TwoTerrors - A common problem with admission criteria is that the tie breaker is stated as if it is an additional category. So, for example, a fairly common set of criteria is looked after children, then siblings, then distance. What they really mean is that it is looked after children, then siblings, then everyone else with distance being used as a tie breaker. So if the total of looked after children and siblings is more than the number of places, they would use distance to determine which siblings get in. I've seen this repeatedly with criteria for LA schools as well as faith schools and academies.
I'm so used to it that I automatically assume that any final criteria which looks like a tie breaker is actually the tie breaker for any category that needs one. So yes, the random allocation does seem to refer to category 5 but in practise if admitting everyone in category 4 would take them over the limit for the band they would use random allocation there. Of course, unlike scholarship places there is no limit on the number of places in each band that can be given to children in categories 1, 3 or 4 and it would be unusual for there to be so many children in those categories that it took them over the limit for the band, so the tie breaker won't generally be needed for those categories. (I hope that last sentence makes sense!)
Given that random allocation is the only form of tie breaker stated in the admission criteria, they must use that wherever a tie breaker is needed. If they do anything else an appeal panel will not be impressed. So therefore they must use random allocation to determine which scholarship students get in.
The scholarship places being spread evenly through the bands is there because it isn't there, if you see what I mean. There are up to 30 scholarship places. They allocate children into three ability bands and then admit from each of the bands in turn with meeting the scholarship criteria being category 2. Significantly more children meet the scholarship criteria than there are places available. That means there must be a certain number of scholarship places for each band. In the absence of any statement to the contrary, those places must be spread evenly across the bands.
I can understand the school feeling that its admission criteria are too complex to publish in full in the booklet Southwark send out as they only have limited space for everything they want to say. However, they MUST publish their admission criteria in full. If there is some other document that gives more detail than their "admissions policy" they are definitely breaking the rules. Any appeal panel will only look at their published criteria in determining whether they have complied with the Code and operated their admissions correctly. The same is true of the Local Government Ombudsman.
I understand your comments about knowing how many children have passed the scholarship criteria but the school isn't required to publish this information. You are supposed to use what happened last year as a guide to your chances. I accept that in this case there doesn't seem to be any good reason not to publish this information and it would probably be a good idea if they did but they are nonetheless complying with the rules on this point.
Violetqueen - Yes, I agree that what is clear to someone like me who reads loads of admission criteria may be less clear to others, which is why I was interested in understanding the issues. Clearly schools in Southwark vary. Kingsdale is quite complex with the interaction between fair banding, scholarships and random allocation. It could certainly do with being explained more clearly. The criteria for some of the RC schools look complex although I think they aren't as bad as they look. I still think the Harris Academies are pretty straightforward. However, with so many schools using fair banding it is a real pity that Southwark haven't bothered to describe what it is and how it works in their booklet.