Evaluation of Test Results        
  Sample dispatch | Evaluation of results | Rating system | Reporting | Annexes | Test Rounds
Reporting of Results

The laboratory may send the raw data and the means of the test results as they would be reported on an ISTA Certificate, together with the information requested to the ISTA Secretariat on the provided forms either by mail, fax or email. Data report forms will be provided by email and on the ISTA website for download and computerized data entry purposes. For some test rounds the laboratory will be asked to also report the results on an ISTA Certificate which will be provided.
Laboratories not having sent the results to the ISTA Secretariat two weeks before the deadline will receive a reminder. Laboratories that have any delay in reporting the results must inform the ISTA Secretariat before the deadline together with reasons for the delay, and an indication of when results will be reported. It will be decided on a case by case basis whether delayed results may be included or not.

Data Entry

The ISTA Secretariat staff verify whether the results are reported correctly and if the proper reporting procedures have been followed. Deviations may lead to a reduced rating for this round (Examples for basic reporting errors). The reported data are then entered in the data base for analysis and report generation.

Statistical Evaluation

The purity, germination, moisture, and tetrazolium results are statistically analysed using the publicly available and free programme called "R" following the method of Tattersfield (1979) Seed Science and Technology 7, (2), 247-257. The principle of the method is to calculate Z-scores (Zi) based on determining the true value represented by the estimated mean, after outliers have been eliminated.
First analysis is done to identify the outliers, meaning these results differ considerably and may bias the estimate of the 'true' mean. A second analysis of the results is then made, excluding the outliers to find the new mean and standard deviation. The laboratory's mean (xi) is correlated with the overall mean (x), the estimated 'true value' of the sample, and the standard deviation (s) derived from the data of the accredited laboratories only, according the given formula:

Zi = (xi – x)/s

These values are used to generate the tables and graphical presentation on the individual report sheet each of the laboratories obtain to provide them with a standardized comparison of their performance with respect to other laboratories. Details on the interpretation of the report sheets are given in Interpretation of PT report sheets.

The Z-scores provide a good indication of a laboratory's analytical competence. The probability of a Z-score greater than ±2.0 or ±2.68 for a single test component (example pure seed in the purity test) is less than 5 % or 1 % respectively. Results between ±2.0 or ±2.68 provide an indication to the laboratory that it may have an analytical problem and it should verify its state of readiness to perform the test in question. Individual Z-scores between 2.00 and 2.68 (or –2.00 and –2.68) indicate a possible problem which the laboratory should investigate. Results greater than 2.68 are considered unsatisfactory and the laboratory should actively investigate the cause of the discrepancy. The Z-scores for each component are also averaged over the three samples to further clarify the laboratory's performance represented by its accumulated deviation from the expected mean.

The analysis of performance for other seed determination is based on a retrieval and identification rate of the added seeds. The performance in OSD is reported as percentage of retrieved and identified seedsto genus level of seeds added to a round's sample. The relative difficulty of retrieving a species from the sample under test is taken into account as explained below.

Last update: 24.05.2017 (BO)


Latest issue of Seed Testing International Issue 156

STI 156 October 2018
More info »

Rules 2019 now available online

More info »