through coach and sleeper from the Pennsylvanian to the Capitol Limited is 18 passengers per car per trip.  They cannot, however, project one number. At least not if they adhere to valid statistical sampling techniques! If Amtrak knows the current number of coach passengers that transfer from the Pennsylvanian to the Capitol Limited in Pittsburgh and, further, if it has collected the data for a more than five years, it could use regression analysis to project the number of people who would ride a connecting coach and sleeper from points east of Pittsburgh to points west of the Iron City. However, this projection would be biased because it does not take into consideration the market appeal for people who might use a through connection but shun the current arrangement because of the inconvenience of a late night or early morning transfer in Pittsburgh. Again, however, depending on the results of the regression analysis, the researcher could only come up with a range of potential riders and not an absolute number.   Amtrak's numbers don't suggest anything more than relatively high level market surveys, if that, but I could be wrong. One thing, however, is certain.  As soon as it says that the potential ridership for a new service is XX, they probably are wrong or would only be correct coincidentally.  If market trending data and projections are done with the right tools, they can result in some pretty accurate projections.  However, one should never lose sight of the fact that market projections, like political, economic, and stock market projections, are at least partially a fools game.
Tags (Optional)
Tags are keywords that get attached to your post. They are used to categorize your submission and make it easier to search for. To add tags to your post type a tag into the box below and click the "Add Tag" button.
Add Tag