MEMORANDUM
(FINDINGS OF FACT AND CONCLUSIONS OF LAW)
DeMASCIO, District Judge.Plaintiffs filed this class action in November 1985 alleging sex based wage discrimination on behalf of all classified employees of the State of Michigan employed in predominantly female job classifications. The individual plaintiffs are classified employees of the State of Michigan in either the Human Services Service Group or the Clerical Service Group. Plaintiff UAW has been the certified representative of the Administrative Support Unit of the Human Services Group since November 17, 1985. Plaintiffs named as defendants the State of *894Michigan and the Michigan Civil Service Commission, the agency responsible for classifying positions in the classified civil service and fixing the rates of compensation for all such classifications.1
Plaintiffs essentially claim that defendant's early classification system, the Position Comparison System, discriminated in basic compensation paid to predominantly female classifications and that the Benchmark System implemented thereafter incorporated those discriminatory wage rates; that defendant discriminated in the design of the service groups and assignment of classes to service groups within the Benchmark System; that defendants discriminated on the basis of sex in the assignment of pay ranges to predominantly female classes; and. that defendant continues to pay discriminatory wages to incumbents in predominantly female classes.
The testimony and exhibits offered by plaintiffs traced only 3 of the state’s 2500 post-conversion classes. In addition, plaintiffs introduced statistical and circumstantial evidence to establish that the state discriminatorily assigned lower wage rates for jobs in female dominated classes. Pursuant to the factual findings and conclusions that follow, we conclude that plaintiffs’ evidence is insufficient to establish intentional discrimination and we dismiss plaintiffs’ complaint.
Defendants, the Civil Service Commission and its Department of Civil Service, have the sole authority to establish a merit classification system and determine wage rates for classified workers.2 Each individual department of the state has the authority to hire, fire, and promote their employees. Those departments are not parties to this action. Defendant established the Position Comparison System (PCS), its first civil service classification system, in 1938. Under PCS, 10 factors were evaluated to assign positions to classes and classes to levels. The Commission initially established 21 levels and 21 standard pay ranges under the PCS. By 1972, there were a total of 112 different pay ranges.
In the early 1970’s, the Commission concluded that the PCS was no longer a workable classification system.3 Mr. Heuni, the Director of the Department’s Classification Division, testified that it had become cumbersome and difficult to manage and the state was under a great deal of pressure to change it. Heuni engaged Cresap, McCormick & Paget, a consulting firm, to study the PCS. Cresap recommended adoption of a factor ranking system. In October 1972, the Commission hired the firm of Public Administration Service (PAS) to assist it in revising its classification and compensation system. With the assistance of PAS, the Commission selected a Benchmark ranking system of classification modeled after the system described in the report “Job Evaluation and Pay Review Task Force of the United States Civil Service Commission” (Oliver Report). The model consisted of System I, which included all classes that did not require a college degree, System II, which included positions that did require a college degree, and System III which included all managerial employees.4 Using the Oliver Report as a guide, the Classification and Compensation Bureau developed tentative factor guides and class specifications, and tested the system in several pilot studies. The Bureau then attempted to implement the system for all classified po*895sitions at one time, a task that proved too difficult for staff to accomplish. The results of the pilot study are reported in the Interim Project Report on Recommended Classification System. In that report, PAS concluded that the Plan was feasible and workable and a significant improvement over PCS. The Bureau continued to attempt implementation after the report, but encountered difficulty with consistency in factoring the positions. The factoring and job evaluations were being performed by committees comprised of members of defendants’ staff and employees of the individual departments. In 1974, John Heuni, then Director of the Classification and Compensation Bureau, assigned Martha Bibbs and John Hand to review the positions factored under Systems I and II to assure consistency in factoring. Bibbs and Hand testified that the major problems they encountered with the system were the inconsistencies between committees factoring similar jobs, the impracticality of attempting to implement a system for 70,000 employees at one time, staff shortages and turnover, and severe pressure to complete the project, which was now well beyond the one year target date for completion.
Late in 1974, Gordon Chamberlain, a personnel administrator in the Bureau, submitted a proposal recommending that the Benchmark System be modified by creating six occupational service groups. Mr. Heuni circulated the proposal to his staff and appointed a Select Committee to review the proposed modifications. The Committee recommended a revised Benchmark Factor Evaluation Plan comprised of several service groups, each with its own ranking system, to be implemented one by one over a period of time. The Bureau thereafter recommended such a plan to the Commission and the Commission adopted it on June 13, 1975.
The Bureau initially established 7 tentative occupational service groups. It eventually established and implemented 11 service groups over a period of 6 years. These groups for the most part parallel labor market occupational groups. Staff rewrote and revised class specifications and merged, abolished, and created classes within service groups. The process of assigning classes to groups was somewhat subjective, but intended to reflect duties and responsibilities of occupational families or separate labor markets. The factor evaluation grids were revised for each group, but all grids evaluated classes on the basis of five factors: knowledge requirements, nature of work, responsibility, personal relationships, and physical effort and work environment. Points were assigned for each of these factors. The Bureau developed guides for each service group which contained definitions, factors, grids, class specifications, and other information. Each guide was submitted to employees, unions, and the individual departments for review and comment before it was adopted. The guides were used to convert all positions from the old system to the new.
After class specification grids were completed, each of the approximately 70,000 positions in the state classified service was individually factored to determine its proper service group, classification series, and level within a service group. Point totals were used to differentiate positions within a class series and to provide an internal ranking of levels and classes within service groups. If a position did not fit within a particular service group definition, it was held for review until the next service group was implemented. Some positions were reassigned from the service group in which they were tentatively placed; these reassignments could affect the pay range assigned to the position.
At the time the Commission adopted the Bureau’s recommendation on implementation of the Benchmark Factor Ranking System, it adopted the following pay conversion policy:
For initial conversion to the new classification plan, pay ranges presently in existence will be used:
A. Criteria used for selecting the most appropriate pay range are:
(1) The same pay range will be used for all classes falling within the same band of points or grade level unless there is compelling evidence that some classes *896should be paid at a different rate, and any deviation must be based on current market conditions and will be regarded as temporary;
(2) Usually the schedule by which the predominant number of positions in a grade level are now paid will be selected;
(3) A determination will be made as to whether the pay structure is appropriate, i.e., whether there is a reasonable increase in pay from one grade level to another;
(4) Classes of employees that were emphasized in pay surveys as those which would be most likely to reflect current market conditions will be considered.
There were no written criteria for determining when deviations were appropriate and no “temporary” deviation was ever eliminated. If an employee was assigned to a pay range, the maximum of which was lower than his or her current range, that employee’s position was “restricted,” i.e., classified under the PCS and paid at the higher range until the incumbent left the position. Once the position was vacated, it was converted to the proper Benchmark class. Until then “restricted” positions received the same wage and step increases as employees who were converted to the new system. Employees were given the right to appeal the factoring and placement of their position in the system. Employees also had the right to appeal the pay range assigned to their position in the new system. When the service groups were implemented, the Bureau intended to begin a cyclical review of each service group to determine if any adjustments were necessary. These reviews were placed in abeyance in 1983 when the defendant Commission established a Comparable Worth Task Force to review the classification system.
After conversion to the Benchmark system and until 1980, wages for classified positions were set by a public meet and confer wage setting process. The process included public hearings with input from Commission staff, labor unions, and representatives of the individual departments. Between 1972 and 1976, the Compensation Advisory Board (CAB) made recommendations to the Commission for general and special wage increases. CAB was composed of four employee representatives and a three member Board composed of one member each from labor and management and an individual selected by the Governor. The CAB was replaced in May 1976 with the Compensation Hearing Panel (CHP) which held extensive hearings regarding pay. The CHP received evidence and testimony from unions, employees, management from each of the state departments, and civil service staff regarding the labor market, economic conditions, and turnover rates. During these years, the Commission generally awarded across the board increases for all classes with special increases for a limited number of classes.
The CAB and CHP and their pre-1972 counterparts (i.e., the Tripartite Committee on Pay Setting %nd the Pay Steering Committee) worked within the limits of the Commission’s pay policy. That policy required them to “recognize the state’s role as a responsible employer in establishing compensation levels that are competitive with other major employers and that enable the state to recruit and retain employees for all its services.” The various panels used the labor market survey conducted annually by Civil Service Commission staff as a tool for recommending increases. Before 1975, the annual survey included over 500 employers and 200 survey classes. Since 1975, the survey has included less than 100 employers and 50 benchmark classes. Separate surveys of public, private and other state employers are conducted. In the past, the survey was criticized by the Office of the State Employer and by some Civil Service staff members as an inadequate measure of the labor market. The defendant Commission changed the survey methodology in 1986.
In 1980, the Commission granted state employees the limited privilege to collectively bargain with respect to certain terms and conditions of employment, including wages. Although classification is a prohibited subject of bargaining, employees may bargain the pay range assigned to a classification. The governor established the Office of State Employer to represent the *897individual departments in the Executive Branch for purposes of collective bargaining. The Employment Relations Board (ERB), established in 1980, has review powers over grievances and representation disputes. The ERB also has the responsibility to maintain a classified compensation plan for all unrepresented employees. Wages presently are set by collective bargaining for all but the non-exclusively represented employees. Wages for non-represented employees continue to be set by the meet and confer process.
When the Civil Service Commission began to study the issue of pay equity in 1981, it was aware of the fact that the Benchmark implementation brought to light some pay disparities between male and female dominated jobs at the same class and level. David Hand, a former Director of the Classification & Compensation Planning Division, pointed out some of these differences in a 1979 study entitled “The State of Michigan and Sexist Discrimination,” a study he completed after leaving the Department of Civil Service. The Commission was also aware that its work force was segregated along occupational lines. In the late 1970’s, the Department of Labor established the Office of Women and Work (OWW). In 1979, that office commissioned Arthur Young & Co. to do a comparable worth study of the State of Michigan. In August 1980, OWW released a report, which critiqued the Benchmark Factor Ranking System and found that females were concentrated in lower paying jobs. A second report, issued in 1981, concluded that there was a sex based difference in pay for jobs with equal Benchmark point totals. The Arthur Young report, released in December 1981, evaluated Michigan’s classified positions on two scales, not including the Benchmark scale. The study found that predominantly female classes were paid less than predominantly male classes which scored the same based on objective job evaluation measures. The report did not reach any conclusions about intentional discrimination. Civil Service staff reviewed the report and the State Personnel Director, John Heuni, recommended that the Commission endorse the effort to achieve progress in comparable worth.
In January 1983, at the direction of the Commission, Civil Service staff began a comprehensive study of occupational segregation within the state classified service. Heuni presented a report on occupational segregation to the Commission in March of that year. The report confirmed that significant sex segregation existed within the state classified service. It further found that such segregation was a primary factor in the wage differential between men and women in the state service. Staff then began to develop a classification model that would permit an analysis across service groups for differences in pay. Joseph Sli-vensky designed a tentative comparable worth job evaluation model, but it was not developed or presented to the Commission. The model used Position Comparison Equivalency Levels (PCELs) to compare classes across service groups for pay equity analyses.
In June 1983, the Commission reviewed the results of the staffs study of comparable worth and determined that there was a need for a Comparable Worth Task Force for a complete development of the concept. The Commission created the Task Force and adopted a suitable charge on August 19, 1983. The initial charge mentioned discrimination in the classified service but the charge was later revised and the reference to discrimination was deleted. The Task Force issued its report on June 21,1985. It recommended continuing the service group approach for classification but suggested reducing the total number of service groups from 11 to 5 non-gender dominated groups. On June 18,1985, the Commission accepted the Task Force report and adopted its recommendations for immediate implementation. The redesign of the service groups, known as the Merit System Review, is still being implemented.
On January 8, 1986, the Commission approved ratification of UAW contracts containing pay equity adjustments including two 20 cent an hour pay equity increases for female dominated classes and a $2 million pay inequity fund. Similar adjust-*898mente were also approved for female dominated classes in two other bargaining unite. These adjustments will cost the State of Michigan approximately $21 million.
Notwithstanding the defendants’ positive response to conditions that developed during and after its first classification system, plaintiffs allege that the state implemented a classification system and assigned lower wage rates to jobs dominated by women to intentionally discriminate in violation of Title VII. They must proceed, of course, under a disparate treatment theory and must prove that defendants depressed the salary of incumbents in female dominated positions because they were predominantly female. County of Washington v. Gunther, 452 U.S. 161, 204, 101 S.Ct. 2242, 2265, 68 L.Ed.2d 751 (1981); American Nurses Association v. Illinois, 783 F.2d 716, 720 (7th Cir.1986). Plaintiffs also must establish that discrimination “was the [State’s] standard operating procedure— the regular rather than the unusual practice.” Bazemore v. Friday, — U.S.-, 106 S.Ct. 3000, 3008, 92 L.Ed.2d 315 (1987) (quoting Teamsters v. United States, 431 U.S. 324, 336, 97 S.Ct. 1843, 1855, 52 L.Ed. 2d 396 (1977)). Once plaintiffs establish a prima facie case, the burden shifts to the defendant to articulate a legitimate non-discriminatory reason for its action. McDonnell Douglas Corp. v. Green, 411 U.S. 792, 93 S.Ct. 1817, 36 L.Ed.2d 668 (1973).
Plaintiffs attempted to establish their prima fade case at trial through both circumstantial and statistical evidence. Each is a permissible method. See Fumco Construction Corp. v. Waters, 438 U.S. 567, 579-80, 98 S.Ct. 2943, 2950-52, 57 L.Ed.2d 957 (1978); American Federal of State County & Munidpal Employees (AFSCME) v. State of Washington, 770 F.2d 1401, 1407 (9th Cir.1985). Plaintiffs contend that an intent to discriminate was a factor in the development of the Benchmark Factor Evaluation System. They argue that defendants were aware of the occupational segregation and pay disparities in the classified system in the early 1970’s as a result of the 1971 Employment Practices Review conducted by the Civil Service Commission and the Department of Civil Rights, and that when defendants began implementing Systems I and II, they found that although certain labor and clerical positions factored out the same, the predominantly male labor positions were paid more than the predominantly female clerical positions. According to plaintiffs, defendants decided to abandon Systems I and II because they did not want laborers and clerical employees on the same pay scale.
The evidence demonstrates that this was not the reason defendants switched to the service group approach. David Hand testified that he and Martha Bibbs were concerned about the fact that the Labor and Clerical positions were factoring out the same but were paid differently. Their concern was that employees would not understand the difference in labor market conditions for the two groups. Both he and Ms. Bibbs testified that Systems I and II were replaced with service groups because of the problems encountered in trying to implement the systems. They had considerable difficulty achieving consistency in evaluation and factoring among the different committees, were far beyond their deadlines, and perceived the task of implementing an entire system at once to be almost impossible. They, and every staff member called as a witness by plaintiff or defendant, testified that the gender of the incumbents was not a factor in the decision to abandon Systems I and II. In sum, defendants realized that the PCS was outdated and unworkable and they set out to change it. They tried one approach and were not having much success with it. An alternate plan was proposed, which they felt would meet their needs, and they decided to try that approach instead. The evidence does not establish that the gender domination of any classification was a factor in the decision to adopt the Benchmark Factor Evaluation System.
Plaintiffs next contend that defendants discriminated in the implementation of Benchmark. First, they claim that the assignment of positions to service groups and classes within service groups was based in part upon the gender of incumbents. For *899example, the predominantly male class of liquor control clerks was assigned to Labor & Trades although it had clerical responsibilities. If it had been assigned to Clerical, it would have raised the wages for the predominantly female classes in that group. According to plaintiffs, defendant decided to place the liquor control clerks in Labor & Trades in order to depress the pay of women in the Clerical service group. The evidence, however, establishes that defendant was unsure of where these positions should go and sent staff out to observe incumbents. As a result of their observations, and input from the Department of Commerce, defendant concluded that the physical aspects of the job should be emphasized and the group should be placed in Labor & Trades. This decision does not evidence an intent to discriminate.
Plaintiffs further suggest that the changes in the number and formation of service groups is evidence of intentional discrimination. The Chamberlain memorandum suggested only 5 service groups. As the implementation progressed defendant expanded that number and eventually established 11 service groups. Several witnesses testified that the Chamberlain memorandum represented a concept only and was not “written in stdne.” The service groups were formulated from necessity during the implementation process as distinct groups of employees or occupational families emerged. More than the initial 5 became necessary when defendants realized that all employees simply did not fit within the 5 proposed groups or could not appropriately be grouped together. The resulting 11 groups reflect legitimate labor market occupational groupings. Plaintiffs failed to demonstrate that the gender of incumbents was a factor in the formation of particular service groups.
The third flaw that plaintiffs emphasize in the implementation of Benchmark is in the application of the pay conversion policy. The state’s policy permits deviations but does not define the circumstances under which deviations are permissible. Plaintiffs identify one incident where defendants allegedly “bent the criteria” of the policy. When implementing the Domestic Workers group, defendants discovered that the entry level Janitor position and the entry level Domestic Service Aid (DSA) position both factored out to skill level I. If the predominant rate were applied, the DSA’s would have been raised to the level of the Janitors. Defendants admittedly were concerned about the effect this change would have on the budget of the Mental Health Department, the appointing authority of the DSA’s. Defendants also had concerns about the skills of entry level janitors who were required but sometimes unable to operate technical machinery. The solution to these two problems was to raise the entry level janitor to skill level II and to require one year of experience. Mr. Slivensky testified that the one year experience requirement was added to the Janitor II class at the request of the Department of Management and Budget because of the need to hire people capable of operating expensive machinery (Tr. 433-34). While this may seem to plaintiffs to have been a somewhat artificial distinction, it does not evidence an intent to discriminate against predominantly female classes.
Plaintiffs have chosen 3 classifications of the 2,500 classes that were factored as evidence of discrimination. We find, however, that all of the positions were classified and compensated without regard to the gender of the incumbents. We are mindful that we must look at the entire system and all the changes made by defendant during these years. In view of the number of positions classified, the two or three identified by plaintiffs would, at best, appear to be isolated incidents and not the result of the state’s “standard operating procedure.” Bazemore, 106 S.Ct. at 3008.
Through Mr. Norman Willis, plaintiffs’ job evaluation expert, plaintiffs criticized Benchmark generally as a job evaluation system. He testified that the system was entirely consistent, so much so that the group classifications seemed to predetermine a job ranking. He further concluded that the grids lacked the flexibility necessary to quantify critical distinctions between different levels of complexity in a particular job. He pointed out, for exam-*900pie, that the points assigned to a clerk typist were identical to the evaluation of the position of a hearing reporter. Yet, the State of Michigan requires a hearing reporter to take live transmissions at a minimum of 200 words per minute, while the clerk typist takes recorded dictation at 40 words per minute. He further criticized the fact that 35 classes in the Human Services group factored out to 500 evaluation points and 22 others had only slight variations. He testified that when the state attempted to install the original Oliver System it had an acceptable approach to job evaluation but that “somehow in the translation they ended up throwing the baby out with the bathwater.” (Tr. 612.) Mr. Willis produced a chart (exhibit 199) to illustrate that, in its implementation of the Benchmark System, the state followed a consistent pattern that resulted in placing some classes in “one skill level that might properly belong in another.” To point out unwarranted wage disparities, Willis made evaluations across service groups and concluded that if “female dominated classes in the Human Service group were evaluated on Engineering and Scientific grids, 9 of the 11 classes would have higher salary ranges.” Three male dominated classes would have made less (Tr. 626-27). Mr. Willis made cross evaluations between Domestic Workers and Labor & Trades and between Clerical and Business Administration. He found that all Domestic Workers classifications would have received more money if assigned to the Labor & Trades Service group (Tr. 632-33) and 9 of the 14 clerical classes would have been paid more while only 1 would have been paid less (Tr. 630). Mr. Willis testified that such results could not have occurred in his system (the Willis system), nor in the original Oliver system. He admitted that he would have adjusted the point ranges where needed, not as the only pattern to follow but to illustrate “one way to do it.” He further admitted that wage disparities between predominantly male and female groups do exist in the labor market. He could not explain the reason for the wage disparities, he only knew they existed. He conceded that when using the Benchmark factor ranking system, as adopted by the state, he obtained the same results they did.
On the other hand, Mr. Philip Oliver, the author of the “Oliver System,” reviewed the state’s Benchmark system at the request of the defendants. He concluded from his study of the distribution of skill levels within a service group and the point ranges assigned to each class that the system was acceptable and met his standards for validity and reliability. He further concluded that although the number of service groups may be excessive, they do follow a pattern in the labor market. He agreed that in making cross evaluations, he would expect a variance in wages across service groups because each of the service groups is made up of jobs that compete differently in the labor market. He stated that as a professional, he would “fine tune” the system to conform to his philosophy. He would, for example, combine Labor and Trades and Domestic Workers in one service group because the grids for the two service groups are approximately the same.
Although the testimony of Messrs. Willis and Oliver may point to various flaws in the Benchmark system, Title VII does not require that a job evaluation system be perfect in all respects. We are persuaded that the Benchmark system is a valid job classification and compensation system. Neither its design nor implementation evidence an intent to discriminate against incumbents of predominantly female classifications.
Plaintiffs sought to prove at trial that defendant’s own job evaluation studies are probative of an intent to discriminate. The specific studies plaintiffs introduced in evidence are the 1971 Employment Practices Review, the two studies issued by the Office of Women and Work, the Arthur Young report, the report on occupational segregation issued by the Civil Service Commission in 1983, and the findings of the Comparable Worth Task Force. They are correct that each of these studies and reports found that wage disparities and occupational segregation do exist in the Michigan classified service. The existence of disparities, however, does not demonstrate *901discrimination in general or intentional discrimination in particular. Mere awareness of any disparate impact the classification and compensation system might have on incumbents in female dominated classes is insufficient to prove intentional discrimination. Plaintiffs have failed to demonstrate that defendant chose to proceed with Benchmark because of its effect on predominantly female positions. AFSCME, 707 F.2d at 1405. In addition, the studies do not consider the effect of the labor market on pay, which clearly is a permissible factor and one which the state heavily relies upon as part of its pay policy. If anything, the various job evaluation studies conducted by the state demonstrate that the defendants perceived a problem concerning women, they investigated that problem, and through the Task Force sought to remedy the problem. If defendants intended to discriminate against women, it is unlikely that it would have devoted its time, money, and other resources not only to reveal but to publicize the allegedly discriminatory condition.
The final circumstantial evidence of intentional discrimination presented by plaintiffs consisted of sex restricted examination announcements and help wanted ads. The exhibits are of only limited relevance to this action because they were discontinued after 1972. They clearly are insufficient on their own to establish a prima facie case of discrimination and add little to plaintiffs other circumstantial evidence.
Plaintiffs presented extensive statistical evidence to bolster their claims of intentional discrimination. They engaged Dr. Rothman to determine whether the conversion to the Michigan Benchmark Evaluation System resulted in a disparity between wages assigned to female and male dominated classes. Dr. Rothman described the history of Michigan’s classification system and then traced its conversion to the Benchmark system, using charts to illustrate his findings. These charts reflect that when the service groups were first established in 1975, male and female classes that had been grouped together for classification purposes were often separated for compensation purposes. He found, for instance, that 29 of 33 predominantly male classes from the 1968 Labor & Trades occupational groups were assigned to Labor & Trades service groups in 1975. Most of the predominantly female classes were assigned to the Domestic Workers Service Group, and as a consequence, received less money. (PX 198-1C) (Tr. 872-79.) He further noted that predominantly female classes were paid less than predominantly male classes in the same skill level in the Human Service group, the Business/Administrative group and the Engineering/Scientific group at the time of conversion (PX 198-6A-C) (Tr. 908-10).
Dr. Rothman testified that as a result of the Benchmark conversion, predominantly male classes received larger pay increases than predominantly female classes although both were in the same service group (PX 198-5) (Tr. 912-17). This was statistically significant in the Domestic Workers, Engineering/Scientific, Human Services and Business/Administrative service groups and showed that predominantly female classes were paid less than predominantly male classes. He found that when controlling for defendant’s Benchmark evaluation scores, combining Domestic Workers and Labor & Trades service groups produced the same result. He concluded that these results were statistically significant (PX 198-9A-E).
Dr. Rothman studied the entry level classes for certain predominantly female and male classes across all service groups for which there are no entry restrictions. He focused on the maximum rate assigned to a particular class. He found that predominantly female classes are paid less than predominantly male classes. This was true with Labor & Trades and Domestic Workers service groups combined, from 1976-1986. He noted that if laborers had been combined with domestic workers, a predominantly female group would have been paid more. His study demonstrates that when combining Labor & Trades with Domestic Workers and Engineering/Scientific with Human Services, predominantly female classes are paid less than predominantly male classes controlling for skill lev*902el (PX 198-12). The same was true when he combined Clerical and Business/Administrative groups. Rothman further testified that the predominantly female classes in the Human Services group are paid less than predominantly male classes when controlling for PCEL’s.
Defendants countered with statistical evidence from Dr. Abowd, a labor economist. Abowd examined the results of all compensation decisions made by the state from 1976-1986. He examined the amount incumbents were actually paid and not the maximum rate. He found that there was no pattern of changes in the compensation system that adversely affected employees in female dominated classes from 1976-1981 (Tr. 1343). He actually found a pattern of changes in the compensation system that benefited employees in female dominated classes (DX 61-2).
Dr. Killingsworth, an economist, also called by defendants, testified that studies investigating discrimination in compensation generally should focus on individual incumbents, while controlling for differences in individual characteristics. In other words, a proper study should look at people and what they are paid, rather than at jobs. He characterized Dr. Rothman’s studies as “comparable worth” studies because they considered only the maximum rate assigned to classes. Killingsworth did his own studies, focusing on individual employees. Controlling for individual characteristics and PCEL’s, he found relatively small differences in pay between men and women in general and between incumbents in predominantly male and female classes.
We find that Dr. Rothman’s studies, considered alone or in combination with Dr. Willis’ testimony, are insufficient to persuade us that the state intentionally violated Title VII of the Civil Rights Act of 1964. In arriving at his conclusion that the Benchmark system as implemented resulted in wage disparities between predominantly female and male classes, Dr. Roth-man relied upon the maximum pay rate assigned to each class. The maximum pay rate is an administratively assigned maximum for a class that has little to do with the actual rate paid to individual incumbents. The state could easily have increased or decreased the maximum rate for a class without affecting the pay of an incumbent. Dr. Rothman offered an unsatisfactory reason for basing his statistical computations upon the maximum rate. In addition, Rothman’s charts included many unweighted statistics, and he could not determine whether the Benchmark system benefited men more than women (Tr. 986-89). Plaintiffs could have accounted for the number of incumbents in each class being compared and presented the court with weighted statistics but they chose to do so only to a limited extent. Dr. Roth-man’s studies were essentially “comparable worth” studies and are insufficient as evidence of intentional discrimination.
The gist of the statistics and other evidence presented by plaintiffs in this case is that defendants have not consistently paid incumbents of predominantly female positions the same as incumbents in predominantly male positions that have the same point factor scores or the same “value” to defendant. Although they carefully avoided the term at trial, what plaintiffs have presented is a comparable worth case. Like the court in AFSCME, we need not comment on comparable worth as an approach to employee compensation. We must conclude, however, that plaintiffs’ comparable worth evidence fails to demonstrate discrimination actionable under Title VII.
In summary, we conclude that the evidence overwhelmingly preponderates in favor of the defendants. The evidence demonstrates that the defendants have always been sensitive to inequities among state employees in the classified system. When defendants recognized that their early classification system had become so cumbersome it could no longer be managed or understood, they voluntarily sought to replace the system at enormous expense to the state. The Commission’s staff called as witnesses by both plaintiffs and defendants described the decision-making process in adopting the Benchmark Ranking System. The system was designed and implemented with complete consistency and was *903described as valid and reliable. Not one shred of evidence suggests that staff considered or even knew the gender of the incumbents in jobs being classified. Nor, is there any evidence that the state took any action designed to prevent women from applying for or holding any job in the classified system. In fact, as early as 1971, in its Employment Practices Review, the state reviewed its own employment practices to be certain there were no prevailing practices that precluded women from obtaining any position in the classified service for which they were qualified.
Similarly, the pay conversion policy and all of the criteria developed for the Benchmark implementation were applied consistently throughout each service group. One of the criteria required dependence on labor market data to be sure the state could successfully compete with other employers private and public. Prior to conversion, Mr. Perkowski conducted salary surveys as one of the factors to be considered in assigning wage rates. He also considered various economic predictions as well as the state’s ability to pay the various wage rates. In addition, staff obtained comments on the proposed wages from unions, employers and the state’s appointing authorities. The results were then submitted to the defendant Commission, which processed the data through its elaborate public hearing system before setting the wage rates. Finally, to assure greater equity, the Commission granted state employees the right to collectively bargain certain terms and conditions of employment including wages. Throughout this development, culminating in collective bargaining, the Commission employed voluntary affirmative action, created new career ladders for women without college degrees, accepted job experience as a substitute for education requirements needed to transfer into classes requiring degrees, and created the position of departmental trainee.
This conduct is totally inconsistent with a pattern and practice to discriminate against women in predominantly female classes. Rather, the evidence establishes the Commission’s determination to conform its decision-making authority to the state’s constitutional mandate to treat all state employees equitably.
Accordingly, we find in favor of defendants and dismiss plaintiffs’ complaint.
IT IS SO ORDERED.
. The original complaint alleges violations of Title VII of the Civil Rights Act of 1964 and included violations of the Equal Pay Act and Michigan Civil Rights Statutes. The court dismissed plaintiffs’ Equal Pay Act and state claims on July 2, 1987. In that order, the court certified the plaintiff class as "all past, present and future classified employees, male and female, under the jurisdiction of the administration of the Civil Service Commission, who, since May 30,1980, have worked or do work in job classifications that are 70% or more female.”
. Stipulated facts, #17. Although we do not set them forth at length here, we adopt the stipulated facts set forth in the second joint pretrial order.
. In 1969, then Governor William Milliken requested the Department of Civil Rights and Civil Service to conduct a comprehensive joint study of the employment practices in state government. The findings were reported in the Employment Practices Review, dated August 1971.
. Stip. of facts # 32.