PATEL, District Judge.

Plaintiffs, the California State Employees' Association and named individuals ("CSEA"), filed suit against the State of California, the governor of California and the State Department of Personnel Administration, alleging violations of Title VII of the Civil Rights Act of 1964, 42 U.S.C. §§ 2000e-2000e-17. Under American Fed'n of State, County and Mun. Employees v. Washington, ("AFSCME"), 770 F.2d 1401 (9th Cir.1985), plaintiffs must prove discriminatory intent to prevail on disparate treatment claims. Id. at 1405 . Since it was clear from the lengthy pretrial record in this case that the critical period for showing intent was the 1930's, the court indicated it would try that aspect of the case first. The 1930's era was the only period where this court found a palpable showing of discriminatory intent possibly sufficient to support a disparate treatment theory. The focus of this first-stage inquiry was the conduct of the State during the early 1930's and the adoption of a new salary structure in 1938. The chief question was whether discriminatory intent in the early 1930 period could be established and if so, whether it carried over into or was eradicated by the revamped salary structure. The court conducted a trial over a ten-day period on the narrow issue of whether defendants intentionally discriminated against female-dominated classes from the 1931 classification plan to the 1938 restructuring.

After considering the entire record, for the following reasons, the court finds that plaintiffs have failed to show by a preponderance of evidence that defendants intentionally discriminated against historically female classes in the 1931 classification, and that such discrimination carried over into the 1938 restructuring. The court therefore concludes as a matter of law that plaintiffs cannot maintain their claims based on a disparate treatment theory under Title VII, and those claims are dismissed. Plaintiffs have indicated they also intend to proceed on a disparate impact theory. Parties are ordered to submit additional briefing on the status of the pending disparate impact claims in light of this order and the Supreme Court decision in Wards Cove Packing Co. v. Atonio, ___ U.S. ___, 109 S.Ct. 2115 , 104 L.Ed.2d 733 (1989).

This action has a long history, which is summarized here only briefly. The complaint in the instant action was filed in 1984 and alleged that certain female-dominated job classifications were paid less than other job classifications, when evaluated on a "comparable worth" theory. "Comparable worth" evaluates different jobs on such criteria as skills, effort, responsibility, and working conditions. See generally U.S. Comm'n on Civil Rights, 2 Comparable Worth: Issue for the 80's 47 (1984). In 1985, the year after the action was filed the Ninth Circuit issued its decision in AFSCME v. State of Washington, 770 F.2d 1401 (9th Cir.1985), in which the court held that the state's reliance on a prevailing rate system did not establish a violation of Title VII under a disparate treatment theory. AFSCME explicitly required that in order to prove discriminatory intent, plaintiffs in gender-based wage discrimination cases must offer evidence of intent. A mere showing that compensation was paid on market-based rates was insufficient to meet the requirement. 770 F.2d at 1406-1408 .

The court denied defendants' motion for summary judgment by order of December 21, 1987. California State Employees' Ass'n v. California, 682 F.Supp. 1044 (N.D.Cal.1987). In an order dated August 25, 1988, the court denied defendants' motion for reconsideration. The court concluded, however, that there was virtually no evidence to support discriminatory intent after 1938 when a restructuring of the California compensation schedule was instituted. In view of the paucity of evidence for post-1938, the court determined that the only thread upon which plaintiffs' discriminatory treatment claim hung was a showing of discriminatory intent in wage-setting in the early to mid-1930's, of which there was some evidence, and proving that the state intentionally maintained these pre-existing discriminatory salary practices despite the 1938 restructuring. If plaintiffs could establish this link, the court then would allow them to pursue the further historical link to the statutory period in question here. Specifically, this court stated:

Order of August 25, 1988, at 13. The court accordingly ordered that a trial be held on that isolated issue.

Plaintiffs alleging gender-based discriminatory treatment in a class action suit under Title VII bear the burden of showing "by a preponderance of the evidence that sex discrimination was defendant's `standard operating procedure' maintained through systematic intentional discrimination." Penk v. Oregon State Bd. of Higher Educ., 816 F.2d 458 (9th Cir.), cert. denied, 484 U.S. 853 , 108 S.Ct. 158 , 98 L.Ed.2d 113 (1987). (citations omitted). Plaintiffs must demonstrate that an employer followed a particular policy because of its effect on members of a protected class; a showing that an employer was aware that a policy disadvantaged a certain group will not suffice. AFSCME, 770 F.2d at 1405 . To prevail on their disparate treatment claims, plaintiffs therefore must establish that before 1938 defendants intentionally set lower salary levels for female-dominated job classifications and that such discrimination was intentionally maintained after the 1938 state restructuring. See Order of August 25, 1988, at 9 (specifying plaintiffs' burden of proof).

Discriminatory intent can be inferred from circumstantial evidence and, in some instances, from statistical evidence. Forsberg v. Pacific Northwest Bell Telephone, 840 F.2d 1409, 1418 (9th Cir.1988) (citing Furnco Construction Corp. v. Waters, 438 U.S. 567, 576 , 98 S.Ct. 2943, 2949 , 57 L.Ed.2d 957 (1978) and Spaulding v. University of Washington, 740 F.2d 686, 696 (9th Cir.) ("Spaulding"), cert. denied, 469 U.S. 1036 , 105 S.Ct. 511 , 83 L.Ed.2d 401 (1984)). However, courts should treat statistics with caution. Forsberg, 840 F.2d at 1418 (citing Spaulding, 740 F.2d at 703 and International Brotherhood of Teamsters v. United States, 431 U.S. 324, 338-40 , 97 S.Ct. 1843, 1855-57 , 52 L.Ed.2d 396 (1977)).

In 1931, the state of California established a comprehensive pay and classification survey for its civil service system under the direction of Fred Telford, then serving as Executive Assistant of the Division of Personnel and Organization. See Ex. JA-21 [1] (Telford memorandum to State Civil Service Commission and Finance Director dated June 5, 1930). The state collected compensation data from large public and private employers for approximately two hundred "key" classes. Id. at 4; see also Ex. JA-22 at 1.

To demonstrate that this classification system was infected by gender discrimination, plaintiffs rely principally on two documents which they characterize as "smoking guns": the "Game Farm Cook memo," Ex. A-59 and the "Ten Questions document," Ex. JA-25. Plaintiffs also present a master's thesis on the 1931 compensation plan, two state questionnaires, sex segregation documents and other state documents, and a statistical sample compiled by Dr. William Dickens, plaintiffs' expert witness. Defendants offer an alternative survey of 1931 rates, and other state documents. The court now considers the documents and testimony on this issue.

This internal state memorandum dated August 10, 1931 focused on the salary range for the pay class of Game Farm Cook. Exs. JA-59 and JA-59A. Eldred Cocking, a graduate student personnel technician, authored the memorandum and submitted it to state officials responsible for pay classifications. [2] In recommending a modest increase in salary for the Game Farm Cook class, Cocking wrote:

Ex. A-59.

Plaintiffs argue that this memorandum demonstrates that the state followed a policy of paying female-dominated classes at lower rates than those for classes dominated by men. [3] However, the Game Farm Cook memo's probative value is limited. First, Mr. Cocking was employed as a graduate student assistant and it has not been shown that he had the authority to formulate or characterize state pay policies. [4] Plaintiffs have introduced copies of the Cocking memo signed by Ralph H. Moss, the Pay Supervisor, and William Brownrigg, Chief, Division of Personnel and Organization. While these signatures might suggest that Mr. Cocking's comments were at least tacitly approved by those officials, the court declines to infer from this evidence that Mr. Cocking's memo codified state policy. Defendants have cited a state statute and internal documents establishing equal pay for equal work, and other nondiscriminatory principles. See Cal. Const. of 1879, art. XX, § 18, JA-1 ("No person shall, on account of sex, be disqualified from entering upon or pursuing any lawful business, vocation or profession"); Cal. Civil Service Reform Act of 1913, 1927 Cal.Stat. 131, § 5, Ex. JA-3 ("like salaries shall be paid for like duties"); memorandum dated June 5, 1930 to State Civil Service Commission from Fred Telford, Ex. JA-21 ("goal of `equal pay for equal work under like working conditions'"). Against this background of official pronouncements, Mr. Cocking's memorandum casts doubt upon the sincerity of defendants' enforcement, but it does not establish the existence of discriminatory intent throughout the system.

Second, the Game Farm Cook memo concerned a small number of state employees. According to the testimony of Jay Atwood, a current state employee knowledgeable about job descriptions and history, only three incumbents were in the Game Farm Cook class as of 1934. RT 8:1313. [5] As shall be discussed later, the Game Farm Cook job class was eliminated in December 1936. Exs. B-303, 304. In 1936, the category of Game Farm Cook, predominantly female, was merged with the class of Camp Cook, predominantly male, at which point men and women in this category were paid equally. Id. Even assuming arguendo that Mr. Cocking's memorandum accurately represented state policy concerning the Game Farm Cooks, such evidence does not show that discriminatory intent infected the myriad of state pay classes. Furthermore, by the 1938 restructuring there was no longer a separate Game Farm Cook position.

Third, even the Game Farm Cook memo itself suggests that various nondiscriminatory reasons influenced state pay policy. The memo cited responsibilities and duties, as well as working conditions, as reasons for state pay practices and classifications. Although the Game Farm Cook memo, badly overdone at trial by plaintiffs and their expert witness, suggests a flavor of gender prejudice on the part of some state officials in the 1930's, its probative value regarding the pay-setting process in general is limited.

In preparing the 1931 compensation plan, the staff of the state Civil Service Commission compiled a "List of Questions Relating to Financial Policies and Standards to Be Answered by the State Civil Service Commission and the Director of Finance" ("Ten Questions document"), Exs. JA-25 and JA-25A (dated October 14, 1930). Presumably this document was prepared at the behest of Fred Telford, who had been appointed Executive Assistant in the Division of Personnel and Organization effective April 10, 1930, Ex. JA-19, and had proposed a classification and compensation study to assist the staff with pay-setting policies. Ex. JA-21.

Plaintiffs assert that the responses to the fifth question in the Ten Questions document present direct evidence of intentional gender-based discrimination. The fifth question asked: "Shall any differences in pay on account of sex be made?" One of three suggested methods of answering the question proposed that, "[w]hen men and women do the same kind of work to make no difference, but to pay somewhat higher for those occupations filled predominately by men than for those occupations filled predominately by women, where, aside from sex, the qualifications are substantially the same." Ex. JA-25 at question 5. The staff endorsed this procedure but specified that "the differentials ... be limited to those kinds of occupations where in the commercial world distinctions are made in the pay for workers engaged in occupations predominantly filled by men as compared to those predominantly filled by women." Id. In selecting this alternative, the staff rejected the method in which men received consistently higher pay than women. Id. Rather than demonstrating intentional gender-based discrimination, the statement here actually tends to show that the policy makers based pay differentials on prevailing rates, not on gender.

Additionally, the responses to the fifth question must be placed in the context of the whole document. The compensation staff urged that pay scales be adopted "primarily on the basis of the rates paid for the same or comparable work by other large commercial and public employers in the State of California." Ex. JA-25 at 1. The Ten Questions document also recommended following union rates "with a few possible exceptions ... the objective being to make it possible for a worker steadily employed in the state service to earn somewhat more in the course of a year than he would earn under existing conditions outside the state service if paid the regular union rate." Id. at 4 (question # 4). [6] Furthermore, the staff listed additional factors including responsibilities and duties, educational requirements, and working conditions, to be considered in determining specific compensation levels. Id. at 3. Other than the fifth question, there are no references to gender as a factor in salary-setting.

Finally, the impact of the Ten Questions document on state pay policy is uncertain. In letters from Fred Telford to the Director of Finance, Governor-elect Rolph and the Secretary of the California Taxpayers' Association, the Ten Questions document was discussed as a possible basis for compensation levels. Exs. A-27 at 3 (Telford letter to Finance Director, dated November 19, 1930), A-28 at 3 (Telford letter to Governor, dated November 26, 1930), A-29 at 1 (Telford letter to Secretary, California Taxpayers' Association, dated November 27, 1930). However, the tentative nature of the Ten Questions document is emphasized in Telford's communications. See, e.g., Ex. A-29 at 1 (noting that answers to Ten Questions are based on assumption that departures from existing policy will be instituted only when data show need, "or when necessary to remove the inequalities for groups and individuals.") There is no evidence that the recommendations in the Ten Questions document ever were enacted into law, were implemented, or in any way found their way into the final salary schedules. See, e.g., Ex. JA-26 at 1 (Minutes of the Meeting of the State Civil Service Commission dated October 15, 1930) (Telford presented list of compensation policies and list of questions relating to policies and standards, but action deferred); Ex. JA-27 at 3 (Letter from Telford to Finance Director, dated November 19, 1930, noting that Ten Questions have been submitted to Director of Finance); Ex. JA-28 at 3 (Telford Letter to Governor, dated November 26, 1930, noting Ten Questions have been prepared); Ex. JA-29 at 1 (Telford Letter to Taxpayers' Association, dated November 27, 1930, noting same); Ex. JA-33 at 1 (Telford Letter to State Civil Service Commission, dated February 11, 1931) (stating that personnel program "from time to time has been talked over and which, as opportunity affords, is being put into effect.") The Ten Questions document indicates that gender bias existed among some state officials, but it also tends to support defendants' contention that the state relied on prevailing rates in salary-setting.

This internal state memorandum, dated October 25, 1931, was written by Eldred Cocking, addressed to Ralph Moss, Pay Supervisor, and concerned salary rates for state nurses. Ex. A-68. Noting that nurses had objected to the proposed salaries in the Bureau of Registration for Nurses, Mr. Cocking recommended that the salary ranges be increased for the positions of Bureau Chief (one incumbent) and Inspector, School of Nursing (indeterminate number of incumbents). Id. The memorandum recommended a higher salary for the Nurses' Bureau Chief, but not at the level enjoyed by the level of Cannery Inspection Bureau Chief, despite the "somewhat comparable nature" of the jobs. Id. In justifying his position, Mr. Cocking referred to "elements in the situation" including past salaries for the incumbents and the nurses' protests about their salaries. Id. For plaintiffs' case, the critical evidence is the hand-written notation of Fred Telford concerning "other elements," the first of which is "white collar job for women." Id.

On October 31, 1931, Mr. Moss recommended rejecting Mr. Cocking's proposal for an increase in the salary range for Bureau Chief in favor of maintaining the status quo. Ex. JA-71. Mr. Moss' analysis cited various factors for this decision, including the smaller size of the nursing bureau and the high supply of qualified nurses. Id.

Although Mr. Telford's notation suggests that some gender bias may have infected the salary-setting process, the evidentiary value of the exhibit is necessarily limited by its scope (one incumbent) and by the uncertainty of the rationales supporting the state's ultimate determination. For these reasons, the court gives this memorandum limited weight.

Plaintiffs also introduced into evidence the master's thesis of Fred Becker, who worked as a graduate student assistant to Fred Telford beginning in September 1931. F. Becker, Administration of the Personnel Program in the State of California (1934), Ex. JA-91; see RT 5:747-48 (Becker testimony). Mr. Becker's thesis for the University of California political science department included in its appendix three state questionnaires including sex as a relevant category of description. Ex. JA-91 at 92, 93, 150. See also RT 1:70-72 (Dickens testimony). On one of these forms, the "Standard Classification Questionnaire," state employees provided information including sex, marital status, education, and previous work experience. Ex. JA-91 at 92. See also Exs. JA-91 at 93 ("Revised Classification Questionnaire" appended to Becker thesis also asked for information on employee's gender); and JA-115 at 136 ("Occupational Classification Questionnaire" appended to 1937 Telford Classification Manual included sex as category for employees). Another form, the "Employment and Wage Questionnaire," asked private employers in the state to supply information concerning the type of job held by their employees, the breakdown by sex and adults/minors, and the compensation plan then in effect. Ex. JA-91 at 150.

While it is undisputed that the state collected this data concerning gender composition in occupations, its significance for the pay plan is uncertain. See Ex. JA-91 at 154 (personnel statistical card code slip); Ex. JA-163 at 44-45 (Smith master's thesis discussing class requirements includes "personal traits" but no mention of gender); RT 1:79-81 (Dickens testimony admitting uncertainty regarding how gender information was used to discriminate). See Memorandum from Fred Telford, dated June 23, 1930, Tentative Plans for Collecting and Tabulating Compensation Data, Ex. JA-22 (no mention of tabulating sex composition data).

Mr. Becker's thesis lists "sex" as one of many "supplemental factors" affecting salary levels in a classification and compensation plan. Ex. JA-91 at 62. However, Mr. Becker's thesis gives primary importance to many other factors, including market rates, personnel supply and demand, responsibilities and duties, qualifications, and internal relationships. Id. Furthermore, Mr. Becker testified that, at the time he wrote his thesis, he had no knowledge of California's salary-setting practices nor had he observed gender being taken into account by the state. RT 5:758-59. According to Mr. Becker, the reference to sex as a supplemental factor was derived from a secondary source, Civil Service Assembly of the United States and Canada, Classification and Compensation Plans: Their Development, Adoption and Administration 14-16 (1928), Ex. JA-17. [7] The court finds that, given the uncertain source and the ambiguous statement, this reference has little probative value. [8]

In addition to internal state memoranda and analyses, plaintiffs introduced statistical studies prepared by their expert, Dr. William Dickens. Dr. Dickens offered a sample of gender-denominated job classes and the minimum and maximum salaries derived from the pay scales of 1931, 1936, 1938, and 1941, with sex composition data compiled as of 1935. See Ex. A-235; RT 2:200-05 (Dickens testimony on table). On the basis of this table, Dr. Dickens concluded that in 1930-31 the state paid female job classes (here defined by gender in the title) [9] less than male classes. RT 2:204. Moreover, Dr. Dickens asserted that, even if fifty per cent of the wage differential could be attributed to differences in job duties, ten to thirty-three per cent of the disparity derived from gender-based discrimination alone. RT 3:423-25; 10:1515-17.

This statistical sample is plagued by faulty methodology and questionable analysis. First, and most significantly, Dr. Dickens admitted that he did not examine job duties and responsibilities, but attempted to "match" jobs merely on the basis of titles. RT 2:200 ("plausible that ... duties and responsibilities would be very similar, if not identical"); 2:204; 3:406; 10:1518. Job descriptions introduced by defendants indicate that this method of analysis was flawed, since significant differences existed among some of the "matched" classes.

For example, the male-dominated cook classes (Institutions Cook, Camp Cook, and Ship's Cook) cooked large quantities and faced atypical working conditions such as remote locations or work aboard a ship. RT 9:1334 (testimony of Jay Atwood) (relying on Exs. B-310 (Ship's Cook description); B-403 (Institution Superintendent's Cook); B-399 (Camp Cook); B-401 (Institution Cook); B-402 (Institution Cottage Cook)); see also Ex. B-302 (Game Farm Cook). Whereas the lower-paid cook classes, Institution Cottage Cook, Institution Superintendent's Cook, and Game Farm Cook, cooked for no more than thirty people and worked under typical cooking conditions. Id. State documents showed that other matched job classes in 1931 had contrasting duties and responsibilities, or working conditions. See, e.g., RT 8:1315 (Atwood testimony concluding that Janitor had moving and lifting requirements, while Janitress did not) (relying upon job descriptions in Ex. B-329 (Janitor) and Ex. B-331 (Janitress); RT 8:1317 (Atwood testimony that superintendent for boys' school oversaw population four times larger than that of girls' school) (relying on Ex. B-424); RT 8:1320 (Atwood testimony that boys' school supervisor dealt with more dangerous clients than girls' school supervisor) (relying on Exs. B-326 and B-327).

Second, the sample represents a minute fraction of the state work force. After subtracting classes with no salary differential or where Dr. Dickens himself found comparison inappropriate, only sixty-nine women in women-dominated classes remained in the sample, out of a work force estimated at over 14,000. RT 8:1311-12 (Atwood testimony). Since the 1937-38 restructuring eliminated certain gender-denominated jobs in the sample, those also must be subtracted in the consideration of the carry-over effect of any discrimination. Only thirty-seven women thus would remain in Dr. Dickens' sample after the 1937-38 job classification, and defendants offered plausible explanations that any salary differentials for these were based on working conditions or duties. RT 8:1315-21.

Another fact gleaned from Dr. Dickens' key classes is striking. The one position that employed the largest number of women was the class of hospital attendant. This position when surveyed for 1938 had 842 men and 857 women. Ex. A-235. There was no other class among his key classes with such an equal ratio of men to women. Of course, this class did not meet Dr. Dickens' definition of female- or male-dominated, but it is instructive. This category of employees received a significant upward adjustment in salary in 1938 because, according to the testimony of Roy Stephens, a former State Personnel Board ("SPB") employee and the only eyewitness to the 1937-1938 restructuring, there was popular sentiment for an increased salary in part because the state was the dominant employer. While this position is not one of the defined female- or male-dominated classes, because of the large number of women employees in the class, its treatment cannot be ignored. The rationale for setting and increasing the salary of this class clearly cuts against plaintiffs' position.

Simply put, Dr. Dickens' sample of closely related, sex-segregated jobs is not reliable. The court therefore accords it no evidentiary value. [10]

Defendants introduced evidence showing that factors other than gender-based wage discrimination influenced state salaries during this period. Dr. Michael Ward, defendants' expert witness, produced a survey based on data from a study co-authored by the state and CSEA. California State Employees' Association and the Division of Personnel and Organization, A Comparative Analysis of Salaries Paid Employee Groups in Public and Private Enterprise (1933), Ex. B-86; RT 6:905-06. According to Dr. Ward, this 1933 study attempted to update the results from the 1930 compensation survey, in response to political pressure to reduce state salaries, at the outset of the Depression. RT 6:903-05 (also relying on Ex. B-83 at 2) (CSEA bulletin reports on prospects of salary reduction).

In his use of the data from the comparative analysis, Dr. Ward focused on the categories of female-dominated occupations (70% or more women) and male-dominated occupations (70% or more men). Ex. B-232; RT 6:910-11 (Ward testimony). Female-dominated categories included clerks, stenographer-clerks, typist-clerks, and secretary stenographers; male-dominated classes included engineers, accountants, purchasing agents, and medical superintendents. Ex. B-232. Female-dominated classes in the state earned roughly equivalent salaries to their counterparts in the private sector. Ex. B-232; RT 6:910-11 (Ward testimony). On the other hand, state employees in male-dominated occupations (70% or more men) tended to earn considerably less than their counterparts in private employment (with the exception of accountants-auditors, who earned approximately the same salary in state service as received by those in private business). Id. Dr. Ward's work also revealed that every female-dominated occupation in the state earned less than the male-dominated occupations; only the lowest-paid male-dominated class, accountant-auditor, was even close in average salary to the highest paid female-dominated class. Id. [11] Dr. Ward's survey thus supports the conclusion reached in a 1930 report by CSEA, which commented on the results of the compensation survey as follows: "[F]or the most part those in the clerical, mechanical, and unskilled groups were receiving salaries and wages comparable to those in private employment, while those in the professional or highly technical groups were not so well paid." F.A. Taylor, Civil Service and the State Employees' Association, California State Employees' Association, Bulletin No. 1, July 1, 1931, at 9, col. 1, Ex. B-59; RT 6:902.

This evidence further undermines the contention that discriminatory intent infected state salary policies. Additionally, it suggests that factors other than market forces, perhaps internal relativities or notions of fairness, played a role in the state salary structure.

Plaintiffs also presented other documents showing, for example, that the state engaged in some sex-segregated job exams and hiring practices prior to and during the 1930's, Exs. A-15 (1927), A-91 at 91, A-92 (1934), A-94 (1935), even though those practices arguably violated the California Constitution and the Civil Service Reform Act of 1913. The sex-segregated job classifications do suggest discriminatory or stereotyped attitudes concerning men's work and women's work. However, plaintiffs failed to link sex segregation practices to their allegations of depressed salaries for women during this period, so the court gives little weight to this evidence.

Dr. Dickens testified that references in various state documents to "costs of living" implied that the state considered different standards of living for its employees. RT 1:119-24 (discussing Civil Service Commission, Division of Personnel and Organization, Pay Scales for Classes of Positions in the California State Service 3 (1931), Ex. JA-53. Dr. Dickens found significant the fact that the plural term was used, believing that it implied that the state employed different measures for different classes. RT 1:122-23. According to Dr. Dickens, this "costs of living" factor enabled the state to discriminate against women, because it would assume that a male's income would be needed to support a family, and a female's income would not. RT 1:123. While the court certainly could imagine that this might occur, it is absolutely and utterly unsupported in the record.

For instance, the 1931 Pay Scales document, Ex. JA-53, refers to "present and past costs and standards of living." Ex. JA-53 at 3. Obviously, the plural form recognizes that the cost of living does not remain constant over the years. Dr. Dickens' reading of the 1936 pay scales, Ex. JA-101 at 2, 5, the proposed 1937 pay scales, Ex. JA-133 at 4-5, and the 1938 pay scales, Ex. JA-147 at 4, is similarly strained and unbelievable. See RT 6:856 (Ward rejecting Dickens' interpretation of "costs of living."); see also Ex. JA-22 at 3 (Telford refers to "cost of living" and changes since 1913 as important in classification plan). [12]

Plaintiffs' "smoking guns" never held ammunition. They suggest a biased or stereotyped attitude on the part of some state officials which this court would not hesitate to describe as disturbing. [13] However, the 1930 survey and salary setting process has been appropriately called a "black box" and the court finds that the trial shed remarkably little light on the darkness. [14] Assuming arguendo that plaintiffs have shown intent during this period, the court now turns to the 1937-38 restructuring to determine whether any alleged discriminatory intent carried over into that period.

In 1934, California voters approved Article XXIV of the California Constitution, which created a merit civil service system and an independent State Personnel Board to oversee classification and pay matters. Ex. JA-89. The Civil Service Act of 1937, enacted pursuant to the mandate of Article XXIV, required that the personnel board ascertain prevailing rates in the applicable localities and maintain the minimum salary at or above the general prevailing rate. 1937 Cal.Stat. 2094-95 §§ 70-71, Ex. JA-126. The 1937 Act also prohibited gender discrimination, but allowed the reservation of some positions for persons of a particular gender. 1937 Cal.Stat. 2110 § 201, Ex. JA-126.

In January 1937, Board Executive William Brownrigg had identified over 250 different pay ranges in the state, and initiated a compensation survey to reduce that number and to promote equity among state employees. Memorandum to SPB from William Brownrigg, dated January 19, 1937, Ex. JA-108; RT 6:921-22. The state Assembly also adopted a resolution urging increased state salaries in light of the recovery from the Depression. Excerpt from Assembly Daily Journal of Friday, January 22, 1937, pages 50 and 51, Ex. JA-109. The State Personnel Board therefore resolved to conduct an "extensive survey" to find out what changes needed to be made to bring state salaries in line with private salaries, and to ensure that "like salaries be paid for like work." Ex. JA-111.

Following the collection of employment and wage questionnaires from public and private employers in the state, Ex. JA-139 at 9, RT 6:957-58, on September 7, 1937, the SPB published a "Report of Findings, Salary Survey." Ex. JA-132. For this report, the Board selected 258 "key classes" (out of over 1500 in the state) to guide the reform of the pay structure. Id. at 1. Data was obtained for 183 viable classes. RT 6:941.

In October 1937, the Board issued its "Proposed Salary Range Groups" report which used the survey data to design pay groups based on evaluation of duties and responsibilities. Ex. JA-133 at 1. Each pay group included numerous classes, which would be compensated at the same salary level. Id. On October 28, 1937, the Board released a revised "Proposed Salary Range Groups" report with a suggested salary range for each of thirty-six pay groups. Ex. JA-134 at 2-3. Following public hearings and comments by state employees and groups including the CSEA, Ex. JA-147 at 10, the final survey report, containing extensive data for seventy-eight key classes, was published on January 15, 1938. Ex. JA-139 at 4-7.

Plaintiffs contend that, since discriminatory acts infected the state pay system prior to 1937-38, they need only show that its effects carried over past the restructuring. Defendants argue that any possible discriminatory effects of the 1930-31 pay plan were eliminated as a result of the 1937-38 changes. Both parties presented statistical analyses of median salaries for men and women, salary ranges by class during the two periods, and other data. State documents pertaining to the survey and salary setting process were also examined at trial. The evidentiary submissions are considered in turn.

Plaintiffs introduced state documents showing that, on average, women in the state service earned considerably less than men. It is undisputed that as of December 1935, women's median salary was 70.8% of men's median salary. SPB, Twelfth Biennial Report 127 (1937), Ex. JA-137. By 1938, women employees' median salary had risen to 73.9% of men's median salary. SPB, Thirteenth Biennial Report 124 (1938), Ex. JA-164. As of December 1939, the median salary for women had fallen to 71.2% of the men's median salary. SPB, Facts about Employment in the California State Civil Service 4 (1939), Ex. JA-181. [15]

However, these figures provide little support for plaintiffs' case, since the purported significance of these figures is premised on an erroneous assumption: Dr. Dickens' assertion of a 10 to 33% salary differential between men and women. As discussed earlier, Dr. Dickens' calculations were faulty, limited in scope regarding 1930-31, and virtually non-existent for the period following the 1937-38. Therefore, plaintiffs' argument that the quantitative disparities caused by gender bias continued in this period begs the question which salary disparities were caused by gender bias? Indeed, Dr. Ward's chart, Ex. B-233, suggests the obvious alternative explanation for this gap in median salaries: market forces. The aggregated measures of median salaries, without regard to job or class, have no probative value in the instant case.

The evidence shows that, as a result of the 1937-38 restructuring, women's salaries actually increased relative to men's. See Ex. B-377 (calculating average pay adjustment by employees, and finding women received roughly 9.5% raises, while men received roughly 7% raises); RT 7:1081-82. Furthermore, female-dominated classes reaped greater advantage from the 1938 restructuring than did male-dominated, or "mixed" classes, as Dr. Ward's calculations show. Ex. B-378 (classes with 70% or more women enjoyed 1%-2% greater pay adjustment, with specific figure depending on which sample of classes surveyed); RT 7:1082-85. [16] Additionally, after 1938 female-dominated classes tended to do better, relative to the comparable market rate for private employees, than did male-dominated classes. Exs. B-379-381: RT 7:1085-89.

In short, women and women-dominated classes benefitted from the 1937-38 restructuring, undercutting plaintiffs' efforts to show the perpetuation of any discriminatory taint.

Plaintiffs rely on the testimony of Roy Stephens, state documents and other evidence to demonstrate that the alleged discriminatory features of the 1930-31 were reproduced in the 1937-38 restructuring. These items are discussed in turn.

Plaintiffs make much of the testimony of Roy Stephens, the former SPB employee who was the only eyewitness to the 1937-38 pay restructuring. In response to a series of questions from the court concerning salary Group 1, in JA-133 at 8, Mr. Stephens admitted the initial grouping of sheepherder, housemother, and junior messenger derived at least in part from the existing state rates. RT 5:686-88. However, Mr. Stephens also testified that the individual occupation was considered by state officials. RT 5:686. Additionally, he explained that other factors, such as prevailing rates, and the occupational level of education, supervisorial and other duties and responsibilities were evaluated before the plan was submitted for public hearings. RT 5:686-90 (housemother eventually moved to a higher salary group).

Although Mr. Stephens indicated that existing rates were sometimes used as a basis for grouping unlikely occupations, his testimony does not assist plaintiffs. First, he emphasized that many factors contributed to the grouping process. Second, he stated that gender-related data was used to anticipate inquiries from state legislators and administrative bodies, but that gender data was never tabulated. RT 5:664-67, 673-74. He flatly denied that the state engaged in a policy of paying women or women-dominated classes differently from men. RT 5:696; Stephens Dep. Vol. 1 at 73.

Plaintiffs also introduced the transcript of a hearing where a Miss Jane Sedgwick asked Louis Kroeger if it was true that her job class, Food Administrator, Department of Institutions, had been grouped in a certain class on the basis of its present salary. Ex. A-143 at 1. Kroeger replied affirmatively, saying, "We didn't have anything to the contrary, and left it where it was." Id.

The probative value of this document is extremely limited. Not only does the statement only apply to one person in one class, but its meaning is uncertain. See RT 8:1185-87 (Ward testimony).

Plaintiffs also point to the following reference in the report, dated October 28, 1937, Ex. JA-134 at 1:

Plaintiffs interpret this passage as an acknowledgment that current salary was central to the state's classification scheme. However, they ignore the fact that the "original grouping" apparently referred to the October 1, 1937 report, Ex. JA-133, which had elaborated on the importance of grouping classes by duties and responsibilities. Moreover, no broad inference regarding the classification process about the carryover from 1930-31 can be drawn from the October 28, 1937 document's mention, Ex. JA-134 at 1, para. 7, that classes with unusual compensation provisions were not included in this grouping. What is clear is that the Personnel Board had not yet determined the appropriate salary level for those occupations.

According to plaintiffs, the facts that the outside survey and the salary grouping process officially began at the same time and the outside survey ended one month later show that the state did not undertake a class-by-class evaluation. RT 6:957 (Ward testimony regarding schedule of outside survey). Plaintiffs would have the court infer that the salary grouping ended at the same time.

This interpretation is flatly contradicted by the evidence. The Thirteenth Biennial Report, Ex. JA-164 at 23, does not say that the salary grouping ended simultaneously with the outside survey. It states that the "early stages" of arranging all state classes into thirty-six groups occurred at the same time as the outside survey started. Id. Available records and testimony show that the salary grouping process and revisions actually continued until January 1938. See, e.g., Exs. B-119; B-121-23; B-125; see also discussion above at II. Plaintiffs' assertion is without merit.

Dr. Dickens produced a regression analysis using existing 1937 salary rates, 1938 salary rates, and the market rates for the 78 key classes used by the state in 1938. Exs. A-239-241, A-259-261. The regressions showed a high correlation between previous salary and current salaries, and found the predictive effect of surveyed rates relatively insignificant. RT 2:271-79; 3:299-304. However, Dr. Ward's testimony effectively discredited Dr. Dickens' regression. RT 7:1070-73. Dr. Ward's analysis showed that, even if female-dominated classes received the ten to thirty-three percent increases allegedly needed to eradicate discriminatory salary differentials, Ex. B-404 at 3 (Dickens' opinion), the correlation between previous and current salaries would still be extremely high. Id. at 6; Exs. B-388, 395. On cross-examination Dr. Dickens admitted that it would be unlikely to find a salary system without a high correlation between previous rates and existing rates. RT 10:1658-60. [17]

The court finds that Dr. Dickens' regression analysis has no probative value in inferring carryover of discriminatory intent.

It may be true that certain internal relationships and certain features of the state salary structure were preserved, at least in part, in the 1937-38 plan. See Ex. JA-137 at 25-26. [18] Ample evidence suggests that state officials involved with the 1937-38 restructuring aimed to eradicate any vestiges of discrimination. For example, the SPB's document, "Proposed Salary Range Groups," published in October 1937, stated:

Ex. JA-133 at 2 (discussion of "importance of grouping classes.") After discussing the anomaly of paying the identical minimum salary for two auditors' classes having varying responsibilities, the Personnel Board explained, "We feel that this error will be repeated many times unless the whole matter of salaries is approached in general, keeping in mind the relative responsibilities of all positions and make corresponding adjustments throughout the salary structure...." Id. This explicit discussion of the Board's goals, and the evidence as seen in Dr. Dickens' sample, Ex. A-235, indicates that gender-denominated inequalities were phased out during this process. [19] Although it is not entirely clear the extent to which the 1937-38 process broke with the past, the evidence suggests that many ad hoc arrangements that had been grafted upon the state service were evaluated and rationalized. RT 8:1184-85, 8:1195, 8:1208-10; Ex. JA-133 at 3 (October 1, 1937 report stating emphasis on "degree of responsibility"). [20]

Plaintiffs have failed to carry their burden of proof of showing that defendants engaged in intentional discrimination against women state employees. See Texas Dept. of Community Affairs v. Burdine, 450 U.S. 248, 250 , 101 S.Ct. 1089, 1092 , 67 L.Ed.2d 207 (1981); McDonnell Douglas Corp. v. Green, 411 U.S. 792, 802 , 93 S.Ct. 1817, 1824 , 36 L.Ed.2d 668 (1973); Penk v. Oregon State Bd. of Higher Educ., 816 F.2d 458 , 461 (9th Cir.), cert. denied, 484 U.S. 853 , 108 S.Ct. 158 , 98 L.Ed.2d 113 (1987). The court concludes that plaintiffs have not met the standard articulated in AFSCME : the evidence does not demonstrate that the state selected particular pay policies because of any alleged discriminatory effects upon women state employees. 770 F.2d at 1406-1408 . See also Personnel Adm'r. of Mass. v. Feeney, 442 U.S. 256, 279 , 99 S.Ct. 2282, 2296 , 60 L.Ed.2d 870 (1979). On the other hand, defendants have produced credible evidence showing that the pay classification system in effect after 1938 reflected market rates and the duties and responsibilities survey undertaken by the state.

Given the record before it, the court concludes that only a very small number of state jobs in 1930-31 were affected by gender bias or prejudice on the part of some state officials. Even before the 1937-38 restructuring, many of these disparities had been eradicated by merger of groups and salary adjustments. After the restructuring, virtually all vestiges of intentional discrimination as contemplated by AFSCME had been eliminated. Indeed, the evidence suggests that women in women-dominated classes enjoyed improved treatment relative to other state employees as a result of the 1937-38 restructuring.

Price Waterhouse v. Hopkins, ___ U.S. ___, 109 S.Ct. 1775 , 104 L.Ed.2d 268 (1989), cited by plaintiffs, is inapposite to the instant case. There the Supreme Court held that "when a plaintiff in a Title VII case proves that her gender played a motivating part in an employment decision, the defendant may avoid a finding of liability only by proving by a preponderance of the evidence that it would have made the same decision even if it had not taken the plaintiff's gender into account." Id. 109 S.Ct. at 1795. Plaintiffs here have not shown that gender played a motivating part in any classification decisions after 1930-31, so Price Waterhouse is not controlling.

Since plaintiffs have failed to prove discriminatory intent for the critical period left open by the court, and since the court has already determined in its earlier orders that intent cannot be established, plaintiffs' disparate treatment claims are DISMISSED. This order is deemed the court's findings of fact and conclusions of law pursuant to Federal Rule of Civil Procedure 52(a). All findings of fact included under conclusions of law are deemed findings of fact and all conclusions of law included among the court's findings are deemed conclusions of law.

The parties are ordered to submit a status report of not more than 15 (fifteen) pages within 30 (thirty) days of the date of this order. No replies shall be filed. In their papers, parties should address the applicability of Wards Cove Packing Co. v. Atonio, ___ U.S. ___, 109 S.Ct. 2115 , 104 L.Ed.2d 733 (1989) to any pending disparate impact claims.

IT IS SO ORDERED.

[1] Hereinafter, for purposes of this order: a designation of "JA" refers to plaintiffs' trial exhibits to

which plaintiffs and defendants have stipulated jointly; a designation of "A" refers to plaintiffs' trial exhibits; and a designation of "B" refers to defendants' trial exhibits.

[2] Mr. Cocking worked for the state Division of Personnel and Organization as a Student Personnel Assistant under an arrangement whereby graduate students received academic credit for work experience.

[3] Dr. Dickens' method established a dichotomy between female-dominated classes, and other classes, which included both male-dominated classes and "mixed" classes.

[4] Plaintiffs introduced into evidence the classification manual authored by Fred Telford, which stated in relevant part that:

The present day arrangement of classes into groups to facilitate the comprehension of the classification plan by the managing and operating officers, employees, and the public, and to show schedules of pay, was worked out by R.O. Beckman in connection with the classification and pay project carried on the by the City of Philadelphia in 1920. The refinements for a large, complex, and diversified service were later developed for the service of the State of California by Eldred M. Cocking.

F. Telford, The Telford Classification Manual: How to Make and Use an Occupational Classification of the Positions in a Public or Commercial Organization 7 (1937), Ex. JA-115. Although this comment suggests that Mr. Cocking played an important role in personnel matters, it does not clarify at which point or to what extent Mr. Cocking shaped state policies with regard to any gender-based pay differentials. Dr. Dickens conjectures that this statement referred to the 1930-31 salary-setting process, but he asserts little basis for linking Mr. Cocking to policies at issue. See Reporter's Transcript 1:65-66.

[5] Hereinafter, the designation "RT" refers to reporter's transcript.

[6] Dr. Dickens testified that this method of pay-setting tended to benefit male-dominated classes, since trades jobs tended to be male-dominated. RT 2:175. Yet, in response to a series of questions from the court, Dr. Dickens acknowledged that this effect could reflect the forces of supply and demand. RT 2:177-182, 185-86. Nevertheless, he stated that he "suspects ... that [state officials] choose the criteria that they want at a particular point in time to get the results that they want." RT 2:182 (discussing other criteria including internal relativities); see also RT 2:186 (stating that defendants would "lower the woman's job.") The court found Dr. Dickens' testimony on this point impressionistic and without any discernible foundation.

[7] Mr. Becker testified that neither Fred Telford nor any other official at the Division of Personnel and Organization assisted him in the preparation of his master's thesis. RT 5:751. He also stated that he drew primarily on secondary sources RT 5:752-53. This testimony contradicts Dr. Dickens' inference of collaboration between Messrs. Telford and Becker in regard to the thesis. See RT 1:75-77 (Dickens testimony regarding Becker's acknowledgment of Telford at beginning of thesis). The court must evaluate the evidence before it, and declines to indulge in untoward speculation.

[8] In response to a series of questions from the court, Dr. Dickens admitted that the presence of the sex category in the questionnaires may have only been used for statistical purposes. RT 1:83-84. He failed to present any specific reasons linking the inclusion of sex and the defendants' alleged discriminatory intent.

[9] For this small sample, gender was derived simply by looking at the job title. Dr. Dickens' more extensive studies created two categories: female-dominated classes, and others. Female-dominated classes were defined as those more than 70% females, and others represented the remaining classes. Exs. A-258 & A-277 (mislabelled). In Dr. Ward's studies, male-dominated classes were defined as those having more than 70% male incumbents, and female-dominated classes were defined as those with more than 70% female incumbents. See Exs. B-392-94; RT 7:1069. The difference in this methodology is discussed extensively in Section II, below.

[10] Following Jay Atwood's testimony, the court observed that this exhibit "has essentially been decimated by the testimony. By the 1930's, most of these jobs were merged. By the late 1930's there were either very few distinctions or the distinctions bear some rational relationship to the job specification." RT 9:1421-22.

[11] Although this evidence might assist plaintiffs had they attempted to make a "comparable worth" study of various jobs, it also emphasizes that they have not shouldered their evidentiary burden in this case. That burden is, as the court has cautioned plaintiffs during the various rounds of summary judgment motions, to demonstrate that intentional gender-based discrimination, apart from that existing in the market, motivated state salary-setting policies.

[12] Throughout the trial, Dr. Dickens lost credibility as a witness by his efforts to act as an advocate and his apparent unwillingness to acknowledge weaknesses in his data or his interpretation. In particular, he appeared to have difficulty recognizing the differences between academic models of behavior and empirical evidence.

[13] One example of this sexism is found in the Master's Thesis of William K. Smith, The Development and Establishment of the Classification Plan of the California State Service: 1930-32 (1938), Ex. JA-163, at 18. In discussing the personal data section on the questionnaire, Mr. Smith states that

spaces were provided for the recording of such information as the sex and date of birth of the incumbent in the job, the extent of his education, the number of months he had been in the state service, and the method of his appointment to his present position.

Id. Mr. Smith worked as a Student Personnel Assistant (on the same basis as Eldred Cocking) in the Division of Personnel and Organization from February 1931, and later assumed a civil service job with the agency.

[14] Dr. Ward used that term to describe the data collected in 1930 and the manner in which it was utilized by the state. RT 6:901.

[15] Dr. Ward explained this drop in women's salaries by stating that the influx of women into the lower clerical grades resulted in both lower median salaries and lower seniority levels. RT 7:1095-98 (relying on Ex. JA-181 at 13). Whether this effect should have been seen before the December 1939 report depends at least in part on the timing of data collection and the timing of the effects themselves. See Plaintiffs' Post-Trial Brief at 13 n. 23 (refuting Ward's hypothesis). Moreover, Dr. Ward also discussed the role high turnover among women employees played in reducing women's wages. RT 7:1098.

[16] The court rejects Dr. Dickens' calculation showing that "female-dominated classes" actually experienced a two per cent drop in average salary, relative to "other" classes, after 1938. Ex. A-258. Not only were numerous errors found in Dr. Dickens' statistical work, but his insistence on grouping all non-female dominated categories under the rubric of "other" ignores the significance of such mixed classes as Hospital Attendants, who received a large salary increase in 1938. RT 7:1083 (Ward testimony).

[17] Dr. Dickens retreated to the position that, had the state "switched completely from a ... duties and responsibilities system to a completely ... prevailing rate system" he would not have expected to see the magnitude of correlation found in his regression. RT 10:1659.

[18] The court notes that current salaries undoubtedly were used as "benchmarks" or "signals" during the 1937-38 plan restructuring. See RT 7:1111, 8:1187 (Ward testimony); RT 8:1270 (Atwood testimony). Plaintiffs' task is to show that the state reproduced the same relationships to the detriment of women employees, not simply to demonstrate some link between the past and present salaries. Any contention that the state was obligated to reconstruct its entire method of evaluating job priorities would be not only profoundly ahistorical, but impracticable.

[19] That same 1937 document also points up the fallacy of Dr. Dickens' "costs of living" hypothesis: the Personnel Board clearly calculated one cost of living, with some accounting for geographical differences. Ex. JA-133 at 5.

[20] For example, the salary relationships among the eight office machine operators suggests that duties and responsibilities, rather than the incumbents' gender, motivated salary setting. The two highest paid classes, Bookkeeping Machine Operator and Tabulating Machine Operator were female-dominated; one male-dominated and one female-dominated class were placed at the lowest salary level. Ex. B-444; RT 9:1380-83, 1387-89. The court found persuasive the testimony of Jay Atwood that the differences in pay could be explained by distinctions in the type of work. RT 9:1383-89.