HARLINGTON WOOD, Jr., Circuit Judge.

This is an appeal from the district court's denial of a preliminary injunction in a Fourteenth Amendment and 42 U.S.C. § 1983 "equal services" class action alleging that the Chicago Park District discriminatorily administers parks in black and latino neighborhoods which are inferior in condition and maintenance as compared to those parks situated in predominantly white neighborhoods. In their motion for a preliminary injunction, plaintiffs sought to restrain the Chicago Park District from adopting any budget, plan or scheme which fails to allocate funds to parks without regard to the race or ethnicity of their clientele, fails to indicate the breakdown of funds on a per-park basis, or fails to allocate funds for the immediate repair of unsafe facilities in certain of the parks; the motion also generally sought to restrain the expenditure of park funds in a discriminatory fashion. Emphasizing the weakness of the plaintiffs' evidence of discrimination and the balance of equities, the district court determined that the prerequisites for preliminary equitable relief had not been demonstrated and so denied the motion. A full trial on the merits of the complaint is scheduled to commence almost simultaneously with the issuance of this opinion. Especially in view of this imminent opportunity to argue the full merits of this case, and in view of the complicated, non-self-executing nature of the relief sought, we affirm.

I.

The centerpiece of plaintiffs' evidence at hearings on the motion was a comparative survey study of fifty-seven Chicago parks, executed over a period of several days by four individuals with no background in park administration or evaluation. The study purported to demonstrate, largely through undefined and subjective criteria, inter alia, that sixty-eight percent of the parks in white neighborhoods were "very clean" while only five percent of black and latino parks could be so described; that sixty-two percent of the parks in white neighborhoods were "very well kept," while only ten percent of black and latino parks were so maintained; that seventy-five percent of the white parks were in "good" condition compared with only fifteen percent of the black and latino parks; and that only black and latino parks were "hazardous," "poor" or "unacceptable." In addition to this survey, there was also submitted photographic evidence and depositions of two newspaper reporters attesting to racially disparate park conditions observed in their simultaneous investigations. Plaintiffs and defendants also called expert witnesses in survey evaluation to respectively defend and attack the survey's methodology and integrity.

The district court found that the evidence offered by the plaintiffs was not sufficient to establish a reasonable likelihood of success on the merits of their discrimination claim. The district court particularly focussed on the lack of methodological rigor and sophistication in the conduct of the survey, the evidence chiefly relied upon by the plaintiffs. It concluded that numerous deficits combined to undermine the survey's credibility.

First, the district court noted, the evidence concerning the survey's timing, administration and staffing did not suggest either professionalism or minimal rigor. The director and author of the survey, for example, was a political scientist located in Washington, D.C. who had no experience in the evaluation of parks or park facilities, and no acquaintance with the Chicago Park District. Even more troublesome to the district court was the fact that the survey director was never present to control or even monitor survey administration, and instead communicated through a law student who was nominally in his employ but who operated out of the offices of plaintiffs' attorney. The director played no part in the hiring or training of the four surveyors, nor did he participate in post-inspection consultations and review, thus depriving him of a means to ensure that the conduct of the survey hewed to his design. Staff selection and training, the district court found, was also not such as to inspire confidence in the survey's reliability. The four surveyors were hired after brief interviews with the director's Chicago proxy (in at least one case, directly following the interview); two of the surveyors were demonstrably without any experience in the evaluation of park services, and there is no indication that the others possessed such training. Nevertheless, after two and a half hours of oral instruction, within two days the surveyors began their inventory in teams of two, during and around the Labor Day weekend in 1981, and exceptionally busy period for the public parks. Only one visit was allocated to each park, with no followup visits to test whether observed conditions were representative over time, and even these visits were concluded rapidly, in some cases at the rate of six per day. The hasty and cursory nature of the visits and the survey design, the district court noted, also calls into question the validity of specific evaluations. To take but two examples, the surveyors took no note of the recreational programs taking place in the parks, and, although the survey elicited descriptions of the availability of park lighting, all visits were made by daylight, thus preventing the surveyors from determining whether the erected lighting was functional.

Second, the district court noted, the lack of experience and training of the surveyors was exacerbated by shoddy survey design and tabulation. Most egregiously, the evidence established that the standards for site evaluation were neither precisely defined nor applied consistently throughout the course of the survey. Although the survey form contained evaluation checklists concerning cleanliness ("very clean," "clean," "in need of cleaning," and "unacceptable") and maintenance ("very well kept," "satisfactorily kept," "in need of care," and "unacceptable"), no definitions were furnished for these terms. Not surprisingly, the testimony established that the surveyors' standards of evaluation changed significantly over time, and that as a result of consultation with other surveyors, at least two survey answers were modified after the fact to reflect a shift in relativistic perspective. The elasticity of the evaluation categories is further suggested by the response of one surveyor who labelled at least one park as "excellent" even though he had noted broken water pipes, graffiti, broken glass, blocked lights, poor blacktopping and poor sidewalks. Further fluctuation in standards was virtually guaranteed by the fact that, although each park was surveyed in teams of two, only one evaluation checklist was provided for each park, making it quite possible that the relative persuasiveness of the two surveyors, rather than consistent direct observation, helped shape the final result. The district court found this fluidity and lack of objective indicia for evaluation especially troublesome in view of the bias inherent in the sharing of results among surveyors and the testimony of at least one surveyor that he had inferred the racial designations of some parks in advance of his survey and had allowed general neighborhood features to influence his park ratings. Finally, the district court found, whatever integrity these evaluations may have had was further eroded by the survey director's telescoping of the four-point evaluation spectrum into two-column data (e.g. the collapsing of "very clean," "clean," "in need of cleaning," and "unacceptable" into the categories "very clean" and "less than very clean"). And not only were the park evaluation criteria of questionable value, the district court found, but also the validity of the survey's classification of each park's racial character was open to question. For example, the category of "white" parks included only those located in neighborhoods of whose residents more than ninety-nine percent were white, while a park was denominated "black" if located in a neighborhood of whose residents eighty percent threshold was used to designate the "black" parks in connection with some of plaintiff's deposition testimony. Further, there was a lack of rigorous correspondence between the park's character and the neighborhood's character; at least one park designated as "white" was found to have been frequented by an equal number of hispanic, black, oriental and white people.

Most of these criticisms of the survey's design and execution were supported at the hearing by one of defendants' expert witnesses, an academic specialist in urban resources who had published several articles concerning the distribution of park resources and had made a detailed study of the allocation of Chicago Park District resources. To buttress the survey's credibility, plaintiffs called as an expert witness an academic survey specialist who was retained by the plaintiffs the day before his testimony and had spent less than three hours reviewing the hundreds of pages of handwritten documents and the survey report and questionnaire.

Aside from the survey, other evidence of the discriminatory allocation of resources among Chicago parks was before the district court and, like the survey, was found to be of questionable value. Plaintiffs, for example, submitted numerous photographs purporting to depict disrepair and heaps of detritus in certain parks, but many of the sites photographed were shown to be outside of park district property, or refuse dumps which are cordoned off from the main areas of the parks; some of the purported disrepair was shown to be in fact construction in progress. Plaintiffs also introduced the depositions of two Chicago Sun-Times reporters purporting to describe racially disparate park conditions observed in their investigations, but these observations were also shown to lack even a modicum of rigor: one reporter testified that he had classified parks as white or black on the basis of his impressionistic sense of the host neighborhoods and that his visits were also short and unguided by any systematized method; the other reporter admitted to having prejudices and preconceptions concerning the relative conditions of Chicago parks and confessed that he had difficulty finding a very clean park in any Chicago neighborhood.

Defendants also offered positive evidence refuting the allegations of racially guided distribution of park resources. Defendants' expert witness, an urban resources specialist with a background in the distribution of park resources, undertook extensive review of Park District documents, including repair and maintenance orders, interviewed Park District personnel, and offered his conclusion that race was not a determining factor in the allocation of Park District resources. This neutral pattern was to some extent confirmed by the testimony of the Park District's director of safety who noted through statistical evidence that injuries and accidents were reported twice as often from "white" parks as from "black" parks. Finally, former and present Park District Commissioners testified to the general absence of discrimination against black parks and noted special affirmative park programs designed to reach disadvantaged communities.

Defendants also offered evidence pertaining to the effects of compliance with preliminary relief requested by plaintiffs: creation of a per-park budget, and the freezing of expenditures pending verification of the racially neutral character of Park District disbursements. There was evidence that a budget allocating specific sums for each of 574 parks would be too inflexible to allow response to special and unforeseen developments. And it was noted that since no expenditure is made unless specifically authorized in a valid budget, a restraint on the 1982 budget of the Park District would result in great interference with Park District operations.

II.

The legal standards governing the issuance of a preliminary injunction are too familiar to merit extended discussion. To obtain such relief, the plaintiff must demonstrate that he has no adequate remedy at law or will be irreparably harmed if the injunction does not issue; that the threatened injury to the plaintiff outweighs the threatened harm the injunction may inflict upon the defendant; that the plaintiff has at least a reasonable likelihood of success on the merits; and that the granting of the injunction will not disserve the public interest. Atari, Inc. v. North American Philips Consumer Electronics Corp., 672 F.2d 607, 613 (7th Cir.), cert. denied, ___ U.S. ___, 103 S.Ct. 176, 74 L.Ed.2d 145 (1982); Fox Valley Harvestore, Inc. v. A.O. Smith Harvestore Products, Inc., 545 F.2d 1096, 1097 (7th Cir. 1976). The standards governing review of the district court's denial of a preliminary injunction are equally clear; we may not reverse unless there is demonstrated to be a clear abuse of the trial court's discretion or clear error in the trial court's findings. Atari, Inc., 672 F.2d at 613. Even though the district court substantially adopted the defendants' proposed findings of fact and we are thus properly enjoined to exercise a somewhat more critical review of those findings, Photovest Corp. v. Fotomat Corp., 606 F.2d 704, 731 (7th Cir. 1979), cert. denied, 445 U.S. 917, 100 S.Ct. 1278, 63 L.Ed.2d 601 (1980); Flowers v. Crouch-Walker Corp., 552 F.2d 1277, 1284 (7th Cir. 1977), we must still defer in large measure to the district court's findings where, as is often the case here, they hinge greatly on the court's assessment of the credibility of observed witnesses, Klockner, Inc. v. Federal Wire Mill Corp., 663 F.2d 1370, 1375 (7th Cir. 1981); Trabert and Hoeffer, Inc. v. Piaget Watch Corp., 633 F.2d 477, 479 (7th Cir. 1980). We find ample evidence in the record to support the district court's conclusion that plaintiffs have not demonstrated the threshold requirements for a preliminary injunction.

The most serious bar to relief below, and one which we have often described as the threshold requirement which must be met before other factors may be considered, see, e.g., Reinders Bros., Inc. v. Rain Bird Eastern Sales Corp., 627 F.2d 44, 49 (7th Cir. 1980); Kolz v. Board of Education of City of Chicago, 576 F.2d 747, 749 (7th Cir. 1978), was the inability of the plaintiffs to show a reasonable likelihood of success on the merits. To prevail in a constitutionally based "equal services" case like the one before us, the plaintiff must demonstrate the existence of racially identifiable neighborhoods, substantial inferiority in the quality or quantity of the municipal service, and discriminatory intent or motive. Dowdell v. City of Apopka, 511 F. Supp. 1375 (M.D.Fla. 1981), aff'd, 698 F.2d 1181 (11th Cir. 1983); Johnson v. City of Arcadia, 450 F. Supp. 1363, 1379 (M.D.Fla. 1978).

Here, the plaintiffs offered three major clusters of evidence to meet the second requirement of its equal services claim, the inferior quality of black and latino parks: the commissioned survey, the depositions of two Chicago Sun-Times reporters, and photographic evidence. But, as the district court found, there was substantial reason to believe that each of these items were unreliable. The survey was designed and directed in absentia by a political scientist who had no previous experience in the evaluation of park systems and who had no direct contact with those who conducted the survey; the training of the equally inexperienced surveyors was perfunctory and brief; the site visits were hastily conducted over one of the busiest periods of the year; the survey itself consisted of fluid, undefined criteria which not only elicited a single response from two surveyors but also invited substantial shifts in the surveyors' relative evaluation framework and even the post hoc amendment of answers as well as bias from discussions with other surveyors and from the appearance of the surrounding communities; even these fluid tallies were subsequently collapsed, thus altering after the fact the width of the response spectrum; different criteria existed for the definition of black and white host neighborhoods; and there was evidence that even these categorizations did not accurately reflect the clientele of the respective parks. Similarly, there was evidence that some photographs submitted to prove allegedly substandard conditions in some parks in fact did not depict park property at all, or depicted only segregated refuse disposal or construction areas of the parks at issue. Finally, the deposition evidence of the newspaper reporters was found to be anecdotal, unsystematic and marred by the admitted preconceptions of the observers. In response, the defendant produced evidence based upon a detailed review of Park District practices that race was not a factor in park resource allocation, and offered evidence of heightened solicitude toward black park clientele.

Plaintiffs challenge as clear error twenty-one of the district court's findings cited by the district court as undermining the credibility of the survey and other evidence. We need not address each of these claims at length, for an examination of those challenges reveals that either they attempt to overturn conclusions reached from ample evidence, much of which concerned witness credibility, or they concern minor shadings of inference which do not affect the gist of the findings. For example, plaintiffs protest that the district court, in noting that one of the surveyors had no background in the survey or analysis of parks, ignored that the surveyor had training in participant observation of educational programs in schools and playgrounds; we are not told, however, how this background refutes the conclusion that he was not familiar with the peculiar factors involved in the maintenance, repair and design of a city-wide park system about which sophisticated qualitative judgments were required. Plaintiffs also challenge the district court's finding that the immediate supervisor of the survey was a law student in the employ of plaintiffs' attorney; while in some administrative sense this student was responsive to the survey director's instructions, it was undisputed that she was paid by plaintiffs' attorney in the first instance (although her salary was later credited against the survey director's billings) and that she operated out of plaintiffs' attorney's office, factors which suggest her closer responsiveness to the strategic needs of the litigation rather than the detachment crucial to objective survey science. Likewise, plaintiffs challenge the district court's finding that plaintiffs used differential statistical criteria for the designation of black and white parks, noting that a lower threshold for black parks was necessary to increase the potential sample of black parks; left unexplained, however, are the reasons why the criteria for "white" parks could not have been equally relaxed, or why a wholly inconsistent sixty percent threshold was used in connection with black parks discussed in the deposition.

Several other findings of the district court — e.g., that the survey director had no direct control over, or even ability to monitor, the execution of the survey; that the survey offered no precise or consistent standards for evaluation; that one surveyor was able to discern the racial designation of some parks before completing the survey; that surveyors' comparative evaluation framework changed and that questionnaire responses were modified after group consultation; and that certain questionnaire categories were recombined after data was gathered — are essentially uncontradicted by the plaintiffs, who nevertheless now dispute the level of emphasis properly attaching to each item. However, even after an independent review, and indulging every presumption in plaintiffs' favor, we believe that the district court did not err in concluding that the use of such dramatically open-ended criteria and post hoc manipulation or responses significantly undermined the credibility of the survey. Other findings challenged by plaintiffs — e.g., that the survey director did not truly believe that his total insulation from the hiring and training of surveyors and the execution of the survey was usual and proper social science methodology; that the expert witness called to buttress the flawed survey after a cursory review of the underlying materials was not credible; and that the testimony of defendants' expert witness, who had concluded after an extensive documentary review and inventory of Park District practices that no evidence of discrimination existed, was to be heeded — clearly rest upon the district court's observation of these witnesses, and as such should receive great deference. Trabert and Hoeffer, Inc. v. Piaget Watch Corp., 633 F.2d 477, 479 (7th Cir. 1980). In short, even if some of the findings concerning the gaps in the design and execution in the survey were in error, there is ample and essentially uncontradicted evidence in the record to leave us with the distinct impression that the survey and other evidence was incapable of supporting a finding of discrimination at trial. Whether or not these credibility gaps may be mended or otherwise compensated for at the full trial scheduled to commence shortly, we agree with the district court that they do not at this preliminary stage meet the plaintiffs' burden of proving that they are reasonably likely to prevail on the merits.

In addition to finding that plaintiffs were unable to demonstrate a reasonable likelihood of success on the merits, the district court also found that considerations of the public interest and the balance of hardships cautioned against the issuance of an injunction at this stage. It should be noted in this connection that plaintiffs have not intelligibly challenged the district court's finding that the proposed injunctive relief would have the short-term effect of severely hampering, if not freezing, the normal expenditure of Park District funds until such time as any proven racial disparities are eliminated. Plaintiffs assert that they merely seek to halt "the practice of masking unconstitutional discrimination behind a non-specific budget," but their requested relief by its very terms would immediately restrain the Park District from "[e]xpending or causing to be expended public funds in the allegedly discriminatory pattern," a restraint which would, if discrimination were proven, make impossible the expenditure of any Park District funds as the busy summer season approaches until the necessarily complicated process of ferreting out discriminatory allocations was completed. Paradoxically, then, the immediate effect of the requested relief would be to seriously threaten the provision of any Park District services to the very population which plaintiffs claim is currently underserved, a result hardly consonant with the public interest.

But even if a change in the budget format were the sole relief requested, it is difficult to perceive how the balance of hardships or the public interest compel relief at this preliminary stage. As a threshold matter, we note that there is no evidence in the record indicating that any practical benefit, much less an immediate one, would inure to the plaintiffs from the creation of a per-park budget; indeed, plaintiffs' survey author specifically refused to avow that any of the allegedly disparate conditions identified in his report could be linked to the Park District's aggregated budgetary categories. By contrast, the Park District produced evidence that per-park budgeting would be both impractical and inflexible in view of the likelihood of unforeseen developments and the economic need to deploy and account for park services on an areal, rather than site-specific, basis. However, even if some link between park budgetary practices and allegedly discriminatory conditions were indicated by the evidence before us, it is hard to see why it would make any practical difference whether an ultimate remedy were postponed pending the conclusion of the full trial scheduled to commence shortly. Indeed, plaintiffs concede that a conversion to per-park budgeting will require an appropriate interval of readjustment. Thus, even if plaintiffs had identified cognizable discrimination and had identified budgetary non-compartmentalization as its source, in view of the necessarily complicated and gradual implementation of any such budgetary relief that could be provided at this stage and the grave operational disruptions that would be occasioned by an immediate order placing the Park District's budget and expenditures in legal jeopardy, the balance of equities suggests that such extraordinary relief should, if justified, follow the conclusion of the imminent full trial.

III.

Plaintiffs have alternatively urged this court to reverse and remand the order denying relief because the trial judge allegedly erred in his decision, contained in an order filed several months after the denial of the preliminary injunction, not to disqualify himself pursuant to 28 U.S.C. § 455(a). However, we note that an order rendered under 28 U.S.C. § 455(a) is not an interlocutory order appealable as of right under 28 U.S.C. § 1292(a). We have also previously noted that strong judicial policy considerations caution against allowing piecemeal appeals through the review of otherwise unappealable orders in the course of considering substantively unrelated interlocutory appeals. Helene Curtis Industries, Inc. v. Church and Dwight Co., Inc., 560 F.2d 1325, 1335 (7th Cir. 1977), cert. denied, 434 U.S. 1070, 98 S.Ct. 1252, 55 L.Ed.2d 772 (1978); Rosenfeldt v. Comprehensive Accounting Service Corp., 514 F.2d 607, 611 n. 5 (7th Cir. 1975). As noted in 9 J. Moore, Federal Practice ¶ 110.25[1] at 272 (1973),

Even where an interlocutory order such as an injunction has been properly appealed and other incidental orders or questions non-appealable in themselves are sought to be reviewed, it must be remembered that the appellate court will usually review only that part of the order which relates to the injunctive relief afforded or denied and those questions basic to and underlying the specific order which supports the appeal.

Since the plaintiff will be able to obtain appellate review of the order denying disqualification after a final decision on the merits, thus allowing for a full contextual assessment of any possible appearance of impropriety, and in view of the fact that plaintiff has already obtained review of the order from this court through a mandamus proceeding, we decline to contravene settled judicial policy in order to permit intermediate appeal on this issue.

For the foregoing reasons, the district court order denying preliminary injunctive relief is affirmed.

AFFIRMED.