Competition - post a photo you've taken of nature >>

Importance of considering the accuracy of data

My local council has rejected my application for a school place based on the distance between where I live and the school being (and I quote) 0.657316789 miles.
Now am I going mad or is that ridiculous?
Is use of this number of significant figures not suggesting the data used has an accuracy of +/-5 ten billionths of a mile or +/- 0.03 thousandths of an inch?
I know distances of this order can not be measured to this accuracy and therefore to compare applications on this basis must be misuse of data like I instinctively know our Sun is not 1m away - but I can not make them understand.
I am weary and beginning to question my logic.
Any suggestions on how to explain the issue or verification of my logic would be appreciated.

Thanks,

Ian.

Thanks,

Ian.

Reply 1

TenOfThem

Original post by ianeck

I guess your argument is only valid if the distance they are comparing it to is given to a similar level of accuracy

What do their rules say?

Reply 2

ianeck

Yes but they are. The implications are that distance tie breaks (which would be decided on age) are artificially avoided because even if they occur there is only a 1 in 10 billion chance they would be revealed. If the data for the closest 50 applicants is arranged to a more reasonable two decimal places distance ties are frequent.

The rules say that they measure from the school gate to the applicant's house using a combination of a Land Gazetteer and the Postal Systems (hardly Nasa technology) and I have asked them to confirm the accuracy of their data but they can not.

I suspect they are using pythagoras to calculate the hypotenuse of the triangle formed by the difference in longitude and latitude coordinates. The nine decimal places is the default Excel format.

Thanks for your reply.

Ian

Reply 3

TenOfThem

Original post by ianeck

So it is not that you are outside of a circle but that another person is closer

Is that correct?

Reply 4

BabyMaths

Original post by ianeck

Perhaps the people at More or Less would be amused.

Reply 5

ianeck

Original post by TenOfThem

So it is not that you are outside of a circle but that another person is closer

Is that correct?

No I am not outside of a circle. They rank applications based on distance only and there isn't a limit as long as places exist. My concern is to rank applications using distances to a billionth of a mile when the data surely can not justify that accuracy will make any ranking unsound If I take the data say of 50 applicants and round to a more realistic two decimal places ties pop up all over the place. A tie indicates to me that the accuracy of the data is such that one application can not be distinguished from the next in the tied pair. The order in the pair therefore can not be determined. If one of a tied pair is given the last place and one loses out it is not certain whether the successful applicant is really the closest of the pair. In a tie situation different rules apply and childs age is used. Of course using nine decimal places ties never occur because even when they exist there is more chance of winning the lottery than a tie being revealed.

Maybe if we ignore the intricacies of school admissions law and focus on my concern that presenting distances in miles to nine decimal places infers an accuracy of better than 1/1000 of an inch. Is it likely distances of around half a mile can be measured with an accuracy of better than 1/1000 of an inch?

Thanks for your interest.

Ian

(edited 10 years ago)

Reply 6

TenOfThem

Original post by ianeck

Maybe if we ignore the intricacies of school admissions law and focus on my concern that presenting distances in miles to nine decimal places infers an accuracy of better than 1/1000 of an inch. Is it likely distances of around half a mile can be measured with an accuracy of better than 1/1000 of an inch?

The level of accuracy as given is clearly nonsense

I was trying to establish how it works so as to best advise you

So, now it comes down to the decimal place that distinguished the in and out line

So, did they give you the cut-off value or just yours

To clarify, if the cut-off value is given as 0.6 miles then the nonsensical accuracy is irrelevant ... if the cut-off value is given as 0.657316788 miles then you have real grounds for appeal

Did they issue a number?

Reply 7

ianeck

Sorry I am not making this clear. Thanks for persevering.

There are 30 places in the class which will be given to the closest applicants in order of rank based on distance (9 decimal places) from the school. The rules say that if two applicants are the same distance from the school (tie break) and only the last place is available the oldest child (of the tied pair) should be given that place

When the applications are ranked based on distances to nine decimal places no ties are evident. But when ranked based on distance to a more realistic two decimal places multiple ties emerge.

My thought is that by using the data without taking into account how accurate the data really is is artificially hiding tie breaks. For example say 0.678923467 and 0.678923469 are be ranked 30 and 31. Now there is only literally a hairs breadth between the two applications and so by any normal means this will not be resolvable. The real order could be the otherway round but we simply can not tell. If we can not tell a tie break should be declared. It never is for reasons already discussed.

Ian