Work Sampling: Definition, Theory and Confidence Level of Work Sampling!
Definition:
"Work sampling is a process where a large number of instantaneous measurements are conducted at random intervals over a certain time period or a number of devices, workers or operations. Every observation records what is going on at that time and the percentage of observations recorded for a particular activity or delay / idleness is a measure of the percentage of time that is taking place".
Work sampling has a long and impressive list of applications but all of them fall into one of the following three categories:
(i) Job measurements can be used as a working and idle times ratio test.
(ii) It can be used as a performance sampling analysis to assess work and work idleness and prepare a performance index.
(iii) It can be used as a technique for measuring work.
Theory of Work Sampling:
It states that the proportion of observations recorded during the operation / process in any State is reliable, provided "sufficient observations are made at random," estimate the percentage of time that the operation / process is in that state.
Here, the terms "random" and "adequate number of observations" should also be emphasized. Some errors may occur, but as the number of samples increases, the magnitude of the error tends to decline.
Job sampling is a method of sampling and is subject to probability law. The population distribution is well estimated by a random sample taken from a large population. Let's take the example below to make it clearer.
A worker does the task assigned to him during his shift or is left idle for one reason or the other. The table below shows that 45 working observations and five idle observations have been made of a total of 50 observations.
State of worker | No of observations |
Working Idle | 45 5 |
This table indicates the working time and idle time.
In this Example, the idle time percentage would be 5/50 x 100 = 10%
Working time would be 45/50 x 100 = 90%
This investigation is for one worker for a shift of 8 hours a day and indicates that the operator was idle for 10% or 48 minutes in a shift of 8 hours (480 minutes) while working for 90% or 432 minutes in one shift.
Confidence Levels:
The results obtained through the technique of work sampling differ considerably from the results achieved through continuous time recording. The exactness of the tests depends on the number of comments and the confidence level limits because a certain degree of error exists during the sampling procedure used. Therefore, the final "Job Sampling" results will determine which degree of trust is needed.
During a investigation, if we increase the number of observations considerably and in each observation then number of activities are large we can obtain a smoother curve called normal distribution curve as shown in Fig. 7.1.
The most common confidence level is 95%. The area under the curve at 2 sigma or two standard deviations is 95.45% which is rounded off gives 95% This indicates that the probability is 95% of the time the random, observations will be true or represents the fact and 5% of the time false or will not. For majority of cases, an accuracy of 5% is considered satisfactory. This is usually referred to as the percentage standard error.
Sample Size Determination. To obtain a desired accuracy level an analyst is required to take sufficient number of observations. Following formula may be used for finding the requisite number of observation in order to achieve the desired accuracy:
Limit of error = SpN
Where x =1, 2 or 3 for confidence level of 68%, 95% and 99% or one sigma, two sigma three sigma confidence levels respectively.
S = Desired relative accuracy.
P = Percentage occurrence of an activity or delay expressed in decimal e.g. 10% =0.10
N = Sample size or total number of random observations.
