How To Take An Activity Sample

To conduct an activity sample, two important rules should be applied:

1. The sample must be conducted over a representative time period. This could be as short as one day, a week or even longer, but the important thing is that it represents a normal working period. Often the workers select some of the dates and the management the others, so the sample is seen as fair by all areas. If the sample is not taken over a representative period, it may be statistically accurate but not reflective of the normal working situation.
2. The sample must meet the recognised statistical tollerance levels to be valid. To do this we need to take enough observations at set or random points, during the representative time period. Calculating the number of observations can appear to be confusing, given the formula used, but in reality this is quite straight forward.
When preparing to sample we need to reach the statistical tollerance of +/-95% confidence level for our desired accuracy. If we are using the sample to generate a time standard then we should aim for an accuracy of +/-5%, and if the sample is being used for calculation of contingencies or analysis of lost time we should aim for +/-10% accuracy.
The formula for the amount of samples we need to take is: