Forcing y-intercept through the origin (0,0) and doing a Robust Regression (w error bars)

I'm not sure that Excel has this sort of thing "built in", but I'm reasonably sure that Excel can be programmed to perform the analysis. I cannot say whether the spreadsheet will be the "easiest programming language" for the job, but I expect it can be done. I also will add that I am a little hesitant to respond because I am not enough of a statistician to know all that may go into a "robust" regression.

A recent similar discussion http://www.excelforum.com/excel-gene...-question.html that you may find interesting.

Because of you requirement that this be "robust", I would probably suggest something built around Excel's built in Solver utility. Since you have not provided any real details, here's a brief overview.

A) Choose regression equation and objective function.
B) Using your raw data and your chosen regression equation, compute your desired objective function. For example, a least squares regression will compute the sum of the squared differences. I would expect that your OF will include the weighting factors and/or x,y error values in some fashion. This is more statistics than Excel, so I will assume that you know how these should be part of your OF.
C) Call Solver and tell it to
C1) Set target cell -- OF
C2) to a minimum ( or maximum if appropriate)
C3) By changing the regression parameters of the regression equation.
C4) subject to any necessary constraints.
D) Evaluate the resulting regression to see if it is reasonable and meets your requirements.
E) Compute any additional statistics and error values using appropriate formulas. Again, I assume that you as the statistician on this project know how to do this.

What part do you get stuck on?