Henbury loop analysis (well, moan....)

I have started to get a bit annoyed at the unscientific way in which the case for the Henbury spur has been dealt with, so I thought I'd blog about it, and the lack of science used in calculating the results


http://www.westofenglandlep.co.uk/meetings/joint-scrutiny-committee

 

The papers recommend:

 

Recommendations

 

1) To endorse the Preliminary Business Case and the progressing of Option 1A

 

(Henbury Spur plus Yate Turn-back), without Constable Road Station, to the

Outline Business Case (Programme Entry).

 

 

 

Essentially, the report uses something called a GRIP2 (or it might have been a GRIP4) analysis of information to come up with a benefit:cost ratio of the Spur v the loop.

 

I don't intend to go in to the detail but at a basic level:

 

• To produce a benefit cost ratio of take up of particular service will involve a large number of factors. Populations, catchment areas, policy decisions, growth rates in train usage, variations in such take up by location etc etc
• Each factor will have a level of accuracy (or inaccuracy) associated with it 
• The accumulation of all these factors will result in a large error
• The analysis is apparently done over a 60 year period. This will exacerbate the problem significantly.

  

From my memories of doing science (ok, a long time ago....), you would calculate an error to a 95%confidence limit.




This means you are 95% confident that your answer is correct between (Your value minus the error limit) to (Your value plus the error limit). Each value in that range is equally likely.

 

Hence a BCR of 2 with an error of 1 means the value is somewhere between one and three, you can not say with any more accuracy than that.

 

The values for the loop are 'BCR with WBS' = 1.35.


If the error margin is 0.7, it gives you a range of 0.65 to 2.05 (which is significant because a 'BCR' of 2 is required to qualify as a scheme.


Likewise, the BCR for the spur with WBS is 2.46. If it had an error margin of 0.7, the range would be 1.76 to 3.16. You'll note that some of these values are lower than some of those above. Given that each value in either range is equally likely, then you simply cannot say that one scheme is better than the other.


The obvious question to ask then - is what error margins/confidence limits did those advocating the Henbury spur come up with?


Answer: they didn't calculate one.