## How to analyze grain samples

by Teresa Acklin

A representative sample is achieved in part by collecting grain from a number of different positions in a bulk. Often, these samples are mixed together and divided down, and a single, sub-sample is analyzed.

This method gives an average value and is probably acceptable when measuring spot values for moisture content or protein. But it gives no clue as to the level of variation or how well the sampling method is estimating variation. It can be a very poor approach when detecting pests, as the key is the size of the sample examined.

A second approach is to assess samples individually. Simple statistical analysis then can be done on the results to estimate the range of variation and the probability of the result being correct. If the results from assessing a number of batches are compared and analyzed, the data can also be used for statistical analysis to give a more accurate picture of the quality of the whole bulk.

The first step in data analysis is to calculate an arithmetic mean, or average, and the extent of the range (maximum and minimum) about that mean. The next step is to calculate the standard deviation; most computer spreadsheets have a formula for standard deviation.

It is generally assumed that 95% of the variation can be covered by the range of two standard deviations about the mean.

For example: If the mean moisture content for 10 samples is 15.2% and the standard deviation is 0.2%, then 95% of the grain will be within the range 14.8% to 15.6% (two standard deviations plus and minus the mean)

Obviously, the larger the standard deviation, the greater the variability within a bulk of grain.

The next step is to use the standard deviation to calculate the standard variance. This indicates the degree of uncertainty in the estimate for a particular set of samples and covers the range of error likely to occur using an identical sampling pattern. Most computer spreadsheets also have a formula for standard variance.

The table shows actual results for moisture content of samples from a truckload of wheat. It also shows how to calculate the number of samples needed to reach a specific rate of Acceptable Error (AE), which is added to or subtracted from the mean value to obtain the range.

Table

 Sample 1 Sample 2 Sample 3 Sample 4 Sample 5 Moisture content (%) 14.5 15.0 16.0 15.5 14.9 Mean moisture (%) 15.18 Maximum 16.00 Minimum 14.50 Standard deviation (%) 0.52 Standard variance (SV) (%) 0.27
 The number of samples required to attain a specific rate of Acceptable Error (AE) is calculated by this formula: 4(SV/AE)2 = Number of samples If the Acceptable Error is set at 0.2% (plus or minus 0.2 percentage points from the mean), seven samples would be required, as follows. 4(0.27/0.2)2 = 7.28 If the Acceptable Error is set at 0.5%, only one sample is required: 4(0.27/0.5)2 = 1.16