Ask an Expert


Ask an Expert
Question and Answer Page
Statistics

Submitted 3/18/04
Q: I am writing an FSS plan that calls for 10 percent of gross alpha/beta direct (static) measurements to be repeated for QC purposes. What is the allowable variation (I was going to use +-20 percent)? Where can one find guidance on this?

A: Replicate surface activity measurements can be a valuable part of your QA program. QC measurements should be done in such a way that they capture precision as well as bias in the measurement. The MARSSIM addresses the issue of assessing measurement precision and bias in Section 4.9.1: "In this case, detecting bias might incorporate field replicates by having a second operator to revisit measurement locations and following the same procedure with the same instrument as was used by the first operator." To capture instrument bias it further states that “A field instrument’s calibration can also be checked by one or more operators during the course of a survey and recorded on a control chart. Differences in setup or handling of instruments by different operators may reveal a significant source of bias that is …..” Setting acceptable precision levels (variability) should be done using the DQO Process. The following points should be considered.

Given that background radiation levels may vary between the initial and repeat measurement, net surface activity levels should be compared. Second, the allowable variation should consider the uncertainty in individual surface activity measurements. This can be determined by making a series of measurements at the same location. If the uncertainty in surface activity measurements is greater than 20% (which is quite likely at low levels), then it may not be appropriate to set a limit on measurement variation at 20%.

Careful consideration should be given to how measurements that indicate zero net activity (i.e., background levels) are compared. That is, initial and repeat measurements that both indicate background levels (or are both less than the MDC for that matter) provide confidence in the initial measurement, but at the same time may have a variation much larger than 20%. Indeed, MARSSIM states in Section 6.2.2, "Determining precision by replicating measurements with results at or near the detection limit...is not recommended because the measurement uncertainty is usually greater than the desired level of precision."

Appendix N of MARSSIM expresses flexibility in determining the level of measurement precision, stating that "the level for each of these performance measures should be specified during development of DQOs." That being said, if the individual measurement uncertainty of surface activity measurements is determined to be close to 20%, then setting a performance measure of 20% variability between replicate measurements may be appropriate.

Submitted 7/14/03
Q: When using the Wilcoxon Rank Sum Test I have levels (in pCi/g) that are not reported (below the MDC) on the laboratory analysis. What value should I enter into the spreadsheet? The MDC value or zero?

A: According to the guidance provided in MARSSIM, reporting requirements for individual surveys should be developed during planning and clearly documented in the QAPP. The Health Physics Society has developed several suggestions for reporting survey results. According to their suggestions: report the actual results of the analysis (do not report data as "less than the detection limit"); report the measurement uncertainty for every measurement; and report the minimum detectable concentration (MDC) for the measurement system as well as the method used to calculate the MDC.

The WRS test allows for "less than" measurement to be present in the reference area and the survey units. As a general rule, the WRS test can be used with up to 40% "less than" measurements in either the reference area or the survey unit.

While applying the WRS test, if there are n "less than" values, they are all given the average of the ranks from 1 to n, they are all assigned the rank (n+1)/2, which is the average of the first n integers.
 

Submitted 10/4/02
Q: Here is a theoretical question that I have not been able to find a clear answer on:

I am performing a Final Status Survey (FSS) on a Class 1 land area that has been remediated. I have drawn systematic samples and I am performing a gamma walkover to look for small areas of elevated activity.

I find some questionable areas, and draw biased samples. The samples exceed the DCGL-EMC and must be remediated. The systematic sample results are a small fraction of the DCGL-W.

Here's the question. Once I remediate the hot spot(s), do I need to re-evaluate (e.g. re-sample) the entire survey unit, or may I use the existing systematic data "as-is"?

A: This question comes up frequently in the MARSSIM courses that we teach. The first point that I'd like to make is that the order of final status survey (FSS) activities should be scanning, biased sampling as a result of scanning, and finally systematic sampling. The idea being that you'd want to perform your statistical sampling after you've addressed any hot spots that need to be remediated because they exceed the DCGLemc.

But to answer your question directly, it is possible to use the existing systematic data "as is" provided that two important conditions are met. First, does the survey unit pass the statistical test with the systematic sample data? If not, then the entire survey unit fails and needs attention, not just the elevated areas that you've identified during scanning. In your problem statement you state that the systematic sample results are a small fraction of the DCGLw, so it very likely that the survey unit passes the statistical test.

Second, assuming that the systematic samples resulted in the survey unit passing the statistical test, it is critical that the remediation does not result in the spread of contamination to other land areas in the survey unit. The whole point of collecting statistical samples is that they accurately represent the population (in the survey unit) at the time of the FSS. If conditions change (i.e., due to re-contamination in the unit) following the collection of the samples, then the data set would be invalidated. Therefore, the hot spots must be carefully remediated and those remediated hot spot locations scanned and sampled to ensure that they were adequately remediated. Any indication that contamination was spread to other locations during this process would undermine the defensibility of the initial systematic sample data set as representative of the survey unit population.

 

Please visit this page again soon for additional questions submitted to our D&D experts and the answers that may assist you!

Return to Q&A Page