Mathematics desk
< July 30	<< Jun \| July \| Aug >>	August 1 >

Welcome to the Wikipedia Mathematics Reference Desk Archives
The page you are currently viewing is an archive page. While you can leave answers for any questions shown below, please ask new questions on one of the current reference desk pages.

July 31

Matrices are unique?

Why is the mattrix corresponding to a linear transformation $L:\mathbb {R} ^{n}\rightarrow \mathbb {R} ^{m}$ unique ?--130.56.71.52 (talk) 11:09, 31 July 2012 (UTC)[reply]

It is unique with respect to the choice of basis. The columns of the matrix correspond to where the basis vectors map to, so once you know the linear transformation those columns are uniquely specified. Staecker (talk) 12:26, 31 July 2012 (UTC)[reply]

The key fact is that every vector can be written as a unique linear combination of the basis vectors. (Which is true by definition.) This guarantees that the columns that Staecker mentions are themselves unique. Which tells you that your n-by-m matrix is itself unique, with respect to the fixed basis. — Fly by Night (talk) 18:53, 31 July 2012 (UTC)[reply]

Quality of polling

The Gallup Organization, Pew Research Center, Angus Reid Public Opinion, and ComRes. These are four of many listed in Opinion_poll#Polling_organizations. Is there a site anywhere that compares their Survey methodology? Like a consumer reports listing them by rank of quality? Some may just phone 10 people a week and sell those numbers. Two I found two that claim 1000 a day.--Canoe1967 (talk) 12:25, 31 July 2012 (UTC)[reply]

I think I figured it out. Depending on the value of the data they pay more for a better sample. For public opinion in a news story they don't need much. For building a mall or where to election campaign harder they need better numbers. I will resolve this unless someone else wants to kick it.

Resolved

--Canoe1967 (talk) 13:13, 31 July 2012 (UTC)[reply]

Provided one has a halfway decent number of samples you are normally far better off paying to do it well rather than make it bigger. Telephone surveys of tens of thousands of people for instance can be pretty useless whereas surveying a carefully selecting sample of a hundred people can be very informative. Dmcq (talk) 13:10, 2 August 2012 (UTC)[reply]