Wrapping up Wrisc.
And I know some you want to Wrap up Wrisk (RISC) and throw it out with the fish heads. Anyhow, if you have been able to use RISC, you can follow through and have some fun(?) with this. If not, just glance through this sub module. But the last few pages, especially page 3 are very important conceptually. The use of Monte Carlo, or similar analysis is relay the only scientific way to express uncertainly.
If you need to do some Monte Carlo analysis, I strongly recommend buy the program Crystal Ball It's a little expensive, but you may be able to get a trial version. We won't do any more with computing by Monte Carlo in this course, but you need to be familiar with the concept.
Monte Carlo
As you worked through RISC you came across several options we did not use. In the Source Concentrations, we used a single value. You can also put in all your lab data and let RISC average it for you. Here are the input screens:
Arithmetic average and using the 95% upper confidence level (UCL)(based on the t-test),we did in 10A. Geometric and 95th UCL Lognormal are similar if the log of the data is "more normal" than the data itself. You could also assign a weight to the data, say one sample is 5 times more important than the others, you could weight it at 5 and the others at 1.
For handling the non-detects we are offered some choices as well:
However once we are all done inputting the data and the machine averages it, RISC will then chug through using the computed average.