Work sampling is a set of short-term observations that are based on random sampling for the purpose of recording frequencies of predetermined flow types in the same kind of work system (see more in Barnes 1980).
Obviously work sampling is just that—sampling work. At various times during a work shift the activity of each person is recorded as a code number, similar to taking a Polaroid snapshot of what the person is doing. Hundreds and even thousands of such instant observations are recorded over a period of many days or weeks. From these, a percentage of time spent on each activity is determined. It is important to understand that the activities are not timed. No stopwatches are used. An activity that takes a long time will simply show up more often (will be sampled more often), resulting in a higher percentage.
The activities of any one person are not important. What is important are the activities of the whole group, and how the mix of activities changes from month to month over the course of a year. The information is used to improve workflow and to schedule people properly.
How accurate is work sampling? After all, we are only observing activities at random instances. The statistical accuracy of a work sampling study depends mostly on how many observations are taken, both total and of a particular activity. With sufficient observations, work sampling is very accurate.
You, the subject
If you are a subject, a person being sampled, what should you do? Just what you always do. That includes such things as breaks and personal activities as well as your normal duties.
It is almost impossible to influence a work sampling study by changing your individual behavior. Work sampling measures groups, not individuals, so just follow your usual routine.
You, the observer
If you are going to record activities of others as an observer, you will be given training. You will be given a list of coded activities and shown how to use an electronic data collector to record observations of those activities.
What will the results be used for?
The work sampling results will not be used to assess individual performance. In fact, we have set up the data collection so that every day the subject numbers are rotated and the daily number assignments are destroyed.
The results will be used to help a team of line station employees identify opportunities for improvement and provide management with statistical backup for their recommendations. For example, if the team found that an activity such as preparing paperwork for shipping parts took up 25% of an employee’s time, they could look into a software solution to automatically generate the paperwork, perhaps reducing this to 15% of the person’s time. Looking at these data together with other data being collected, the team could conclude that this recommendation could result in reducing service times to mechanics by an average of five minutes.
In addition, the results will be looked at hour by hour and compared with call volume. Since the line stations’ activities are significantly driven by pushes, it will be very important for the team to look at recommendations based on busy times. In other words, when a push comes, what takes the most time, how can we make that faster, and what impact will that have on reducing service times and preventing delays?
How will we do this?
Having used this technique, the surveyor takes a tour within the business at predetermined times and observes flow types of specific work systems (for instance, the machine runs or the machine stops). A real picture of real flows based on several random observations can be attained by referencing to the work sampling technique’s testimony for a myriad of goals.
A typical work sampling study is generally extended to a few weeks. Thus, if the results are to be really useful, meticulous planning is a prerequisite.
- • The first goal of work sampling is to formulate and to select and describe the work systems to be observed. Work sampling is successfully used in identification of numerical indicators relating to the business such as mechanization level, utilization rate, loading rate, and workforce requirements; investigation of work flows in connection with the direction and planning of production; and identification of distributions of time.
- • Then, to reach the desired results, the distribution of observed system elements to different flow types is determined. Flow types herein are generally divided in more detail than needed.
In this way, a better overview can be obtained regarding the observed event.
- • In the next step a tour plan is generated. This is a sketch-like representation of the observation points and sequences.
- • The issue of the number of required observations is resolved by estimating the desired reliability level and share of flow type to be investigated most.
- • At this point, the time for each round is set. What is important here is the random timing. Thus, both the statistical requirements are fulfilled and the results are not manipulated by the observer. Tour times are determined with the help of hour-minute-coincidence charts.
- • After approximately 500 observations, generally an interim evaluation is conducted and whether the estimated number of required observation is appropriate or not is inspected.
For example, two supply attendants from each shift and location (one each day to provide coverage for seven days) will act as the observer for that shift. The data collector will be set up to beep randomly. The observer will do a tour and record what each worker is doing against a list of coded activities. No worker names will be recorded. A typical work sampling study form is shown in Figure 5.11.