We have a large set of data (approximates to normal distribution), which we are using to create a set of three ratios (Ra, Rb, Rc), for which we can calculate standard errors. These ratios are then being applied to a fixed survey(x), which has a known standard error, to move the date of the sample to a standard date.
We have to present the errors on our overall count, and this has raised two questions:
1. How should we calculate the standard error for the output after we have multiplied the data together as below?
Y=x*Ra*Rb*Rc
2. How should the standard error be calculated from this when we add multiple sites (different values for Y) together?
Many thanks for your help.