[obog365]

On day 1 my class did an experiment to determine concentration of a solution, on day 2 we did a repeat experiment to determine the concentration of the same solution. So now there is data on the concentration from Day 1 experiment and day 2 experiment. The data for the entire class is collected and a plot of the day 2 concentration vs. day 1 concentration is made (a data point for the plot would be (day 1 concentration, day 2 concentration) and each data point would be from one student). A line is fitted to this plot and my professor says that the standard deviation of the residuals is an estimate for the  indeterminate error in the experiment. I don't understand how this could be true. What is his reasoning? thanks.

[ARGOS++]

Dear Obog365;

It can be used as an Estimator, - But it depends How you define “Indeterminate Error of ….”.

But it seems to be not a very well defined Estimator, as several Assumptions/Hypothesis are prior made.
And assumptions/hypothesis requires a “statistical” Test, including consequences!, before their future use.

Only a very few assumptions/hypothesis which needs a Test:

• Is the size of the “basic” population (# of Students; # of Experiments) large enough to take it as an Estimator for what you prefer?
• Are both daily means really independent, - or must the total mean be used instead? (t-Test).
• How much “systematic” Error(s) is incorporated in the difference of both daily means anyway?
  (Better trained Students at the second day; really identical samples/reagents; etc., etc.)
• Are both daily “Standard Deviations” really measure the same? (F-Tests).
• And so on.

I hope I have you not more confused as you was before!
And! – Also Estimators and their Results need a “Validation”!

Good Luck!
                   ARGOS⁺⁺

P.S.:  NO!  -  I'm not a real Statistician!
.