1/X Weighting works for MS peak area data. Any good references as to why?
During an installation of an Acquity /Mass Spec system, I selected a 1/X data transformation of peak area in order to get the coefficient of variation of my standard curve (R squared) to maximize and approach 1.000. My customer is a long time user of UV data and was uncomfortable with the notion of massaging data. As one who came from that school, I can share that misgiving. Though as an empiricist, I tend to use what appears to be useful and this certainly does. I recall picking up a statistics text by Stuckey back in the 80s that argued that an elegant transformation of data to make it amenable to gaussian statistical analysis is a valid and powerful tool for understanding the data. I pointed out that UV data is itself the 1/logTransmittance transformation of photon data, making it amenable to Gaussian analysis, but my customer was unsatisfied. Sadly, I am unable to find that text by Stuckey, or even remember its title exactly. Does anyone have a good basic text or monograph reference on this point? A propagation of error discussion would be icing on the cake. Matt B the FSE