Diagnostic data evaluation. Part III. Collaborative study evaluation: standard deviation considered to be a constant

The diagnostic evaluation of the crude protein collaborative study identified possible problems with a standardized interval for the Kjeldahl digestion, with sample preparation, and with excess sample weights. The standard deviations for the ham and the beef samples cannot be considered to be representative of the method because of these problems. The standard deviations for the frankfurter and the pork sausage samples for all analysts met the performance standard of 0.24. There was no evidence of analyst bias.