Diploma in Business Management

This chapter teaches you how analyze the reliability of 
compensation surveys, as well as how to conduct a survey for your organization.

Examining how an organization designs and conducts its own survey provides information useful in appraising surveys made by others. The steps involved may be outlined as follows:

A. Plan the survey

Determine the purpose of the survey
Determine the jobs to include
Determine the markets to survey
Determine the firms to survey
Determine the information to be obtained
Make the schedules
Determine the survey method

B. Conduct the survey

Collect information
Insure job comparability

C. Tabulate, analyze, and present results

Even when an organization decides to conduct its own survey there is much to be said for attempting to interest other organizations in co-sponsoring the effort. If this is possible, the costs may be shared. The joint effort may also result in an informal group of firms that will see the advantage of periodic surveys.

Whether or not other organizations accept co-sponsorship, a steering committee of some of the firms certain to be included is a good idea. These committee members will be helpful in planning the survey and in securing cooperation.

This will be useful in determining the jobs, markets, and firms to be included and the information to be obtained. It also determines the accuracy needed and the time limits of the survey. Obviously, if information is needed on only one or two jobs or an overtime policy, a much less elaborate survey is called for than if a picture of an area or industry market is sought.

The jobs on which pay data are sought must be selected. For a number of reasons, compensation surveys do not attempt to obtain information on all of an organization's jobs. First, some jobs are unique to the organization and unlikely to be found elsewhere. Second, many jobs are always filled from the internal labor market and market data may not be needed. Third, compensation policy and practice in most organizations are based on a limited number of key jobs.

Key jobs are reference points in the job structure. In most wage and salary surveys, a limited number of them (25 to 30) are chosen to represent the entire range of jobs. It should be noted, however, that organizations placing more emphasis on external competitiveness tend to survey a larger number of jobs, sometimes as many as 80, if possible.

Key jobs are selected on a number of criteria. They should represent the entire range of jobs in the organization or as determined in the purpose of the survey. They should be numerically important in terms of surveyed firms having them and in terms of organizational positions. They should be readily definable. They should be relatively stable in content. They should represent good reference points in the job structure in terms of difficulty and responsibility. They should be well known to managers and labor leaders. They should be jobs that at least some organizations fill from external sources.

Using these criteria, the organization or the steering committee will choose the jobs to be surveyed. Obviously, jobs involving recruitment or turnover problems will become key jobs in the survey.

The next step is to determine the relevant labor markets for these jobs. Both the normal recruiting area for the jobs concerned and the area within which employees have been lost to competitors are apropos. The relevant labor-market area is the area where the labor supply for the jobs is most likely to be found.

The labor supply for some jobs is local, for others regional, and for still others national or international. Most companies confine their wage and salary surveys to local or regional markets. Surveys containing regional, national, or international data are usually secured from trade or professional organizations or from consultants.

Most white-collar and blue-collar employees are hired locally, and the labor-market area is defined as normal commuting distance. For these employees, normal commuting distance, if in question, can usually be determined from the addresses of employees presently filling these jobs.

Technical, administrative, and professional employees may be attached to the local area. But often they come from and go to other areas of the state or even adjacent states. Some managers and professionals operate in national or even international markets. Obviously, local surveys are of limited usefulness for these jobs.

Given the geographic area of the survey, the next step is to choose the organizations to be surveyed. An attempt is first made to obtain firms in the same industry. Such organizations are more likely to have all of the jobs of the surveying organization and usually have similar wages.

Organizations hiring similar skills is another criterion. An attempt is made to include firms hiring substantial numbers of employees for jobs being surveyed.

A third consideration is organization size. Compensation varies by company size. Ideally, the survey will include a balance of organizations of varying size. But very small firms may represent non-comparable labor markets.

If possible, organizations selected should have formal compensation systems administered by a compensation professional. Even better would be organizations using pay systems similar to that of the surveying organization.

The number of organizations included in the survey is very much a function of the purpose of the survey. If only a limited number of firms are adjudged to represent an organization's labor-market competition, including only these firms makes sense. If, however, the objective is to achieve a realistic picture of the labor market in the area (or industry), a representative and balanced sample is called for. The latter approach would require taking a census of establishments in the universe, and then drawing a sample. The Bureau of Labor Statistics follows this approach.

In most surveys made by companies or groups of companies, the participants are usually chosen on the criteria of (1) industry, (2) comparable work, (3) competition for workers and (4) size. In most private-company surveys, the number of cooperating establishments is limited to a small number on the assumption that these organizations represent the pertinent labor-market competitors.

This determination is driven by the purpose of the survey. Although it is possible to obtain information on pay and benefits as well as pay policies in one survey, the need for cooperation from other organizations argues against such an approach. A better practice is to make separate surveys of wages and benefits, and perhaps of average wage changes.

But compensation surveys require some information on pay policies and practices to permit interpretation of compensation data; they also require sufficient organization-identification data to permit decisions on organization comparability. The latter information consists of location, industry, size (usually number of employees), and union or nonunion status. Information sought on wage and salary policy and procedures may be limited to general increases, wage structure adjustment, and individual pay change. Alternatively, information on job evaluation plans, number of wage structures, incentive-plan usage, hours and work schedules, and general benefit policy may be sought.

The major information contained in compensation surveys is pay information on specific jobs. Here, the major requirement besides job descriptions or job briefs is careful definitions of the data sought. For example, base rates for regularly employed day workers are defined as not including overtime, shift differentials, or nonproduction bonuses, but including cost-of-living increases. These rates should be reported prior to deductions for Social Security, other taxes, and any employee contributions to benefits.

Directions for calculating base rates for employees on incentive or salary should be specified. Incentive base rates are obtained by dividing earnings by hours worked, increasing the hours to remove overtime premiums. Monthly or weekly salary is reduced to a common period.

Hiring rates should reflect the starting wage rates for particular jobs and classes of employees. Minimum, standard, and maximum rates are usually sought. Actual wage rates may be sought, together with the number of incumbents at each. The actual minimum and maximum paid for the job are usually obtained.

Benefit surveys are usually conducted separately. The information sought is quite detailed but usually applies to all employees or broad employee groups. The purpose is usually to determine the prevalence of benefits and benefit practices rather than costs, but some benefit surveys attempt to obtain the latter as well.

Some surveys are concerned only with wage changes since the last survey. Thus they carefully specify the method of calculating the information sought.

After it has been decided what information to seek, the next step is making the schedules to be used in collecting the data. The schedules are designed to permit recording the data as conveniently as possible and to permit rapid tabulation and analysis.

Normally two schedules are prepared: one for information about the organization and its compensation policy and procedures, another for compensation data. Only one of the former is required per organization. But one wage-data schedule is required for each job included in the survey. A common practice is to present the job description or job brief in the schedule prior to specifying the wage information sought. Exhibits 11-1 and 11-2 are examples of such schedules. Obviously, both schedules could be shortened or lengthened depending upon the information called for.

Exhibit 11-1. General Information: Policies and Practices Affecting Compensation

Wage and salary surveys may be made by telephone, mail, visit or group meeting. Telephone surveys may be used to acquire some compensation policy and practice data and wage information on a very few jobs when both the caller and the respondent are well acquainted compensation professionals. Such surveys may be common practice in well-defined industry groups where job comparability has been recently established or can be easily accomplished. But outside such informal networks such surveys can be unproductive or dangerous.

Mailed surveys may be safely used for securing compensation policy and practice data. But although this is the most common survey method, assuring job comparability and thus useful compensation data is a continuing problem. Careful construction of schedules and precise definitions help. But the quality of the data always depends on the qualifications of the respondent and the general press of business. As in the case of telephone surveys among professionals in the same industry, reliable data may be secured. In fact, organizations surveying a widely dispersed industry may have no other choice. But local surveys, unless job comparability by some other method has recently been assured, should probably not be done by mail.

It is usually agreed that the most productive method of gathering accurate compensation data is a visit by qualified personnel. Because determining job comparability is the key to obtaining useful occupational-compensation data, the visit is probably best conducted by trained personnel using a standardized data-collection method on site, at least in the original survey. These people can compare the survey job descriptions with company job descriptions. They can question compensation professionals. They may talk to job incumbents. They may apply a standard job evaluation scale. They collect compensation data only after personally verifying job comparability. They interview the proper person to obtain information on wage policies and procedures.

In the late 1990s, the Internet began to have its effect as a natural medium through which surveys might be conducted. The common thread that dissuades organizations from participating in surveys is the staff time required to complete survey input. The Internet brings computer technology and survey software into the office. Survey questionnaires are built in HTML and posted on the Internet for use in the collection of data. Most surveys still collect this data in the form of fixed length files where data is reviewed both statistically and visually. Some of the newer surveys check inputted data. If it falls within a preset standard error (range), the data is automatically integrated into the survey. The result is that Internet based surveys can be interactive, real-time, and reflect data gathered up to the minute.

Some compensation surveys, usually local ones, are made in group meetings. The job descriptions or job briefs of survey jobs are distributed in advance. At the meeting compensation professionals thrash out problems of job comparability and provide the surveying organization with compensation data. Although this method seems superior to mailed surveys and is much less costly than visit surveys, some questions about it can be raised. For example, in large organizations with hundreds of jobs, can even the most competent professional answer questions of comparability with survey jobs in the meeting? Also, does high turnover among compensation professionals have any effect on the reliability of the data?

From this discussion of survey methods it should be obvious that determining job comparability is the Achilles' heel of compensation surveys. Job comparability can be improved by personal visits, group meetings, use of common job evaluation plans, and use of job-scope data. But cost and coverage requirements probably ensure that many surveys will continue to be made by mail. Does this mean that survey reliability and validity depend on the conscientiousness and time pressures of respondents? K.E. Foster argues that developing Z-scores from the range midpoints of benchmark jobs and the mean and standard deviation of the company's overall range-midpoint distribution provides a useful measure of job comparability among reporting organizations. These job-value indexes JVI) can be supplemented by policy-capturing statistical techniques (such as multiple-regression analysis) and then used to explain how and why different firms value key jobs differently. Measures such as a JVI would permit much quicker and easier determination of job comparability than would use of a common job evaluation plan.

The survey method in large part determines the steps in conducting the survey. Regardless of the method, however, it is useful to separate the initial contact from the collection of the data. The initial contact is made to ensure the participation of the firms selected for the survey. It usually involves providing information on (1) the purpose of the survey, (2) the jobs covered, (3) the information sought, (4) the survey method, and (5) the report of the results. Participation is invited. Confidentiality of the data is assured. A copy of the report is offered. The latter is usually enough to secure cooperation if the potential participant has a use for the information. Participating in a survey is much less costly than conducting your own.

In mailed surveys obtaining compensation policy and practice information, determining job comparability, and obtaining compensation data depend largely on the schedules developed. Constructing the schedules with ease of completion in mind pays off in rate of return and ease of tabulating results.

In other survey methods the compensation policy and practice schedule can be mailed back to the surveying organization without loss of accuracy. In visit surveys, they are usually completed before the visit is made.

In both the group meeting and the visit survey job, comparability is determined by discussion between the representative of the surveying organization and an organization�s compensation professional. In the latter method, compensation data may be secured from the payroll department by the surveying company's representative.

When the data from all cooperating organizations are in, they are tabulated, summarized, and presented in the form of results. All information is edited for comparable terminology and units before tabulation. Consolidating the data involves separate tabulations for compensation data and compensation policy and practice data.

There are several ways to tabulate and present compensation data, depending on the purpose of the survey. Data summaries in surveys focusing on separate jobs, as is usually the case, differ from summaries in surveys focusing on the job structure.

Data summaries by job. In most surveys, data are presented separately for each job. Table 11-3 illustrates the format for this approach.

One of the advantages of this method is that it permits comparisons with other is that it permits comparisons with other companies individually or collectively. Organizations may, if they wish, exchange code numbers and custom-build their comparisons. The method also has the advantage of presenting several measures of actual rates and ranges in one tabulation.

Frequency distributions. Somewhat less common is a frequency distribution, presenting the total number of incumbents at each wage rate or class for each job. This is the method used by the Bureau of Labor Statistics in some of its surveys. See table 11-4.

This method has the advantage of presenting sufficient detail to permit the user to calculate several summary measures. But it has the disadvantage of lacking means of identifying specific company data. Other information, such as hiring rates or ranges, can be tabulated in the same way.

Graphs. If the major purpose of the survey is to compare job structures rather than individual jobs, graphs may be used. In this approach, a graph depicts a summary wage line (made up by consolidating all the information in the survey) and a company wage line. The chief advantage of this method is that each organization can compare its wage level and structure with the summary wage line and the wage line of each participant. A disadvantage is the averaging necessary to develop the graph.

Constructing such graphs is facilitated when the participating organizations are using the same job evaluation plan. But the method does not depend upon this. Any consistent method of plotting the jobs on the horizontal axis will do.

Many variations are also possible in the tabulation of compensation policy and practice data. They may be presented in narrative form. They may be tabulated question by question following the schedule used. For maximum usefulness of wage-survey data it is advantageous to use company codes for compensation policy and practice data. In this way unusual wage positions may be interpreted by reference to compensation policy and practice information.

In the post 1999 era, the presentation of survey data takes on special importance with the needed addition of "rate of error." In 1975, the U.S. Congress passed Federal Rule of Evidence 702 so that a threshold standard for the admission of testimony was established for federal courts. Based on the concept that experts should use methodologies that are "generally accepted" by a discipline�s practitioners, the rule states: "If scientific, technical, or other specialized knowledge will assist the trier of fact to understand the evidence or to determine a fact in issue, a witness qualified as an expert by knowledge, skill, experience, training, or education, may testify thereto in the form of an opinion or otherwise." Following this, the Supreme Court issued an opinion in Daubert v. Merrill-Dow Pharmaceuticals, 509 U.S. 579, 113 S. Ct. 2786, 125 L.Ed.2d 469 (1993) that has become the standard for the admission of "general acceptance." This standard is now adopted by Federal and most State Courts, and where the admittance of expert witness testimony and evidence requires the following two-step analysis. Evidence must be relevant and reliable.

The "relevance" is a subjective judgment and simple logic may be applied (e.g., salary survey data for use in wage analyses). For "reliability" the Court established four separate, nonexclusive tests:

In March 1999, the United States Supreme Court issued a ruling in the Carmichael vs. Kumho Tire Co. case that further defined when a Daubert reliability challenge applies. In Carmichael, the Supreme Court ruled that reliability must be established in all types of expert testimony, both scientific and nonscientific/technical. The Court held that the role of a trial judge was that of "gatekeeper" regarding both the relevance and reliability of all expert testimony. The Court stated that the Daubert case was not intended to be limited to scientific cases only. Instead, it would/should apply to all fields of expert testimony, including those who utilize surveys.

Developing, analyzing, and using compensation surveys may be the most common practice in compensation administration. Organizations use compensation surveys as the measurement of the labor market.

Organizations may choose to use compensation surveys developed by others or to participate in groups that develop wage information. These methods cut down on the time expended and may provide the organization with a broader range of information. However, the question of comparability becomes greater with surveys developed by others.

Given the importance and common use of compensation surveys, planning their design and use is of utmost importance. Developing a compensation survey consists of making a series of decisions on:

The major issue in compensation surveys is the comparability of the organization's jobs to those being surveyed. It is difficult to ensure this comparability in any way other than analyzing the jobs in both organizations.

While the process of collecting data through compensation surveys is well defined, the analysis of the data is done in a variety of ways. Every organization uses its own analysis to answer its particular compensation question.

Finally, the Internet is revolutionizing the field of compensation. As general employees become more knowledgeable about what the labor market pays, it is more important than ever that Human Resources professionals are well-versed in compensation survey data.

		Diploma in Human Resource Management