Analyzing Labor Productivity through Work Sampling

Introduction

Construction companies are constantly searching for ways to improve labor productivity. Since labor is one of the greatest risks in a construction contract it must be controlled and continuously improved. The construction company with the most efficient operations has a greater chance to make more money and deliver a faster construction project to the project owner (Orth & Jenkins, 2003). There are several factors that affect labor productivity on a jobsite such as weather conditions, workers’ skill level, overcrowding of work crews, construction methods used, and material delivery/storage/handling procedures (Jenkins & Orth, 2003).  Several methods exist for measuring and analyzing worker productivity. These methods include work sampling, audio-visual methods, and the method productivity delay model. This study attempted to isolate areas of potential labor productivity improvement by observing the labor force using work sampling.  Work sampling is a method that evaluates the amount of productive, supportive, and non-productive time spent by the trade workers engaged in performing their assigned work activities.  Work sampling also helps identify any trends affecting labor productivity.

This paper presents the guidelines used by the authors to develop and implement a work sampling study for work performed on a hospital construction project. This particular study took place during July 2004 and was used to evaluate the work conducted by the company’s three profit centers: fireproofing, waterproofing, and insulation on a hospital project in the southeastern United States.  The authors believe that work sampling is a useful method for analyzing site productivity, providing solutions for existing productivity problems, and establishing a baseline for comparing future studies, due to the value of information obtained from this study.  Although this paper focuses on applying work sampling techniques to a hospital construction project, the guidelines presented in this paper may be modified to conduct such a study on any type construction project.

Work Sampling

Work sampling is a series of instantaneous observations, “snap shots,” of work in progress taken randomly over a period of time (Jenkins & Orth, 2003).” This method provides information on the amount of time workers spend performing productive, supportive, and non-productive work (Jenkins & Orth, 2003). Work sampling is a technique that provides valuable information to a construction manager regarding areas of low productivity that need corrective action (Thomas & Napolitan, 1999). Work sampling’s main advantage is that it allows for simultaneous monitoring of labor productivity in a variety of trades (Noor 1998).  When monitoring more than one trade, the observer(s) need to distinguish the different trades so observations are not incorrectly categorized.  Work sampling can be conducted by anyone with basic knowledge of both construction and work sampling methods.  However, the authors suggest that the person(s) conducting the work sampling study be a neutral party and not employed by the company being evaluated. This will help to reduce bias of the study and reflect the actual conditions of jobsite labor productivity.

In conducting a work sampling study, the observer(s) start at a random time and follow a pre-planned route throughout the jobsite. As each trade worker is encountered, the observer categorizes and records the activity of the individual based on the instant the observer came into contact with the trade worker.  Categories of work are classified as being productive, supportive, or non-productive.  By definition, the categorization of the activity performed by each individual trade worker at the time they are observed is considered to be a “work sample”.  Productive work is defined as direct, hands-on activity to physically construct the project, such as insulating a pipe or painting a door.  Supportive work is defined as any activity used to support the productive work such as measuring a pipe or stirring paint.  Non-productive is defined as wasted time spent on the jobsite that can be recovered as productive time such as waiting for instructions, starting late or quitting early on a work activity.

The study continues until observer(s) record the pre-determined number of samples.  The data obtained for a work sampling study will become more reliable as the quantity of work samples increases. Statistical tables relaying the quantity of samples required for a high probability of accuracy are available to help determine the required number of samples (Adrian & Adrian, 1995).  At the conclusion of the study, all observed work samples are compiled together and analyzed. Recommendations for improving productivity are then based upon these findings.  While collecting data it is important to take work samples throughout the entire day or shift (Thomas & Daily, 1983). If the observer does not take into account the time at the beginning and end of the day or before and after lunch then the data will show an inflation of productive work by eliminating set up times and times when workers are putting tools away (Thomas & Daily, 1983).

Methods

A meeting was held with the general manager of the contractor in question to describe the procedures of the work sampling study.  The data collection method was described as well as the type of information that could be extrapolated during the analysis phase.  After the general manager was familiar with the process and the information that could be obtained from a work sampling study, an objective was determined.  The contractor wanted to have a baseline of the labor productivity for the company’s profit centers:  insulation, waterproofing, and fire proofing.  He wanted to know if a particular profit center required more immediate attention due to lower direct work percentages.

A large hospital in the southeastern United States was chosen as the subject of this study’s focus between the dates of 7/13/2004 through 7/20/2004. This project was selected due to the fact that the contractor had crews from each of the three profit centers scheduled to work at the same time. The hospital was also a typical job for the contractor with regard to schedule and scope of work.   

Once the objective was determined and the project was chosen, the next step was to define the work categories.  This meeting included the three superintendents for each of the three profit centers.  The superintendents were briefed on the objective and purpose of the work sampling study before defining the work categories.  The goal of the study was to determine the amount of time spent by the trade workers performing productive, supportive, and non-productive work and to identify situations unflattering to productivity.  Therefore, defining work categories is an important part of the work sampling process. This is a large undertaking as there are many work activities performed on a construction site at any given moment. The challenge of defining work categories is to be detailed enough to incorporate all of these anticipated site work activities, but broad enough so the quantity of work activities is less cumbersome for the observer. Furthermore, these definitions must be clear and concise to ensure the ease of consistent data recording throughout the study. 

The following is the list of work categories and definitions that were developed during the meeting.  The examples listed below were common activities observed during the work sampling study:

Productive Work:

This category includes the installation and fabrication of material such as fireproofing – spraying material, waterproofing – caulking, tooling, and scraping and insulation – placing insulation, taping, placing hat channel, and spraying fire stop caulk.

Supportive work:

This category includes any work that directly contributes to the placement of materials such as receiving or providing instructions, fireproofing – loading machine, waterproofing – refilling caulk gun, insulation – cutting material, equipment mobilization, material handling, cleaning tools, sweeping, and safety procedures.

Non-productive work:

This category includes any work that is non-productive such as walking empty handed, waiting, unscheduled breaks, late starts, early finishes, no contact or failure to observe worker who is assigned to specific work location, and handling material or tools not within immediate vicinity of its final use.

Each day the observer pre-planned multiple tour routes and different starting points.  The observer also started the work observation tours at random times throughout the day.  The tours needed to begin at random times and at different starting points to ensure jobsite workers did not anticipate the exact moment the observer would tour that particular work area.  This will provide a more accurate picture of the trade worker’s use of time.

The construction drawings were used to determine where work was to be performed over the duration of the study.  Once the work locations were identified several routes were determined to adequately cover the areas where the work was being performed.  It took between fifteen to twenty minutes to complete one work observation tour.  The observer randomly chose a duration of zero to forty-five minutes between tours.  If a trade worker could not be found in their work area on the first tour the observer would walk a second tour using an alternate random route covering the entire building.  Only after a second tour would the observer mark a trade worker as no contact.  No observations were made 30 minutes before or after a scheduled break due to the number of workers fireproofing and the waterproofers working on swing stages. Fireproofers can not stop spraying until the hopper of the spraying equipment is empty, otherwise the material would harden in the hose.  For this reason the fireproofing crew did not take breaks at scheduled times.  The waterproofing crews working on swing stages also took breaks only when it is convenient to stop.  With that taken into consideration the observer decided not to start a tour 30 minutes before or after scheduled breaks.

Before the observer started the first work observation tour of the day, he would check with the superintendents assigned to the hospital to find out the number of trade workers that were supposed to be on site for the day.  The observer also determined the locations where they were assigned to work.  Throughout the day the observer would check in with the superintendents to confirm the number of workers on site and their locations because the contractor had a tendency to pull workers off the job throughout the day and place them at different job site.

Each trade worker encountered on the work observation tour is considered to be one work sample.  When a trade worker was observed, the observer documented in which pre-defined work category the trade worker was involved the first instant he or she was seen by the observer.  This observation process took place over a period of six days until an adequate number of samples (505) had been recorded.

Findings

Upon completion of 30 random tours, 535 samples were recorded. Of these 535 samples, 211 were recorded as productive work or 40 percent, 177 samples were recorded as supportive or 33 percent and 147 samples were recorded as non-productive work or 27 percent (see Figure 1).

Figure 1: Overall Labor and Productivity Results

In Figure 2, the subcategories are shown with the resulting percentage of time spent on each subcategory.  The definitions for each subcategory can be found in the Appendix.

Figure 2: Overall Productivity Results for Subcategories

Samples were also recorded separately for each of the three different profit centers to determine if one profit center was being less productive than the others.  This study found that the productive work performed by the fireproofing crew was significantly less than the other two profit centers. The fireproofing crew had 43 of 154 samples recorded as productive work equaling only 28% (see Figure 3).  This is well below the productive work of the waterproofing crew’s 47% (see Figure 4) and the insulation crew’s 40% (see Figure 5).  However, the supportive work for the fireproofing crew was at 41% compared to the waterproofing crew’s 29% and the insulation crew’s 31%.  Another interesting note is that the non-productive work for all three was comparable with the fireproofing crew at 31% compared to the waterproofing crew’s 24% and the insulation crew’s 29%. 

Figure 3: Fireproofing Profit Center Labor and Productivity Results

Figure 4: Waterproofing Profit Center Labor and Productivity Results

Figure 5: Insulation Profit Center Labor and Productivity Results

According to the results of this study, the amount of time one worker spend per week on productive work is 16.0 hours per week, on supportive work is 13.2 hours per week, and on non-productive work is 10.8 hours per week (see Figure 6).  Figure 6 is based on a 40-hour week for one worker.

Figure 6:  Number of Hours Spent Per Week Per Category

It should be stressed that while the study results presented here accurately reflect the productivity conditions on the hospital site, one should not arbitrarily apply these results to any other jobsite.  The site-specific factors that affect productivity will differ between jobsites located in different regions and countries.

Recommendations

After reviewing the data collected, the observer determined that the contractor is operating with above average labor productivity.  This is based on other similar work sampling studies conducted by O’Brien (32 percent), Borcherding (34 percent), Orth and Jenkins (29 percent).  It is also common knowledge that labor productivity for the construction industry is approximately 30 percent.  The authors believe that contractor’s productivity is high due to these trades, fireproofers, insulators, and waterproofers, having a complete and defined scope of work, meaning they are usually following another trade such as a pipefitter or steel worker.

Although the contractor performed with a comparatively high level of productivity, there was one profit center that stood out from the others. The fireproofing crew’s productive work percentage fell well below the other profit centers. The fireproofers were only at 28% compared to the others’ 47% and 40%.  This suggests that the fireproofing crew may need to be restructured.  If two fireproofing crews worked on a site simultaneously and shared two crewmembers; one for loading material into the machines’ hoppers and another helping the two sprayers, there should be a decrease in idle time and an increase in supportive time. This would in turn increase the crew’s productivity. However, the crew’s productive work may not be affected by a change in crew size or structure.  

This study also found that the fireproofing crew spent 27% percent of their time idle. This was particularly true of the fireproofing crew applying scrim to the pencil joists.  It is the opinion of the observer that the idle time was due to a lack of supervision, since a supervisor is only on the job for a short amount of time each day.  When the workers arrived to place the scrim, they were confused as to how and where to apply it.  This confusion continued throughout the remainder of the study.  Compounding the problem is that most of the foremen speak English as a second language.  Also, all of the fireproofing crews’ foremen are also the sprayers, making it impossible to direct and keep track of their crew members while spraying.

Further work sampling studies need to be conducted to determine if the hospital job was truly above average for the contractor with respect to productivity for several reasons.  The first reason is that no observations were made 30 minutes before or after a scheduled break due to the number of workers doing fireproofing and the waterproofers working on swing stages.  A system needs to be developed so early finishes and late starts for scheduled breaks are not omitted from the study. A second reason is that the superintendents did not always provide accurate details pertaining to the number of workers expected to be on site and their work locations.  This may have contributed to errors or omissions of trade workers during the work observation tours.  It was also difficult to communicate with the craft workers due to the fact that most of the workforce had limited English speaking skills.  A third reason is that management did not give adequate time or resources to properly conduct the study.  More samples should have been obtained to result in better accuracy.

Conclusion

The objective of this study was to develop a baseline of labor productivity for the three profit centers:  insulation, waterproofing, and fire proofing.  The contractor also wanted to know if a particular profit center required more immediate attention due to lower direct work percentages.  The fireproofing profit center did have a considerable lower productivity then the other two profit centers.

When compared to other work sampling studies, the contractor’s labor productivity is relatively high.  However, most work sampling studies done in the past focused on the operations of general contractors working on large industrial projects such as power or pharmaceutical plants. This high level of direct work could be attributed to the fact that the contractor is a specialty contractor and its workers generally have a defined scope of work that will not change. This focused approach allows the workers to become intimately familiar with how to perform their work tasks as efficiently as possible.

The length of time required to obtain the desired number of samples for the study will vary among different jobsites.   It is suggested that the observers practice and time the routes to be used before beginning the study. Therefore, modifications can be made to adjust these routes as needed without wasting any actual study time. However, it is suggested that observers coordinate their tour paths with site supervisory personnel on a daily basis to ensure their workers will be easily located.

It was important that this study and its intentions were well publicized to the craft workers to avoid adverse reactions resulting from misunderstandings. “Backlashes resulting from the misinterpretation, misunderstanding, and misuse of work sampling data can lead to serious personnel conflicts within the construction organization itself (Thomas & Daily, 1983).” A “dry run” is typically performed on work sampling studies to familiarize the workers with having someone observing their actions. This was not necessary the case in this study due to the fact that the observer had spent time in the field prior to the study making observations to get familiarized with the nature of the contractor’s work.

Each jobsite has site-specific factors that affect worker productivity.  Management must make educated decisions in order to control these factors on their particular jobsite.  One simple, but effective method available for evaluating productivity problems on the jobsite is through the use of a work sampling study.  After a work sampling study is conducted, one will be able to determine the amount of time spent productively or wasted by jobsite workers.  Using this data, managers will be able to make educated decisions to govern the factors that positively and adversely affect jobsite productivity. The guidelines discussed in this paper should be followed to create and implement a work sampling study similar to the successful research performed on this project. Additionally, not only will the information obtained from the initial research be used to correct current productivity problems, it will also serve as a baseline to compare the results of future work sampling studies to be performed.

References

Adrian, J. J., & Adrian, D. J (1995). Total productivity and quality management for construction. Champaign, IL: Stipes Publishing.

Jenkins, J. & Orth, D. (2003). Productivity Improvement through Work Sampling. AACE International, Transactions of the Annual Meeting. CSC.5.1-CSC.05.7.

Liou, F. S., and Borcherding, J. D. (1985). Work sampling can predict unit rate productivity. Journal of Construction Engineering and Management, 112 (1), 90-103.

Noor, I. (1998). Measuring Construction Labor Productivity by Daily Visits. AACE International. Transactions of the Annual Meeting. 5.1-5.6.

Orth, D., & Jenkins, J. (2003). Mechanical and General Construction Productivity Results. AACE International Transactions of the Annual Meeting. CSC.07.1-CSC.07.5.

O’Brien, K. E. (2001). Improvement of on-site productivity.  Toronto, Canada: K. E. O’Brien & Associates, Inc.

Thomas, H.R., & Daily, J. (1983). Crew Performance Measurement Via activity Sampling. Journal of Construction Engineering and Management, 109, (3), 309-320.

Thomas, H.R., & Napolitan, C. (1999). Quantitative Effects of Construction Changes on Labor Productivity. Journal of Construction Engineering and Management, 124, (2), 290-296.

Appendix

Work Sampling Categories and Subcategories Definitions