determine / eliminate influence of confounding variable on 10 data series


New Member
Any help with the problem below would be greatly appreciated. I have very little knowledge of statistics. I am using excel.

I have measured concentrations for 10 different pollutants, in 44 wastewater samples taken over a year (including some 'not detected' which I have set to 0mg/l). The amount (mass) of each pollutant entering the wastewater is independent of the amount of water consumed and varies from week to week. Due to a malfunctioning flow meter I do not know the amount of water consumed in each week, which is also likely to have varied from week to week and of course inversely correlated to concentration.

Looking at the graph of the pollutants concentrations there are obvious peaks and trophs that are similar for all pollutants, and so likely to be the result of low or high water consumption in those weeks, but also some that are different for different pollutants and therefore likely a result of high or low pollutant disposal in that week.

I am interested in the variation over time of the amount (mass) of each pollutant disposed via the drain.
Is it possible to estimate the influence of water consumption on the concentrations, based on 'common patterns' (is this covariance?) in the concentrations between all pollutants?

Many thanks
Last edited:


No cake for spunky
I don't think you can statistically show that concentration is related to water consumption because you say you could not measure water consumption reliably. If you show that concentration is in fact varying enough significantly (a substantive not statistical definition here) then you can argue that something logically should be causing this. But you can't actually show it was water consumption without data for water consumption.