Job Analysis

By Milkovich, G.T., Newman, J.M., Gerhart, B.

Edited by Paul Ducham


Potential uses for job analysis have been suggested for every major personnel function. Often the type of job analysis data needed varies by function. For example, identifying the skills and experience required to perform the work clarifies hiring and promotion standards and identifies training needs. In performance evaluation, both employees and supervisors look to the required behaviors and results expected in a job to help assess performance. IBM recently identified every role (490 in all) performed by its 300,000-plus workers, managers, and executives. For example, IBM’s vice president for learning has the roles of learning leader and manager. IBM also measures and monitors 4,000 skill sets.

An internal structure based on job-related information provides both managers and employees a work-related rationale for pay differences. Employees who understand this rationale can see where their work fits into the bigger picture and can direct their behavior toward organization objectives. Job analysis data also help managers defend their decisions when challenged.

In compensation, job analysis has two critical uses: (1) It establishes similarities and differences in the work contents of the jobs, and (2) it helps establish an internally fair and aligned job structure. If jobs have equal content, then in all likelihood the pay established for them will be equal (unless they are in different geographies). If, on the other hand, the job content differs, then the differences, along with the market rates paid by competitors, are part of the rationale for paying jobs differently.

The key issue for compensation decision makers is still to ensure that the data collected are useful and acceptable to the employees and managers involved. As the arrows in Exhibit 4.3 indicate, collecting job information is only an interim step, not an end in itself.



Job titles, departments, the number of people who hold the job, and whether it is exempt from the Fair Labor Standards Act are all examples of information that identifies a job.

While a job title may seem pretty straightforward, it may not be. An observer of the U.S. banking system commented that “every employee over 25 seems to be a vice president.” A study accuses the U.S. government of creating more new job titles in a recent 6-year period than in the preceding 30 years. Some of the newer positions include deputy to the deputy secretary, principal assistant deputy undersecretary, and associate principal deputy assistant secretary. Most of these titles were created at the highest levels of government service, often to attract a specific person with unique skills. Our personal favorite is at the Jet Propulsion Laboratory in Pasadena, California, where the head of the Interplanetary Network Directorate is, naturally, the director of the Directorate. On the other hand, your tax dollars are paying the wages of 484 deputy assistant secretaries, 148 associate assistant secretaries, 220 assistant assistant secretaries, and 82 deputy assistant assistant secretaries.


This is the heart of job analysis. Job content data involve the elemental tasks or units of work, with emphasis on the purpose of each task. An excerpt from a job analysis questionnaire that collects task data is shown in Exhibit 4.7 The inventory describes the job aspect of communication in terms of actual tasks, such as “read technical publications” and “consult with co-workers.” The inventory takes eight items to cover “obtain technical information” and another seven for “exchange technical information.” In fact, the task inventory from which the exhibit is excerpted contains 250 items and covers only systems and analyst jobs. New task-based questions need to be designed for each new set of jobs.

In addition to the emphasis on the task, the other distinguishing characteristic of the inventory in the exhibit is the emphasis on the objective of the task, for example, “read technical publications to keep current on industry” and “consult with co-workers to exchange ideas and techniques.” Task data reveal the actual work performed and its purpose or outcome.



We can look at the kinds of behaviors that will result in the outcomes. Exhibit 4.6 categorizes employee data as employee characteristics, internal relationships, and external relationships. Exhibit 4.8 shows how communication can be described with verbs (e.g., negotiating, persuading). The verbs chosen are related to the employee characteristic being identified (e.g., bargaining skills, interpersonal skills). The rest of the statement helps iden tify whether the behavior involves an internal or external relationship. So both Exhibit 4.7 and Exhibit 4.8 focus on communication, but they come at it with different approaches.

The excerpt in Exhibit 4.8 is from the position analysis questionnaire (PAQ) , which groups work information into seven basic factors: information input, Mental processes, work output, relationships with other persons, job context, other job characteristics, and general dimensions. Similarities and differences among jobs are described in terms of these seven factors, rather than in terms of specific aspects unique to each job. The communication behavior in this exhibit is part of the relationships-with-other-persons factor.

The entire PAQ consists of 194 items. Its developers claim that these items are sufficient to analyze any job. However, you can see by the exhibit that the reading level is quite high. A large proportion of employees need help to get through the whole thing.

Another, more nuanced view of “communication” focuses on the nature of the interactions required plus knowledge underlying them. Interactions are defined as the knowledge and behaviors involved in searching, monitoring, and coordinating required to do the work. Some interactions are transactional—routine; “do it by the book.” The nine steps of a McFry job, shown in Exhibit 4.9 , seem transactional to us. Other interactions are more tacit—complex and ambiguous. Work content that involves more tacit interactions is believed to add greater value than more transactional tasks.

The content of communications that occurs between the Merrill Lynch financial advisor and a client to complete a stock transaction differs substantively from that between a Merrill Lynch senior vice president investor and client who aims to build a long-term relationship to manage a client’s $10 million in assets. Communication in both settings includes interactions with clients, but “building long-term relationships” versus “complete transactions” reveals substantive differences in content.

However appealing it may be to rationalize job analysis as the foundation of all HR decisions, collecting all of this information for so many different purposes is very expensive. In addition, the resulting information may be too generalized for any single purpose, including compensation. If the information is to be used for multiple purposes, the analyst must be sure that the information collected is accurate and sufficient for each use. Trying to be all things to all people often results in being nothing to everyone.


The McFry Nine-Step Program

1. Open a bag of fries.

2. Fill basket about half full (at McDonald’s, a machine does this step because we humans might make a mistake. At most places, the task is manual.)

3. Place basket in deep fryer.

4. Push timer button to track cooking time.

5. Play Pavlov’s dog—remove basket from fryer when buzzer rings and tip so fries go into holding tray. Be careful; this takes two hands, and hot grease can be flying about. Don’t spill even a drop of grease on the floor or you will be skating—not walking—in it for the rest of the day.

6. Salt fries.

7. Push another button that signals when 7 minutes are up, the “suggested holding time” for fries.

8. Check screen for size fries requested on next order.

9. Fill the corresponding fry container with fries and place in holding bin.



In addition to the job description having sections that identify, describe, and define the job, the Americans With Disabilities Act (ADA) requires that essential elements of a job—those that cannot be reassigned to other workers—must be specified for jobs covered by the legislation. If a job applicant can perform these essential elements, it is assumed that the applicant can perform the job. After that, reasonable accommodations must be made to enable an otherwise-qualified handicapped person to perform those elements.

ADA regulations state that “essential functions refers to the fundamental job duties of the employment position the individual with a disability holds or desires.” The difficulty of specifying essential elements varies with the discretion in the job and with the stability of the job. Technology changes tend to make some tasks easier for all people, including those with disabilities, by reducing the physical strength or mobility required to do them. Unfortunately, employment rates for people with disabilities are still low.

The law does not make any allowances for special pay rates or special benefits for people with disabilities. Say a company subsidizes paid parking for its employees. An employee who does not drive because of a disability requests that the employer provide the cash equivalent of the parking subsidy as a reasonable accommodation so that the money can be used to pay for alternative transportation.

While the law does not require any particular kind of analysis, many employers have modified the format of their job descriptions to specifically call out the essential elements. A lack of compliance places an organization at risk and ignores one of the objectives of the pay model.


The job analysis terms defined in Exhibit 4.4 are arranged in a hierarchy. The level at which an analysis begins influences whether the work is similar or dissimilar. The three jobs described in the beginning of the chapter—customer representative, engineer, account analyst—all involve use of computers, but a closer look showed that the jobs are very different. At the job-family level bookkeepers, tellers, and accounting clerks may be considered to be similar jobs, yet at the job level they are very different. An analogy might be looking at two grains of salt under a microscope versus looking at them as part of a serving of french fries. If job data suggest that jobs are similar, the jobs must be paid equally; if jobs are different, they can be paid differently.

Does this mean that the microscopic approach is best? Not necessarily. Many employers find it difficult to justify the time and expense of collecting task-level information, particularly for flexible jobs with frequently changing tasks. They may collect just enough job-level data to make comparisons in the external market for setting wages. However, the ADA’s essential-elements requirement for hiring and promotion decisions seems to require more detail than what is required for pay decisions. Designing career paths, staffing, and legal compliance may also require more detailed, finely grained information.

Using broad, generic descriptions that cover a large number of related tasks closer to the job-family level in Exhibit 4.4 is one way to increase flexibility. Two employees working in the same broadly defined jobs may be doing entirely different sets of related tasks. But for pay purposes, they may be doing work of equal value. Employees in these broadly defined jobs can switch to other tasks that fall within the same broad range without the bureaucratic burden of making job transfer requests and wage adjustments. Thus, employees can more easily be matched to changes in the work flow. Recruiter, compensation analyst, and training specialist could each be analyzed as a separate, distinct job, or could all be combined more broadly in the category “HR associate.”

Still, a countervailing view deserves consideration. A promotion to a new job title is part of the organization’s network of returns. Reducing the number of titles may reduce the opportunities to reinforce positive employee behavior. E * Trade experienced an increase in turnover after it retitled jobs. It reduced its vice presidents and directors to 85, down from around 170 before the retitling. Moving from the federal government job of assistant secretary to that of associate assistant secretary (or reverse) may be far more meaningful than people outside Washington, DC, imagine.


Many companies post a sample of job openings on their Web sites. Compare the job postings from several companies. How complete are the job descriptions included with the postings? Are “essential elements” listed? Are job titles specific or generic? Can you get any sense of a company’s culture from its job postings?

Links to fast-growing small private companies can be found via the Inc 500 link at Are there any differences in job postings between large and small companies?



The most common way to collect job information is to ask the people who are doing a job to fill out a questionnaire. Sometimes an analyst will interview the jobholders and their supervisors to be sure they understand the questions and that the information is correct. Or the analyst may observe the person at work and take notes on what is being done.

Exhibit 4.10 shows part of a job analysis questionnaire. Questions range from “Give an example of a particularly difficult problem that you face in your work. Why does it occur? How often does it occur? What special skills and/or resources are needed to solve this difficult problem?” to “What is the nature of any contact you have with individuals or companies in countries other than the United States?” These examples are drawn from the Complexity of Duties section of a job analysis questionnaire used by 3M. Other sections of the questionnaire are Skills/Knowledge Applied (19 to choose from), Impact This Job Has on 3M’s Business, and Working Conditions. It concludes by asking respondents how well they feel the questionnaire has captured their particular job.

The advantage of conventional questionnaires and interviews is that the involvement of employees increases their understanding of the process. However, the results are only as good as the people involved. If important aspects of a job are omitted, or if the jobholders themselves either do not realize or are unable to express the importance of certain aspects, the resulting job descriptions will be faulty. If you look at the number of jobs in an organization, you can see the difficulty in expecting a single analyst to understand all the different types of work and the importance of certain job aspects. Different people have different perceptions, which may result in differences in interpretation or emphasis. The whole process is open to bias and favoritism. As a result of this potential subjectivity, as well as the huge amount of time the process takes, conventional methods have given way to more quantitative (and systematic) data collection.



Increasingly, employees are directed to a Web site where they complete a questionnaire online.  Such an approach is characterized as quantitative job analysis (QJA) , since statistical analysis of the results is possible. Exhibits 4.7 and 4.8 are excerpts from quantitative questionnaires. In addition to facilitating statistical analysis of the results, quantitative data collection allows more data to be collected faster.

A questionnaire typically asks jobholders to assess each item in terms of whether or not that particular item is part of their job. If it is, they are asked to rate how important it is and the amount of job time spent on it. The responses can be machine-scored, similar to the process for a multiple-choice test (only there are no wrong answers), and the results can be used to develop a profile of the job. Questions are grouped around five compensable factors: knowledge, accountability, reasoning, communication, and working conditions. Knowledge is further subcategorized as range of depth, qualifications, experience, occupational skills, management skills, and learning time. Assistance is given in the form of prompting questions and a list of jobs whose holders have answered each question in a similar way. Results can be used to prepare a job profile based on the compensable factors. If more than one person is doing a particular job, results of several people in the job can be compared or averaged to develop the profile. Profiles can be compared across jobholders in both the same and different jobs.

Some consulting firms have developed quantitative inventories that can be tailored to the needs of a specific organization or to a specific family of jobs, such as data/information processing jobs. Many organizations find it practical and cost-effective to modify these existing inventories rather than to develop their own analysis from ground zero. But, remember, as we have said, the results depend on the quality of the inputs. Here, the items on the questionnaire matter. If important aspects of a job are omitted or if the jobholders themselves do not realize the importance of certain aspects, the resulting job descriptions will be faulty. In one study, the responses of high-performing stockbrokers on amounts of time spent on some tasks differed from those of low performers. The implication is that any analysis needs to include good performers to ensure that the work is usefully analyzed.


Collecting job analysis information through one-on-one interviews can be a thankless task. No matter how good a job you do, some people will not be happy with the resulting job descriptions. In the past, organizations often assigned the task to a new employee, saying it would help the new employee become familiar with the jobs of the company. Today, if job analysis is performed at all, human resource generalists and supervisors do it. The analysis is best done by someone thoroughly familiar with the organization and its jobs and trained in how to do the analysis properly.


The decision on the source of the data (jobholders, supervisors, and/or analysts) hinges on how to ensure consistent, accurate, useful, and acceptable data. Expertise about the work resides with the jobholders and the supervisors; hence, they are the principal sources. For key managerial/professional jobs, supervisors “two levels above” have also been suggested as valuable sources since they may have a more strategic view of how jobs fit in the overall organization. In other instances, subordinates and employees in other jobs that interface with the job under study are also involved.

The number of incumbents per job from which to collect data probably varies with the stability of the job, as well as the ease of collecting the information. An illdefined or changing job will require either the involvement of more respondents or a more careful selection of respondents. Obviously, the more people involved, the more time-consuming and expensive the process, although computerization helps mitigate these drawbacks.

Whether through a conventional analysis or a quantitative approach, completing a questionnaire requires considerable involvement by employees and supervisors. Involvement can increase their understanding of the process, thereby increasing the likelihood that the results of the analysis will be acceptable. But it also is expensive.


What happens if the supervisor and the employees present different pictures of the jobs? While supervisors, in theory, ought to know the jobs well, they may not, particularly if jobs are changing. People actually working in a job may change it. They may find ways to do things more efficiently, or they may not have realized that certain tasks were supposed to be part of their jobs.

3M had an interesting problem when it collected job information from a group of engineers. The engineers listed a number of responsibilities that they viewed as part of their jobs; however, the manager realized that those responsibilities actually belonged to a higher level of work. The engineers had enlarged their jobs beyond what they were being paid to do. No one wanted to tell these highly productive employees to slack off. Instead, 3M looked for additional ways to reward these engineers rather than bureaucratize them.

What should the manager do if employees and their supervisors do not agree on what is part of the job? Differences in job data may arise among the jobholders as well. Some may see the job one way, some another. The best answer is to collect more data. Enough data are required to ensure consistent, accurate, useful, and acceptable results. Holding a meeting of multiple jobholders and supervisors in a focus group to discuss discrepancies and then asking both employees and supervisors to sign off on the revised results helps ensure agreement on, or at least understanding of, the results. Disagreements can be an opportunity to clarify expectations, learn about better ways to do the job, and document how the job is actually performed. Discrepancies among employees may even reveal that more than one job has been lumped under the same job title.

Top Management (and Union) Support Is Critical

In addition to involvement by analysts, jobholders, and their supervisors, support of top management is absolutely essential. Support of union officials in a unionized workforce is as well. They know (hopefully) what is strategically relevant. They must be alerted to the cost of a thorough job analysis, its time-consuming nature, and the fact that changes will be involved. For example, jobs may be combined; pay rates may be adjusted. If top managers (and unions) are not willing to seriously consider any changes suggested by job analysis, the process is probably not worth the bother and expense.


To avoid starting from scratch (if writing a job description for the first time) or as a way to cross-check externally, it can be useful to refer to generic job descriptions that have not yet been tailored to a specific organization. One readily accessble source is the Occupational Information Network, or O*NET ( ). Exhibit 4.11 shows the information O*NET provides using the job of computer programmer as an example.


Use O*NET to find the knowledge, skills, and other characteristics needed to be a computer programmer (or an occupation of your choice).

Go to

Choose: Find Occupations

Enter the occupation name into space under “Keyword or O*NET-SOC code”

Click on “go”

Then click on the name of the occupation to see the knowledge, skills, etc. required.

Would this information from O*NET be useful to you if you needed to write job descriptions in your organizations?



Descriptions of managerial/professional jobs often include more-detailed information on the nature of the job, its scope, and accountability. One challenge is that an individual manager will influence the job content. Professional/managerial job descriptions must capture the relationship between the job, the person performing it, and the organization objectives—how the job fits into the organization, the results expected, and what the person performing it brings to the job. Someone with strong information systems and finance expertise performing the compensation manager’s job will probably shape it differently, based on this expertise, than someone with strong negotiation and/or counseling expertise.

Exhibit 4.12 excerpts this scope and accountability information for a nurse manager. Rather than emphasizing the tasks to be done, this description focuses on the accountabilities (e.g., “responsible for the coordination, direction, implementation, evaluation, and management of personnel and services; provides leadership; participates in strategic planning and defining future direction”).


Job Description for a Manager

Title: Nurse Manager

Department: ICU

Position Description:

Under the direction of the Vice President of Patient Care Services and Directors of Patient Care Services, the Nurse Manager assumes 24-hour accountability and responsibility for the operations of defined patient specialty services. The Nurse Manager is administratively responsible for the coordination, direction, implementation, evaluation, and management of personnel and services. The Nurse Manager provides leadership in a manner consistent with the corporate mission, values, and philosophy and adheres to policies and procedures established by Saint Joseph’s Hospital and the Division of Patient Care Services. The Nurse Manager participates in strategic planning and defining future direction for the assigned areas of responsibility and the organization.


Education: Graduate of accredited school of nursing. A bachelor’s degree in Nursing or related field required. Master’s degree preferred. Current license in State of Wisconsin as a Registered Nurse, Experience: A minimum of three years’ clinical nursing is required. Minimum of two years’ management experience or equivalent preferred.


The final step in the job analysis process is to verify the accuracy of the resulting job descriptions (step 6 in Exhibit 4.5 ). Verification often involves the jobholders as well as their supervisors to determine whether the proposed job description is accurate and complete. The description is discussed, line by line, with the analyst, who makes notes of any omissions, ambiguities, or needed clarifications (an often excruciating and thankless task). It would have been interesting to hear the discussion between our nurse from 100 years ago, whose job is described in Exhibit 4.13 , and her supervisor.

The job description paints a vivid picture of expectations at that time, although we suspect the nurse probably did not have much opportunity for input regarding the accuracy of the job description.


Job Description for Nurse 100 Years Ago

In addition to caring for your 50 patients each nurse will follow these regulations:

1. Daily sweep and mop the floors of your ward, dust the patient’s furniture and window sills.

2. Maintain an even temperature in your ward by bringing in a scuttle of coal for the day’s business.

3. Light is important to observe the patient’s condition. Therefore, each day, fill kerosene lamps, clean chimneys, and trim wicks. Wash the windows once a week.

4. The nurse’s notes are important in aiding the physician’s work. Make your pens carefully, you may whittle nibs to your individual taste.

5. Each nurse on the day duty will report every day at 7 a.m. and leave at 8 p.m. except on the Sabbath on which day you will be off from 12:00 noon to 2:00 p.m.

6. Graduate nurses in good standing with the director of nurses will be given an evening off each week for courting purposes, or two evenings a week if you go regularly to church.

7. Each nurse should lay aside from each pay day a goodly sum of her earnings for her benefit during her declining years, so that she will not become a burden. For example, if you earn $30 a month you should set aside $15.

8. Any nurse who smokes, uses liquor in any form, gets her hair done at a beauty shop, or frequents dance halls will give the director good reason to suspect her worth, intentions, and integrity.

9. The nurse who performs her labors and serves her patients and doctors faithfully and without fault for a period of five years will be given an increase by the hospital administration of five cents a day, provided there are no hospital debts that are outstanding.



Offshoring refers to the movement of jobs to locations beyond a country’s borders. Historically, manual, low-skill jobs were most susceptible to offshoring. There are substantial differences in hourly compensation costs across countries for manufacturing workers; this has played an important role in companies’ decisions about where to locate production operations. Similar differences in cost in other low-skill occupations (e.g., in call centers) have had similar ramifications. (So, when you call for an airline reservation or help with your printer, you may well reach someone in another country.) Of course, as we also noted, labor cost is only part of the story. There are productivity differences across countries as well, meaning that lower labor costs may in some cases be offset by lower productivity. Availability of workers with needed education and skills is another potential constraint. Proximity to customers is yet another issue. Sometimes that argues for moving offshore, sometimes it does not.

Increasingly, susceptibility to offshoring is no longer limited to low-skill jobs. Whitecollar jobs are also increasingly at risk. Is there a way to systematically measure which jobs are most susceptible to offshoring? The U.S. Bureau of Labor Statistics has attempted to do just this with respect to service-providing occupations. Exhibit 4.14 shows the list of occupations it found to have the highest and lowest susceptibility to offshoring. The offshoring susceptibility scores are based on the sum of scores on the four items shown. So, jobs are most susceptible to outsourcing when inputs and outputs can easily be transmitted electronically, little interaction with other workers is required, little local knowledge is required, and the work can be routinized.

Interestingly, highly susceptible jobs include not only those that require little education and training, such as data entry keyers and telemarketers, but also computer programmers and tax preparers. Turning to jobs with low susceptibility to outsourcing, we see various managerial positions and also positions where local knowledge is required (e.g., marketing managers presumably need to know consumer preferences in particular regions of the world) or where being “on the ground” (literally, in the case of landscape architects) is necessary.

To our knowledge, the system for assessing susceptibility to offshoring has not been rigorously validated to see how it well it predicts actual offshoring of occupations. Nevertheless, as Exhibit 4.14 indicates, growth rates (in the United States) for jobs on the highly susceptible list are generally small or negative, while jobs on the low susceptibility list have shown strong growth. Unless the two sets of jobs have different growth rates across countries, the differential growth rates seem consistent with the possibility that jobs on the highly susceptible list have lower growth rates, at least in part because they have experienced greater offshoring. Also, there are certainly numerous examples of jobs on the highly susceptible list (e.g., data entry keyers, telemarketers, and computer programmers) being offshored.



As firms spread work across multiple countries, there is an increasing need to analyze jobs to either maintain consistency in job content or else be able to measure the ways in which jobs are similar and different. For example, for a software development team to work equally effectively with programmers in the United States and India, the job descriptions and job specifications need to be measured and understood. One potential challenge is that norms or perceptions regarding what is and what is not part of a particular job may vary across countries. However, a study of three different jobs (first-line supervisor, general office clerk, and computer programmer) in the United States, China, Hong Kong, and New Zealand found that ratings of the importance and amount of work activities and job requirements were “quite similar” across countries, suggesting that job analysis information “is likely to transport quite well across countries.”


If you measure something tomorrow and get the same results you got today, or if I measure and get the same result you did, the measurement is considered to be reliable. This doesn’t mean it is right—only that repeated measures give the same result. Reliability is a measure of the consistency of results among various analysts, various methods, various sources of data, or over time. Reliability is a necessary, but not sufficient, condition for validity .

Research on employee and supervisor agreement on the reliability of job analysis information is mixed. For instance, experience may change an employee’s perceptions about a job since the employee may have found new ways to do it or added new tasks to the job. The supervisor may not realize the extent of change. In such cases, the job the employee is actually doing may not be the same as the job originally assigned by the supervisor. Differences in performance seem to influence reliability. Other research finds that reliability is lower for jobs that are more interdependent with other jobs, and have more autonomy/are less routine. To date, no studies have found that gender and race differences affect reliability. Obviously, the way to increase reliability in a job analysis is to understand and reduce sources of difference. Quantitative job analysis helps do this. But we need to be sure that we do not eliminate the richness of responses while eliminating the differences. Sometimes there really may be more than one job. Training can also improve reliability.


Does the analysis create an accurate portrait of the work? There is almost no way of showing statistically the extent to which an analysis is accurate, particularly for complex jobs. No gold standard exists; how can we know? Consequently, validity examines the convergence of results among sources of data and methods. If several job incumbents, supervisors, and peers respond in similar ways to questionnaires, then it is more likely that the information is valid. However, a sign-off on the results does not guarantee the information’s validity. It may mean only that all involved were sick to death of the process and wanted to get rid of the analyst so they could get back to work.


If job holders and managers are dissatisfied with the initial data collected and the process, they are not likely to buy into the resulting job structure or the pay rates attached to that structure. An analyst collecting information through one-on-one interviews or observation is not always accepted because of the potential for subjectivity and favoritism. One writer says, “We all know the classic procedures. One [worker] watched and noted the actions of another . . . at work on [the] job. The actions of both are biased and the resulting information varied with the wind, especially the political wind.” However, quantitative computer-assisted approaches may also run into difficulty, especially if they give in to the temptation to collect too much information for too many purposes. After four years in development, one application ran into such severe problems due to its unwieldy size and incomprehensible questions that managers simply refused to use it.


To be valid, acceptable, and useful (see below), job information must be up to date. Some jobs stay relatively stable over time, while others may change in important ways, even over short time periods. As Exhibit 4.15 shows, most organizations report that they have up-to-date job information, but a substantial portion report that job information is not up to date. That can not only hinder compensation practice and decision-making, but also employee selection, training, and development. Most organizations do not engage in any regular (e.g., annual or biannual) updating of job analysis information, instead being more likely to update job information when the significant changes are believed to have occurred or when the job is being re-evaluated for compensation purposes. It may be useful to develop a systematic protocol for evaluating when job information needs to be updated.



Usefulness refers to the Practicality of the information collected. For pay purposes, job analysis provides work-related information to help determine how much to pay for a job—it helps determine whether the job is similar to or different from other jobs. If job analysis does this in a reliable, valid, and acceptable way and can be used to make pay decisions, then it is useful.

As we have noted, some see job analysis information as useful for multiple purposes, such as hiring and training. But multiple purposes may require more information than is required for pay decisions. The practicality of all-encompassing quantitative job analysis plans, with their relatively complex procedures and analysis, remains in doubt. Some advocates get so taken with their statistics and computers that they ignore the role that judgment must continue to play in job analysis. Dunnette’s point, made more than 25 years ago, still holds true today: “I wish to emphasize the central role played in all these procedures by human judgment. I know of no methodology, statistical technique or objective measurements that can negate the importance of, nor supplement, rational judgment.”


In the face of all the difficulties, time, expense, and dissatisfaction, why on earth would you as a manager bother with job analysis? Because work-related information is needed to determine pay, and differences in work determine pay differences. There is no satisfactory substitute that can ensure the resulting pay structure will be work-related or will provide reliable, accurate data for making and explaining pay decisions.

If work information is required, then the real issue should be, How much detail is needed to make these pay decisions? The answer is, Enough to help set individual employees’ pay, encourage continuous learning, increase the experience and skill of the work force, and minimize the risk of pay-related grievances. Omitting this detail and contributing to an incorrect and costly decision by uninformed managers can lead to unhappy employees who drive away customers with their poor service, file lawsuits, or complain about management’s inability to justify their decisions. The response to inadequate analysis ought not to be to dump the analysis; rather, the response should be to obtain a more useful analysis.