Abstract
In the first paper of this series, five data processing procedures were compared by a simulation and three recommendable procedures were selected. These procedures were sitpulated in official standards such as ISO or JIS.It was found by an additional smulation that these selected procedures produced biased final values, in the case when the data contained an extremely highly discordant data. These biases were not serious but should not be ignored.In this paper, the causes of producing biased final values were discussed and three modified procedures were proposed. Simulation was carried out on the three procedures selected in the first paper and on the three modified procedures.To summarize the results, the procedure V-1 (modification of procedure in ISO 3271) is the best in these procedures. However in the case that there is a limitation about number of data, the procedure VI (modification of procedure in JIS Z 8402 in clause 3. 3. 2 (1. 2. 2))is the most recommendable.
