National Labor Relations Board v. Cell Agricultural Manufacturing Company

BOWMAN, Circuit Judge. The National Labor Relations Board (“Board”) seeks enforcement of its order requiring Cell Agricultural Manufacturing Company (“CAMCO”) to offer full reinstatement to three of its former employees, to pay back wages and benefits to laid off employees, and to bargain with the Sheet Metal Workers International Association (“Union”). We grant enforcement in part and deny enforcement in part. I. CAMCO makes agricultural machinery, specifically large potato planters and harvesters. Its two facilities are located within a quarter mile of each other in Braham, Minnesota. At its main facility (“assembly plant”), CAMCO fabricates parts and assembles them into machines. The machines are subsequently taken to its other facility (“rubber plant”) where certain parts are coated with rubber or are treated with fiberglass. CAMCO was formed in January 1990 and is wholly owned by the Cell family. Gordon Cell is the president of CAMCO and his son Gary Cell is the vice-president for operations. Prior to 1990, virtually the same business operated at the CAMCO sites under the name Dahlman, Inc. (“Dahlman”). Gordon Cell was president of Dahlman from 1983 until it went bankrupt. While the Dahlman corporation operated the .business, its employees at both facilities were represented as a single bargaining unit by the International Brotherhood of Teamsters, Chauffeurs, Warehousemen and Helpers of America. The relationship between management and the Teamsters was amicable. Since CAMCO was organized, its employees have not been represented by a union. In April 1991, CAMCO employee Scott Witte contacted Michael LaFave, a representative of the Sheet Metal Workers Association. Nineteen of the thirty-three assembly plant employees met with LaFave on April 22. One company supervisor was also present. Seventeen employees signed union authorization cards. Within the next week, five additional employees signed authorization cards. 1 Gary Cell learned about the meeting through an unnamed source on either April 22 or April 23. On April 24, Cell laid off all of CAMCO’s employees without any prior notice. Cell informed the employees that the company was reorganizing and that anyone who wanted to be rehired would have to fill out a new employment application within the next two days and interview with their current supervisor. During previous layoffs, CAMCO had never required employees to submit new applications or to be reinter-viewed. Gary Cell testified that for about a month prior to the mass layoff there had been a work slowdown at CAMCO that was jeopardizing the company’s ability to meet its manufacturing deadlines. Over the next few days, CAMCO rehired all but three of its employees: Gary Hackler, Alvin Thomas, and Witte. Prior to the layoff, all three had lobbied vigorously for increased pay and had discussed what they regarded as CAMCO’s unfair wage structure with other employees. Additionally, Thomas had served as union steward for the Dahlman company employees. CAMCO claimed, with supporting evidence, that Witte was not rehired because of chronic absenteeism and excessive socialization during working hours, that Thomas was not rehired because of excessive socialization and a refusal to weld any more frames until his pay was increased, and that Hackler was not rehired because he *393 refused reasonable work requests by Ms supervisor. All three were arguably instigating or engaging in a work slowdown. In conjunction with CAMCO’s rehiring of its other employees, the company raised the wages of eleven assembly plant employees, nine of whom had signed union authorization cards. Prior to the mass layoff, CAMCO consistently had refused requests for wage increases. In a letter to hourly employees dated April 19,1991, Gordon Cell stated that “we have got to have several good months of shipments to make up the deficits and, unfortunately, until we are profitable, we simply cannot grant wage increases.” NLRB General Counsel’s Exhibit 4. The Union filed charges against CAMCO on May 22, 1991, and filed amended charges and a complaint on June 20, 1991. The Union alleged that the mass layoff and refusal to rehire Hackler, Thomas, and Witte constituted unfair labor practices in violation of sections 8(a)(1) and (3) of the National Labor Relations Act, 29 U.S.C. § 158(a)(1), (3). 2 The Umon further alleged that the post-layoff wage increases violated section 8(a)(1). CAMCO answered the charges in a July 5, 1991, letter from Gordon Cell. On July 22, 1991, the Board ordered a hearing on the charges, and on August 5, 1991, CAMCO filed an amended and consolidated answer. The hearing was held before an Administrative Law Judge (“ALJ”) on October 1 and 2, 1991. The ALJ found that CAMCO laid off its work force, refused to rehire three employees, and granted wage increases to eleven rehired employees in response to union activities protected by the Act. The judge concluded that CAMCO’s mass layoff and refusal to rehire Hackler, Thomas, and Witte violated sections 8(a)(1) and (3) of the Act. The judge further concluded that the post-layoff wage increases violated section 8(a)(1). After determining that the assembly plant was an appropriate bargaining unit under the Act, the ALJ recommended that CAMCO be ordered to cease and desist from any further unfair labor practices, to offer reinstatement to Hackler, Thomas, and Witte, to pay any wages and benefits lost by CAMCO employees as a result of the mass lay-off, and to recognize and bargain with the Union as the agent for the assembly plant employees. The Board affirmed the findings and conclusions of the ALJ and adopted his proposed order. Cell Agricultural Mfg. Co., 311 N.L.R.B. 1228, 1993 WL 288281 (1993). The Board has petitioned tMs Court for enforcement of its order under section 10(e) of the Act, 29 U.S.C. § 160(e). II. In response to the Board’s petition for enforcement of its order, CAMCO challenges the Board’s findings that the mass layoff, the refusal to rehire three employees, and the post-layoff wage increases were unfair labor practices. We review the Board’s factual findings to determine whether they are supported by substantial evidence on the record considered as a whole. Universal Camera Corp. v. NLRB, 340 U.S. 474, 477, 71 S.Ct. 456, 459, 95 L.Ed. 456 (1951); Mississippi Transport v. NLRB, 33 F.3d 972, 977 (8th Cir.1994); see also 29 U.S.C. § 160(e)-(f). TMs is a narrow standard of review, but “[sjubstantial evidence is more than a mere scintilla. It means such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.” Universal Camera, 340 U.S. at 477, 71 S.Ct. at 459 (internal quotations omitted). WMle we might disagree with the Board’s interpretation of the evidence in a given case, we are not free to substitute our view of the facts for the *394 Board’s when it is supported by some credible evidence in the record that is not outweighed by evidence opposed to the Board’s view. See Universal Camera, 340 U.S. at 488, 71 S.Ct. at 464; United Exposition Service Co. v. NLRB, 945 F.2d 1057, 1059 (8th Cir.1991). A. As to the mass layoff, the Board found that CAMCO laid off all of its employees because some employees were engaged in union activities. The Board specifically found that CAMCO hoped the mass layoff would discourage its employees from continuing to seek representation. The Board further found that CAMCO would have continued to employ Hackler, Thomas, and Witte if CAMCO had not engaged in the unlawfully motivated mass layoff, that the termination of their employment was a direct consequence of the unlawful conduct, and that the likely result of the action was “to reinforce the employees’ fear that they would lose employment if they persisted in union activity.” Cell, 311 N.L.R.B. at 1229, 1993 WL 288281. In response, the company contends that the layoff was designed to stop a work slowdown conducted by a number of the .assembly plant employees, and that Hackler, Thomas, and Witte would have been terminated for lawful reasons even if the mass layoff had not occurred. We have reviewed the administrative record. Although there is room for doubt, we cannot say that the Board’s findings are unsupported by substantial evidence when the record is reviewed as a whole. The mass layoff occurred only two days after a union organizing meeting, and the testimony of Gary Cell established that he was aware of the meeting prior to deciding that all of the assembly plant employees would be laid off. The layoff was announced without warning in the middle of the workday. Gary Cell testified that the layoff was the company’s response to a work slowdown by dissatisfied employees. The work slowdown, however, was not mentioned by any company official at the time of the layoff; CAMCO referred only to a lack of work to be performed and a corporate reorganization. The reorganization, if there was one, consisted of a direction to supervisors to exercise authority they already had been given. Additionally, employees were told in their rehiring interviews to be satisfied with the terms of their employment or work elsewhere. Given the circumstances, reasonable employees could interpret this message as an order to forget about getting a union at CAMCO. Of the thirty-three assembly plant employees laid off on April 24, CAMCO refused to rehire only three: Hackler, Thomas and Witte. All three had signed union authorization cards, and all three had made direct complaints to management regarding employment terms. Witte was instrumental in setting up the initial meeting with the Union. Discharging prominent union supporters can effectively defeat a representation campaign. Taken in the context of a mass layoff within forty-eight hours of a union organizing meeting and the unplanned wage-rate increases granted to some of the subsequently rehired employees, discussed below, the Board found that CAMCO was motivated at least in part by anti-union sentiment. CAMCO, on the other hand, argues that these three employees would have been discharged in any event for absenteeism, low productivity, or insubordination. The three employees, however, never were told they were being discharged or why. Prior to being laid off, they were not reprimanded for, nor were they given an opportunity to correct, their alleged shortcomings. We conclude that substantial evidence supports the Board’s findings that the mass layoff was designed to discourage employee support for the Union and that Hackler, Thomas, and Witte, would have retained their jobs but for the organizing campaign. B. CAMCO also argues that the finding that its post-layoff wage increases were motivated by anti-union animus is not supported by substantial evidence on the record as a whole. We disagree. Granting a wage increase is an effective method of defeating an effort to organize a union because it removes from the table a *395 major issue around which an organizing effort is built. A wage increase, motivated by a desire to thwart a representation campaign, clearly violates the National Labor Relations Act. See NLRB v. Exchange Parts Co., 375 U.S. 405, 409, 84 S.Ct. 457, 459, 11 L.Ed.2d 435 (1964) (stating that “well-timed increases in benefits” suggest “a fist inside the velvet glove.”). The timing of the wage increase supports the Board’s finding of an unlawful motive. See McGraw-Edison Co. v. NLRB, 419 F.2d 67, 73-74 (8th Cir.1969). In this case, the wage increase also constituted a sudden reversal in policy. Just three days before nineteen assembly plant employees met with the Union’s representative, Gary Cell stated in a letter to employees that wage increases were not possible until the company became profitable. Less than a fortnight later, CAMCO raised the wages of eleven of its assembly plant employees. The Board found that the wage increases were the “carrot” to the “stick” of the mass layoff, designed- to persuade employees that their differences with CAMCO could be resolved better without a union. CAMCO argues that it was planning to revise its wage scale prior to the April 22 meeting with the Union’s representative. This uncommumcated and undocumented plan, however, does not vitiate the substantial evidence before the Board indicating that the wage increases were motivated by anti-union animus. We cannot say that the Board’s finding is unsupported by substantial evidence. III. We turn now to CAMCO’s contention that, even if the unfair labor practice charges are sustained, the Board’s finding that the as- sembly plant employees constitute an appropriate bargaining unit is not supported by substantial evidence. • As a result, CAMCO argues, the Board lacked authority to issue a bargaining order because the Union never achieved majority support in an appropriate bargaining unit. See NLRB v. Gissel Packing Co., 395 U.S. 575, 614, 89 S.Ct. 1918, 1940, 23 L.Ed.2d 547 (1969). Section 9(a) of the Act provides for the designation or selection of an exclusive bargaining representative “by the majority of the employees in a unit appropriate for such purposes_” 29 U.S.C. § 159(a). Section 9(b) of the Act requires the Board to decide whether a particular unit of employees is appropriate for exclusive representation. Id. § 159(b). The Board must consider whether the employees share a “community of interests.” Cedar Valley Corp. v. NLRB, 977 F.2d 1211, 1217 (8th Cir.1992), cert, denied, — U.S. -, 113 S.Ct. 2334, 124 L.Ed.2d 246 (1993). The relevant factors in the community-of-interests analysis include “bargaining history, operational integration, geographic proximity, copimon supervisor, similarity in job function, and employee interchange.” Id., 977 F.2d at 1217 (internal quotations omitted). The Board, adopting without comment the ALJ’s findings and conclusion, decided that the assembly plant employees constituted an appropriate bargaining unit and that the Union had achieved majority support because twenty-two of the thirty-three assembly plant employees had signed union authorization cards. After carefully reviewing the administrative record, we conclude that the ALJ misapplied the community-of-interests analysis 3 and that the ALJ’s find *396 ings on the relevant factors are not supported by substantial evidence. First, the ALJ considered but confused the geographic proximity and the functional integration factors, stating that “[w]ith regard to functional integration, the record shows a close geographic proximity between the two plants.” Cell, 311 N.L.R.B. at 1237, 1993 WL 288281. These are separate factors, and the ALJ should have considered them separately. With regard to geographic proximity, it is simply irrelevant that the two plants had separate addresses or were apparently divided by a fence; the plants were less than 1500 feet apart. While the ALJ correctly stated that geographic proximity “is not the determinative factor,” it is a factor. In this case, the substantial evidence shows that the close geographic proximity weighs in favor of a finding that the two plants share a community of interests. The ALJ’s notion of “functional integration,” limited as it apparently was to geographic proximity, is clearly too narrow. The evidence shows that every product CAMCO sells requires work in both plants. During the manufacturing process, machines frequently move from the assembly plant to the rubber plant and back to the assembly plant again. Thus there is operational integration of the two plants. See Mayflower Contract Services v. NLRB, 982 F.2d 1221, 1226 (8th Cir.1993) (including bus drivers and mechanics in single unit because “drivers operate the buses that the mechanics maintain and repair”). CAMCO employees transport machines and parts by forklift between the two plants on a daily basis. The shipping, receiving, and purchasing for the rubber plant is all conducted at the assembly plant. Payroll and other human resources functions for both plants are carried out at the assembly plant. The overwhelming evidence shows that the two plants are operationally integrated. With regard to the factor of common supervision, the ALJ noted evidence of “centralized administration of operations” such as “polieies ... uniformly applicable to employees working at both plants” and “uniform wage ranges and benefits ... applied to employees at both plants.” Cell, 311 N.L.R.B. at 1237, 1998 WL 288281. The ALJ, however, discounted this evidence because CAM-CO’s “centralized administration has not left the separate intermediate and immediate supervision of each plant Totally bereft of discretion in personnel matters.’ ” Id. at 1238, 1993 WL 288281 (quoting Spring City Knitting Co., 647 F.2d 1011, 1014 (9th Cir.1981)). We think the ALJ’s focus on intermediate and immediate supervisors who are not “totally bereft of discretion” was too narrow. Because CAMCO is a small, family-owned company, Gordon and Gary Cell have significant authority over the day-to-day operations of both plants. The record shows that while intermediate supervisors at both plants have the authority to recommend wage increases and employment actions, the final decisions are made by the Cells. Gary Cell is also responsible for disciplining employees.. Even though immediate supervisors report discipline problems to him, Gary Cell often deals directly with the employee rather than giving instructions to the supervisor. Additionally, any union or unions representing CAMCO’s employees would have to negotiate agreements with Gordon and Gary Cell. The Cells are thus the relevant supervisors for purposes of determining whether the assembly plant was an appropriate bargaining unit, and the assembly and rubber plants have these supervisors in common. The ALJ’s findings on this factor are thus unsupported by substantial evidence. The ALJ also considered, but did not give sufficient weight to, the bargaining history between Gordon Cell and CAMCO’s employees, calling it “relatively remote” and noting that a “different employer operates the facility.” Cell, 311 N.L.R.B. at 1237, 1993 WL 288281. At the time of the unfair labor practices, less than two years had passed since the employees at both plants had been represented in a single bargaining unit. During the intervening period, these employ *397 ees did not split into separate bargaining units; they were simply unrepresented by a union. The long, two-plant bargaining history is substantial evidence weighing against a finding that a single-plant unit is an appropriate bargaining unit. The ALJ’s cursory consideration of the similarity in job function between assembly plant and rubber plant employees also misses the mark. The ALJ noted that the types of machines used in the two plants are substantially different, and then concluded that “[djistinct operations are performed in each of those plants_” Cell, 311 N.L.R.B. at 1238, 1993 WL 288281. But the function of the employees at the two plants is the same: they are employed to produce potato planters and harvesters. Any assembly line will have substantially different machines at different points on the line; the function of the CAM-CO employees, like the function of employees on an automobile assembly line, does not change simply because the assembly line extends from one plant to another nearby plant. 4 The ALJ erroneously concluded that “distinct operations” are performed at the two plants. In this case, the similarity factor weighs against a single-plant bargaining unit. With regard to employee interchange, we agree with the ALJ that the record in this case shows this factor weighing in favor of a single-plant bargaining unit. This factor alone, however, cannot constitute substantial evidence supporting the Board’s overall finding that a single-plant unit is appropriate. The overwhelming evidence put forward by CAMCO with regard to the other factors shows that a single-plant unit is an inappropriate bargaining unit in this ease. After a proper application of the community-of-interests test, the substantial evidence supports only one plausible conclusion: the appropriate bargaining unit for CAMCO’s employees must include both the assembly plant and the rubber plant. Thus we hold that the Board’s finding that the assembly plant employees formed an appropriate bargaining unit is not supported by substantial evidence on the record as a whole. Our conclusion regarding the appropriate bargaining unit means that the Board’s bargaining order cannot stand. The Board has authority to issue a bargaining order in response to “less extraordinary” unfair labor practices 5 only “when there is also a showing that at one point the union had a majority.” Gissel, 395 U.S. at 614, 89 S.Ct. at 1940. Our decision that the sole appropriate bargaining unit in this case consists of both the assembly plant and the rubber plant employees means that the Union needed authorization cards from twenty-four of the forty-six employees to achieve the required majority support. The Union, however, obtained cards from only twenty-two employees. Because the Union never achieved a majority, the Board lacked the authority to issue a bargaining order, and we therefore must decline to enforce the order. IV. While our decision in this case does not require us to consider the propriety of the Board’s bargaining order had the Union attained majority support in an appropriate bargaining unit, we do want to comment on that part of the Board’s opinion declaring that “[ejvidence concerning employee turnover [after unfair labor practices have been committed] is an irrelevant consideration when assessing the propriety of a Gissel bargaining order.” Cell, 311 N.L.R.B. at 1229, 1993 WL 288281. For over two decades, the Board has maintained a running feud with the courts of appeals regarding whether changed *398 circumstances after an employer’s unlawful conduct, such as employee turnover or the passage of time, which tend to erode majority support for a union, are relevant to the Board’s decision to impose a bargaining order. See, e.g., NLRB v. American Cable Sys., 427 F.2d 446, 448-49 (5th Cir.), cert. denied, 400 U.S. 957, 91 S.Ct. 356, 27 L.Ed.2d 266 (1970). The Board argues that they are not relevant because Gissel contemplated that the propriety of the bargaining order would be “appraised as of the time of the commission of the unfair labor practices, and not currently.” See Gibson Products Co., 185 N.L.R.B. 362, 363 (1970). All but one of the Courts of Appeals that have considered the issue, however, have held that such changed circumstances are relevant to the Board’s decision to issue a bargaining order. See NLRB v. So-Lo Foods, 985 F.2d 123,128-29 (4th Cir.1992) (noting that significant, normal employee attrition is relevant); NLRB v. LaVerdiere’s Enterprises, 933 F.2d 1045, 1054-55 (1st Cir.1991) (passage of time); Texas Petrochemicals Corp. v. NLRB, 923 F.2d 398, 405-06 (5th Cir.1991) (same); M.P.C. Plating v. NLRB, 912 F.2d 883, 888 (6th Cir.1990) (passage of time and employee turnover); Amazing Stores v. NLRB, 887 F.2d 328, 331 (D.C.Cir.1989) (employee turnover), ce rt. denied, 494 U.S. 1029, 110 S.Ct. 1477, 108 L.Ed.2d 614 (1990); Impact Indus, v. NLRB, 847 F.2d 379, 383 (7th Cir.1988) (passage of time, employee turnover, and change of management); NLRB v. Armcor Indus., 535 F.2d 239, 246 (3d Cir.1976) (Board must consider present conditions), application after remand 588 F.2d 821 (1978) (enforcement denied by equally divided court); see also Piggly Wiggly v. NLRB, 705 F.2d 1537, 1543 n. 9 (11th Cir.1983) (holding that changes up to time of hearing are relevant); but see NLRB v. Bakers of Paris, 929 F.2d 1427, 1448 (9th Cir.1991) (refusing to consider employee turnover and passage of time). We have not directly addressed the issue of whether changed circumstances are relevant, but we note that most of our sister circuits have rejected the Board’s position. We are persuaded by this authority from other circuits. Accordingly, we adopt the view that the Board must consider any change of circumstances, including the passage of time, employee turnover, and voluntary statements of cooperation by company officials, when deciding whether to issue a bargaining order. 6 In this case, it is apparent that the Board did not set aside its traditional view regarding the relevance of changed circumstances. Moreover, to the extent the Board gave any consideration to CAMCO’s changed circumstances, it did not carefully weigh the evidence of employee turnover. The Board’s opinion supports our assessment by erroneously connecting the cause of CAMCO’s high turnover rate to its violations of the Act. 7 The undisputed evidence shows that only one in three CAMCO employees stayed at CAM-CO more than a year even prior to CAMCO’s unfair labor practices. Additionally, at the hearing in this case the ALJ stated on the record that evidence of high turnover is “not an argument that’s going to have great force with me or the Board,” even though “it has sometimes with the circuit courts.” We hardly need point out that if an argument “has force” with the courts of appeals, it should “have force” with the Board and its ALJs as well. *399 The Board also failed to consider post-layoff actions by CAMCO that helped to create the conditions necessary for a free and fair representation election. After unfair labor practice charges were filed, Gordon Cell met with employees from both plants and told them that he was personally neutral on whether the employees should vote for the Union and that CAMCO could work with a union. We cannot find anything in the record to indicate that the ALJ or the Board gave any weight to these statements. In fact, the ALJ’s opinion states that “[t]here has been no showing that [Gordon and Gary Cell] have experienced any change in the attitude that led them to authorize and implement the unlawfully motivated lay-offs [and other unfair labor practices].” Cell, 311 N.L.R.B. at 1239, 1993 WL 288281. Like evidence of employee turnover, the Board should have considered any statements made to the employees by the company’s manager-owners that could help to ensure a free and fair election. In light of the changed circumstances at CAMCO since the commission of the unfair labor practices at issue in this case, it is not at all clear that the Board would have been within its discretion in issuing a bargaining order even had the Union obtained a card majority in an appropriate bargaining unit. We see in the record nothing approaching a persuasive showing that a free and fair representation election could not now be held at CAMCO. However, as our holding on the bargaining-unit issue renders the question of changed circumstances moot, we do not rest our decision on this ground. V. To sum up, we grant enforcement of that part of the Board’s order requiring CAMCO to cease and desist from any and all unfair labor practices, to offer to reinstate Hackler, Thomas, and Witte, and to make whole assembly plant employees, including Hackler, Thomas, and Witte, for any pay or benefits lost as a result of the unlawful layoff of April 24. We decline, however, to enforce the Board’s bargaining order. Footnotes [1] Of the five additional authorization cards, four were signed after the mass layoff, discussed below. Two were signed the day of the layoff, one was signed on April 27, and one final card was signed on April 29. [2] Section 8 of the Act provides in relevant part as follows: It shall be an unfair labor practice for an employer — • (1) to interfere with, restrain, or coerce employees in the exercise of rights guaranteed in section 157 of this title; iH í¡! sfc (3) by discrimination in regard to hire or tenure of employment or any term or condition of employment to encourage or discourage membership in any labor organization.... 29 U.S.C. § 158. Section 7 of the Act guarantees the “right to self-organization, to form, join, or assist labor organizations” and "to bargain collectively through representatives of their own choosing, and to engage in other concerted activities for the purpose of collective bargaining.” Id. § 157. [3] The ALJ's misapplication of the community-of-interests analysis was compounded by a misplaced reliance on " 'a presumption that a single plant is an appropriate bargaining unit.’ " Cell, 311 N.L.R.B. at 1237, 1993 WL 288281 (quoting Spring City Knitting Company v. NLRB, 647 F.2d 1011, 1014 (9th Cir. 1981)). We note that the single-plant presumption has not been adopted by this Court. Cf. Cedar Valley Corp. v. NLRB, 977 F.2d 1211, 1217 (8th Cir.1992) (applying community-of-interests analysis to determine whether employees of one employer located in different geographical areas can have different exclusive representatives), cert. denied, - U.S. -, 113 S.Ct. 2334, 124 L.Ed.2d 246 (1993); Kansas City Terminal Elevator Co. v. NLRB, 697 F.2d 269, 270 (8th Cir. 1983) (per curiam) (holding that single plant was appropriate unit after applying community-of-interests analysis). Such a presumption is at odds with the community-of-interests analysis our Court has favored, and we decline to adopt it. Even if we were to adopt the presumption advocated by the Board, the presumption can be rebutted by a showing that the employees of separate plants “have been merged into a more comprehensive unit by a history of bargaining, or that employees lack any meaning *396 ful separate identity or community of interest.” Brief for the NLRB at 37. We think that our analysis of the community-of-interests factors below clearly shows that CAMCO met its burden and rebutted the single-plant presumption. [4] The geographic proximity factor adequately considers the effect of physical separation on the employees' community of interests. [5] The Gissel opinion divides the universe of bargaining order cases into three categories: (1) “exceptional cases marked by outrageous and pervasive unfair labor practices”; (2) "less extraordinary cases marked by less pervasive practices which nonetheless still have the tendency to undermine majority strength and impede the election processes"; and (3) cases of "minor or less extensive unfair labor practices” having a "minimal impact” on the election machinery. Gissel, 395 U.S. at 613-15, 89 S.Ct. at 1939^10 (internal quotations omitted). The parties agree that the unfair labor practices at issue in this case are not exceptional and properly fit into the second category. [6] We also note that the Board further compounded its error when it stated that "[i]f ... employee turnover can be relevant ..., the burden must be on the employer to demonstrate why employee turnover should preclude the imposition of a bargaining order.” Cell, 311 N.L.R.B. at 1229, 1993 WL 288281. That statement of the law is wrong. The preferred method of selecting a representative under the Act is an election. NLRB v. Ely's Foods, 656 F.2d 290, 293 (8th Cir. 1981). If the General Counsel wants the Board to order a company to bargain without an election, it must demonstrate "that the possibility of erasing the effects of past practices and of ensuring a fair election ... by the use of traditional remedies, though present, is slight....” Gissel, 395 U.S. at 614, 89 S.Ct. at 1940; see also Avecor v. NLRB, 931 F.2d 924, 939 (D.C.Cir. 1991) (Board must "explain convincingly why the turnover has not cleared the air of the unfair practices and why traditional remedies could not reasonably ensure a fair election”), cert. denied, -U.S. -, 112 S.Ct. 912, 116 L.Ed.2d 812 (1992). [7] "Indeed, were the Respondent law abiding, who is to say that turnover would not be substantially reduced?" Cell, 311 N.L.R.B. at 1230, 1993 WL 288281.