Double-blind refers to the same collected sample being divided into two samples and analyzed “blind” meaning that lab technician or off site lab does not know that within the samples being analyzed that several are duplicate samples.
When calibrating an analyzer the results can never exceed the laboratory accuracy and sample collection method. Variances can be due to sample collection, head space in sample container, temperature and duration and type of lab method and other variables. All can be minimized in a simple specific procedure outline for the product.
Moisture samples should be collected in a polypropylene or other suitable container with a moisture barrier lid. Container should be full with minimal air head space. Crush samples if possible to increase sample density.
Determine the appropriate sample size 200 – 450g, temperature and duration for weigh dry weigh lab determination of control samples. Many labs feature automated weigh dry weigh analyzers that predict sample moisture. The higher the temperature of the convection oven the quicker the extrapolate result, and usually a less accurate result. While this speeds sample turn around and increases throughput its contribution to variance must be considered. Hence, double-blind sample averaging minimizes this variation and improve calibration.
For example, using a weigh-dry-weigh gravimetric procedure with an oven temperature of 115?C for 6 minutes will yield a different result than running the sample at 80?C two or three hours. There also may appear to be “crisping” or “sample browning” at the higher temperature that adversely impacts accuracy. Again, this is in regards to calibration samples and not necessarily day to day production lab tests that require rapid turn around and high throughput.
Let’s view results for samples tested at 80?C, 115?C and those sent to an outside sample as double-blinds. Outside labs can be subject to the same variation as in-house labs and double-blind samples reveal the accuracy of their results too. The chart below shows the benefit of using double-blind samples for calibration and any subsequent validation to improve accuracy.