This petition for enforcement of an order of the National Labor Relations Board (NLRB or Board) raises two principal issues. The first has its genesis in a mass layoff by the emplpyer in this case. Although an uncommon phenomenon, in labor relations in this country during the period of industrial growth preceding World War II, mass layoffs have not been an infrequent occurrence in the intervening years. The issue is whether the employer, Vemco, Inc., violated the National Labor Relations Act (NLRA or Act) by permanently laying off 60 employees allegedly in retaliation for their union activities. The second issue is whether the employer’s conduct during the organizing campaign by the International Union, United Automobile, Aerospace and Agricultural Implement Workers of America (UAW) so disrupted the election process as to justify a bargaining order.1 The Board concluded that the layoff constituted an unfair labor practice and that the employer’s conduct during the organizing campaign justified the harsh remedy of a bargaining order. *1472The Board has petitioned this court for enforcement of its order.2 We deny enforcement of those portions of the Board’s order relating to the layoff and the bargaining order. We modify the remaining provisions of the NLRB’s order relating to undisputed violations of the Act to include the traditional remedy of a rerun election and order the enforcement of these remaining provisions as modified.
I.
In 1989, the UAW filed charges with the NLRB alleging that Vemco was engaging in certain unfair labor practices in violation of the NLRA during the course of the UAW’s campaign to organize Vemco’s employees. After hearings, an administrative law judge (AU) found that Vemco had engaged in numerous unfair labor practices related to the UAW’s organizing efforts. He therefore ordered Vemco, inter alia, to reinstate with back pay employees laid off allegedly because of union activity and to bargain on request with the UAW. On review, the NLRB affirmed the AU’s decision with only slight modification.'
Vemco, located in Grand Blanc, Michigan, supplies the automotive industry with large plastic exterior car parts requiring injection molding and sophisticated paint application. Vemco’s executive management consists of Larry Winget, principal owner and president, Michael Torakis, vice president of finance and administration and secretary/treasurer, and James Schütz, vice president of manufacturing. Schütz is responsible for the day-to-day operations of Vemco, which first commenced operations in the spring of 1988. Winget, Torakis, and Schütz together set corporate policy and engage in corporate planning.
Vemco’s three operations are. molding, paint, and assembly. The paint plant is considered state of the art. For the 1989 model year, the paint plant itself was subdivided into cladding and fascia lines, with the fascia line further split into a masking area and a paint area.3
At the time it opened its doors and throughout the-1989 automotive year, Vem-co had only one customer, Buick-Oldsmobile-Cadillac Division of General Motors (BOC). BOC contracted with Vemco for fascias and claddings for several of its car models. The contract called for delivery to commence in August 1988, with projected quantity requirements to increase weekly over the fall of the year as BOC increased its own manufacturing. The contractual penalty for a failure to deliver according to the industry practice of just-in-time (JIT) shipping causing downtime on a BOC assembly line ran as high as $25,000 per minute of downtime.4
Vemco’s high-tech paint system did not function as planned when Vemco began operations because of major technical problems in many of the components of the system. The company that designed and built the paint system eventually walked off the project, leaving Vemco to identify and correct these problems with its own personnel and other outside contractors. Ultimately, the system had to be redesigned and rebuilt at the same time BOC’s increasing delivery requirements had to be met.
As a result of its technical problems, Vemco was unable to produce a product that consistently met quality standards the first time through and had to rework enormous numbers of unacceptable parts. Because of the severe economic consequences of a failure to timely deliver to BOC, Vem-co ended up with many more employees working much longer hours than originally forecast. By mid-December, Vemco was running a second shift on both the fascia *1473and cladding lines, partially to relieve the 60 to 84 hour weeks its employees had been working and partially in anticipation of orders from Ford Motor Company, which, if obtained, would have required a fully trained second shift to produce.
By early 1989, Vemco succeeded in “debugging” in part its paint operation. As a result, workhours decreased and productivity increased. These productivity gains resulted in a shortage of workhours for Vem-co’s employees. By mid-January, both shifts on the fascia lines were reduced to a four-day workweek, and Vemco considered going to a three-day workweek. At this point, employees expressed unhappiness with the sharp decline in hours and apprehension over further hourly reductions and possible layoffs.
Vemco has a non-union philosophy that is clearly expressed in its Personnel Policies Handbook that all employees receive and sign for at the time of their orientation. Some Vemco employees, however, became interested in unionizing soon after the company opened, and by late September employees had made telephone inquiries of the UAW. An initial union meeting was held in early October 1988, attracting 50 employees. Periodic meetings were held thereafter, including more or less weekly gatherings between January and March of 1989. Both Schütz and Torakis were aware of general union activity by October 1988. Line supervisors, who were also aware of this general union activity, committed several unfair labor practices during the period between Vemco’s start-up and the March 1989 mass layoff.
II.
A. STANDARD OF REVIEW
The NLRB’s findings of facts, as well as its application of law to fact, may not be disturbed where substantial evidence on the record taken as a whole supports the Board’s findings and conclusions. Universal Camera Corp. v. NLRB; 340 U.S. 474, 488, 71 S.Ct. 456, 465, 95 L.Ed. 456 (1951); Turnbull Cone Baking Co. v. NLRB, 778 F.2d 292, 295 (6th Cir.1985), cert. denied, 476 U.S. 1159, 106 S.Ct. 2277, 90 L.Ed.2d 720 (1986); 29 U.S.C. § 160(e), (f). As a guide to determining whether the substantial evidence standard has been met, this court stated:
Evidence is considered substantial if it is adequate, in a reasonable mind, to uphold the decision. Universal Camera, 340. U.S. at 477 [71 S.Ct. at 459], The appellate court should consider the evidence contrary to the Board’s conclusions, but may not conduct de novo review of the record.. Union Carbide Corp. v. NLRB, 714 F.2d 657, 660 (6th Cir.1983).
Turnbull, 778 F.2d at 295. Questions of law, however, are reviewed de novo. NLRB v. C.J.R. Transfer, Inc., 936 F.2d 279, 281 (6th Cir.1991).
B. REVIEW OF UNCONTESTED . VIOLATIONS
1. Uncontested Violations and the Contested Use of Protected Speech
The General Counsel of the NLRB (GC) points out that Vemco does not contest various section 8(a)(1) violations occurring before both the mass layoff of March 17, 1989, and the representation election at the end of September 1989. Vemco also does not challenge section 8(a)(1) and (3) viola-, tions involving an employee named Greg Hall. We therefore will not review these violations to ascertain whether there is substantial record evidence to support them. Hyatt Corp. v. NLRB, 939 F.2d 361, 368 (6th Cir.1991).
Vemco, however, has vigorously objected to the Board’s reliance on certain employer speech to find violations of section 8(a)(1) because this speech was allegedly protected by section 8(c) of the Act.5 Because the Board then used violations based on this *1474.allegedly protected speech as evidence to support both the layoff violation and the bargaining order, the uncontested violations must be reviewed to the extent that they may contain protected speech. If protected speech is found, the evidentiary value of these uncontested violations supporting the alleged layoff violation and the bargaining order must be reconsidered. This may then affect our determination of whether the Board’s conclusions regarding the layoff and the bargaining order are supported by substantial evidence.
2. The Evidentiary Value of Protected Speech
Statutory interpretation is a matter of law that we review de novo. See, e.g., United States v. Hans, 921 F.2d 81, 82 (6th Cir.1990). Section 8(c) of the Act basically defines protected speech as the expression of views, argument, or opinion, whether in spoken or written form, that contains no threat of reprisal or force or promise of benefit. According to section 8(c), protected speech “shall not constitute or be evidence of an unfair labor practice.” The NLRB generally takes the position that although protected speech, such as an employer’s expression of its views or opinions against a union, cannot be deemed a violation in and of itself, i.e., cannot be used as primary or direct evidence of a violation, it can nonetheless be used as background evidence of anti-union animus or anti-union motive on the part of the employer. See, e.g., Holo-Krome Co. (Holo-Krome I), 293 NLRB 594 (1989), enforcement denied, Holo-Krome Co. v. NLRB (Holo-Krome II), 907 F.2d 1343 (2d Cir.1990). The circuits are currently split regarding the evi-dentiary value of protected speech, and this circuit has not yet addressed the issue directly.
Some circuits rely on the plain meaning of the statute filtered through a narrow, but plausible, reading of the legislative history to preclude any use whatsoever of protected speech. See, e.g., Holo-Krome II, 907 F.2d at 1346 (holding the Board’s use of employer’s protected expressions of opinion against union as a basis for finding animus to be contrary to section 8(c)). Other circuits agree with the Board’s interpretation and therefore allow protected speech to be used as background evidence of animus. See, e.g., Hendrix Mfg. Co. v. NLRB, 321 F.2d 100, 103 (5th Cir.1963) (permitting the Board to consider the employer’s clear expression of opposition to the union as background in order to determine motivation for management’s conduct).6
Despite the noncommittal stance of the Board on this issue in the present case,7 we analyze Vemco’s alleged protected speech under the Holo-Krome I interpretation of section 8(c). Therefore, unprotected speech, whether constituting an independent violation of the Act or not, will be considered as direct evidence of animus in the context of the layoff violation; we regard protected speech, however, only as background in our determination of animus.
III.
Before commencing production, Vemco had projected personnel requirements total-*1475ling 270 employees, 180 in direct labor and 90 in support positions, working a single shift, to meet the delivery requirements of the BOC contract. The financial success of the company depended on meeting these projections. Torakis prepared a report in early November 1988 regarding the financial performance of the company. With 332 people currently employed, Torakis’s report pinpointed a reduction in the labor force of 63 employees as an opportunity to improve the financial performance of the company at its current sales volume. No action was undertaken at that time to reduce the work force, however, because of Vemco’s continuing start-up difficulties and delivery requirements.
The issue of overstaffing came up again at a business planning meeting in January 1989. Various goals were discussed for the 1989 fiscal year, including the necessary goals of increasing productivity and decreasing the work force at Vemco. Management targeted 1989 for work force reductions.
Because BOC had been intensely involved in Vemco’s difficult start-up period and had an economic stake of its own in Vemco’s plans to resolve its problems, it requested a Productivity Improvement Plan from Vemco. Vemco submitted the Plan at the end of January 1989 based on the results of the January meetings. In the Plan, Vemco set forth its goal of reducing the size of its work force and made a commitment to BOC to share productivity gains with it in the form of a price reduction on future orders.
Informal meetings continued throughout February and early March among Winget, Torakis, and Schütz pertaining to the productivity increases in January and February as they related to both the costs of overstaffing and the effect of overstaffing on productivity. Torakis and Schütz both noted that continued gains in productivity were threatened as employees realized that increased productivity meant fewer work-hours or probable layoffs. On March 8, Winget, Torakis, and Schütz collectively acknowledged that the anticipated Ford work would not be forthcoming. Torakis and Schütz testified that the three officers therefore agreed to a layoff, which they tentatively scheduled for March 17, to cut back to minimum staffing levels. To implement the layoff, Schütz first directed the plant managers to calculate the minimum number of employees necessary in each department to staff the plant to meet BOC’s delivery requirements. From these figures, the managers then calculated the number to be laid off.
In the meantime, on Sunday, March 12, employee Greg Hall told his then girlfriend Debbie Reed that the UAW was planning to send a letter officially notifying Vemco of the organizing campaign and providing a list of employees on the organizing committee. Reed, an assistant to Vemco’s Human Resources Administrator, Paula McIntyre, told McIntyre about the letter the following day, March 13. McIntyre asked Reed when the letter would arrive, but Reed did not know. Reed later put the question to Hall and on Tuesday, March 14, she told McIntyre that the letter was tentatively scheduled to arrive on either Friday, March 17, or Monday, March 20.
Betty Harrison, the UAW representative handling the Vemco campaign, however, testified that until March 16 she had no plans to deliver the letter before March 20. She changed her original plan, however, when she surmised from information she indirectly received on March 16 from an inside source at Vemco that a layoff had been scheduled for March 17. She therefore accelerated the delivery date of the letter to March 17.
On March 14, McIntyre had Reed run an attendance report on all production employees in preparation for the selection of the employees to be laid off. While line supervisors and plant managers worked on the number to be laid off, on March 15, Schütz expressed to Winget and Torakis his reservations about implementing a one-shot layoff. He feared that cutting back too far might leave Vemco unable to meet BOC’s delivery requirements; he suggested instead a three-step staggered layoff starting on March 20. Torakis and Winget believed the risk of an excessive cut-back could be *1476avoided by careful calculation; the three company officers decided after additional discussion to hold to the original plan. To-rakis prepared the draft of the letter notifying targeted employees of their immediate, permanent layoff,' which counsel reviewed and revised.
During the week preceding the layoff, Reed and McIntyre frequently spoke about layoff preparations. Reed also overheard McIntyre speak with other Vemco personnel regarding the layoff. Although not directly involved in the selection process, McIntyre made several comments to Reed prior to the layoff, expressing her hope that some of the people laid off would be union people. After the layoff, she indicated that such was the case. Once, when speaking to one of the plant managers and two of his line supervisors, she said she wanted them to think of people affiliated in some way with the Union as they made their layoff selections. One of the line supervisors responded, however, that “it would be hard to do.” On March 16, when Reed asked McIntyre if the layoff was in response to the imminent arrival of the union letter, McIntyre responded that the layoff had not been called because of the letter, but that it had been moved up a week. On this same day, McIntyre instructed two clerical employees not to accept any certified mail or sign for anything at the reception window.
After the plant managers and supervisory staff determined the number to be laid off for each line, they selected the employees to be terminated using a five-point evaluation model based on attendance, cooperation, initiative, job knowledge, and quality of work with consideration of attendance and disciplinary reports drawn from personnel records. At 2:30 p.m. on March 17, the UAW’s Harrison hand-delivered to Schütz the letter announcing the Union’s organizing campaign and naming 50 employees on the organizing committee. The names of all but three employees to be terminated had already been submitted to McIntyre by that time; the three remaining names were submitted by 4:30 p.m., and two of the three appeared in the letter. Reed and another clerical employee prepared termination letters and mailed them by 5:00 p.m. that afternoon. Vemco laid off a total of 60 production workers.
A. APPLICABLE LAW
Section 7 of the NLRA guarantees employees the right to organize and choose representatives for collective bargaining.8 Section 8(a)(1) of the Act implements this guarantee by making it an unfair labor practice for an employer “to interfere with, restrain, or coerce employees” in the exercise of their section 7 rights; section 8(a)(3) in pertinent part makes it an unfair labor practice for an employer to discriminate in regard to the tenure of an employee’s employment to discourage membership in a labor organization.9 Because a discriminatory termination or layoff interferes with or coerces employees in the exercise of their section 7 rights, a breach of section 8(a)(3) automatically violates section 8(a)(1).10
A section 8(a)(3) violation consists of “a discharge or other adverse action that is based in whole or in part on antiunion animus” or, alternately worded, section *14778(a)(3) is violated if “the employee’s protected conduct [under section 7 is] a substantial or motivating factor in the adverse action.” NLRB v. Transportation Management Corp., 462 U.S. 393, 401, 103 S.Ct. 2469, 2474, 76 L.Ed.2d 667 (1983). Breaking the violation down into its simplest elements, the GC must show that animus, whether toward the union or an employee’s protected conduct, produced the adverse action in question. In this case, the GC charges that the March 17 mass layoff constituted the necessary employment action under section 8(a)(3). Cases discussing section 8(a)(3) often speak of motive, with motive encompassing both employer animus and causation, and these are the elements at issue in this case. Animus is often an elusive state of mind, and discerning it and motive “is a tricky business.” White, supra note 7, at 1.
In Transportation Management, the Supreme Court approved of the section 8(a)(3) analysis propounded by the Board in Wright Line, A Division of Wright Line, Inc., 251 NLRB 1083 (1980), enf'd, 662 F.2d 899 (1st Cir.1981), cert. denied, 455 U.S. 989, 102 S.Ct. 1612, 71 L.Ed.2d 848 (1982), for “cases involving charges of employment actions motivated by anti-union animus and employer protestations of legitimate reasons for the actions.” Birch Run Welding & Fabricating, Inc. v. NLRB, 761 F.2d 1175, 1179 (6th Cir.1985). The Birch Run court reiterated the Wright Line methodology in teriiis particularly applicable to the present case:
In a case involving lay-offs, the General Counsel must prove that anti-union animus partially motivated or contributed to the lay-off decision. If this prima facie ease is established, then the employer must show by a preponderance of the evidence that the employees would have been laid-off even if they had not engaged in protected activity.
Id.
B. THE GENERAL COUNSEL’S PRIMA FACIE CASE
The Board can rely on both direct and circumstantial evidence to determine employer motive. Turnbull Cone Baking Co. v. NLRB, 778 F.2d 292, 297 (6th Cir.1985), cert. denied, 476 U.S. 1159, 106 S.Ct. 2277, 90 L.Ed.2d 720 (1986). The Board found that statements made by Human Resource Administrator McIntyre during layoff preparations, together with several pre-layoff section 8(a)(1) violations committed by line supervisors, established anti-union animus. The section 8(a)(1) violations are summed up as follows:
1. Coercive interrogation and threat of plant closure involving one supervisor and one employee;
2. Three incidents of threat of job loss involving two supervisors and three employees;
3. Coercive interrogation by a supervisor of one employee in front of a second employee; and
4. Overbroad no-solicitation rule involving one supervisor and her line employees.
Although Vemco raises questions regarding some of the pre-layoff section 8(a)(1) violations on the basis of protected speech, review of these violations shows that, although protected speech was sometimes mixed with unprotected speech in the incidents described, the violations themselves do not rest on protected speech. These violations can therefore be used, along with McIntyre’s unprotected speech, as primary evidence in our review for substantial evidence of animus in the context of the alleged layoff violation. In addition to this evidence of animus, the Board may appropriately consider under section 8(c) the anti-union policy expressed in Vemco’s personnel handbook as background evidence of animus. ■ A review of this entire body of evidence shows that a.finding of anti-union animus is clearly supported by substantial evidence on the record taken as a whole and should therefore not be disturbed.
The remaining and critical element is causation. Preliminarily, as a matter of causation, adverse employment action generally will not be held to violate section 8(a)(3) unless the employer had knowledge of the targeted employee’s union activities or pro-union sentiments at the time it took *1478the action. Turnbull, 778 F.2d at 296; Birch Run, 761 F.2d at 1179. In Birch Run, this court expressed an alternate theory, however, that is applicable to mass layoff situations. That theory permits the GC to prevail
by showing that the employer ordered general lay-offs for the purpose of discouraging union activity or in retaliation against its employees because of the union activities of some! This theory is viable even though some employees who actually opposed the union ... were laid-off with their pro-union counterparts. Moreover, the theory can be valid even though not all union adherents were laid-off.... The focus of the theory is upon the employer’s motive in ordering extensive lay-offs rather than upon the anti-union or pro-union status of particular employees.
Id. at 1180 (citations omitted).
The GC proceeded in the present case under this alternate theory. The GC did not show employer knowledge with respect to each individual employee laid off. In fact, only 17 of the 60 employees laid off were included on the list of employees belonging to the union organizing committee. Vemco challenges this failure to show specific knowledge as to each laid-off employee as error. However, this challenge is meritless under Birch Run.
There is, nonetheless, a deficiency in the way the GC uses the mass layoff theory in this particular case. In Birch Run, the issue was whether specific knowledge of each laid-off employee’s union sentiments was necessary to support a section 8(a)(3) violation or whether an employer’s general knowledge of union activity would suffice in the context of a mass layoff. The court concluded that general knowledge suffices in the context of a mass layoff in a case where that very general knowledge was the asserted trigger for the layoff. The present case, however, is distinguishable from Birch Run because Vemco’s general knowledge of union activity was not the asserted trigger for the layoff.
The GC contends that the imminent arrival of the union letter, and the increase in union activity it disclosed, triggered Vemco’s attempt to “beat the union to the punch” by implementing the layoff. Therefore, although we agree with the GC that it is not necessary to show knowledge of each specific employee’s union sentiments for the Board to prevail,11 a showing of Vemco’s general knowledge of union activity is insufficient in itself to prove the employer’s motive for the layoff in this particular case. Instead, to satisfy the knowledge requirement of causation, the GC must show that Vemco knew of the forthcoming letter when Winget, Torakis, and Schütz decided to reduce the work force. Torakis and Schütz testified that they set the March 17 layoff date on March 8; additionally, they testified that on March 15 Schütz proposed to Torakis and Winget that the layoff be moved back to the following week, but this suggestion was rejected. McIntyre first heard of the union letter on March 13 and did not know when it would arrive until the following day. Finally, Reed testified to a statement made by McIntyre on March 16 that although the letter did not cause the layoff, it caused the layoff to be moved up a week.
Regardless of the GC’s erroneous reliance on Vemco's general knowledge of union activity, the ALJ, after specifically *1479crediting Reed’s testimony and discrediting the testimony of company officials where it contradicted her, Vemco, Inc., 304 NLRB No. 118, 1991 WL 181869, at *14 (NLRB Aug. 27, 1991), concluded:
It was only after learning of the Union’s plan to send it a letter (which would identify Union supports [sic] and tend to offer them some degree of legal protection against retaliation), that [Vemco] decided to hasten its consideration of an ongoing business review of staffing levels and to move up the date of its action in an attempt to “out maneuver” the Union in its bid to secure the support of a majority of the company’s employees.
Id. at *16. Thus, the AU determined by making a specific credibility evaluation that Vemco knew of the letter before it decided on the layoff date. Ordinarily, this court will not disturb credibility evaluations by an AU who observed the witnesses’ demeanor. Turnbull, 778 F.2d at 295 (quoting NLRB v. Baja’s Place, 733 F.2d 416, 421 (6th Cir.1984)). We will therefore not disturb the AU’s factual finding on this threshold knowledge requirement of causation.
More generally, causation is a part of what the courts refer to as motive; however, ascertaining the motivation for an employer’s action is not simple. Rarely is there direct evidence that the employer’s animus actually caused a layoff decision, and the Board and courts look for illumination to a variety of factors from which anti-union motivation may reasonably be inferred. Several of these factors are:
1. A company’s expressed hostility towards unionization together with knowledge of the employees’ union activities;
2. Inconsistencies between the proffered reason for the discharge and other actions of the employer;
3. Disparate treatment of certain employees compared to other employees with similar work records or offenses;
4. A company’s deviation from past practice in implementing the discharge; and
5. Proximity in time between the employees’ union activities and their discharge.
Turnbull, 718 F.2d at 297; Birch Run, 761 F.2d at 1179. To determine whether substantial evidence supports the Board’s finding that anti-union animus was a motivating factor for the March layoff, it is necessary to look at each of these factors as they apply to the present case. We discuss these factors seriatim.
1. A company’s expressed hostility towards unionization toyether with knowledye of the employees’ union activities.
The existence of anti-union animus and knowledge of the imminent arrival of the union letter is supported by substantial evidence. This factor is thus established and may be used to determine whether causation may reasonably be inferred.
2. Inconsistencies between the proffered reason for the discharye and other actions of the employer. 12
The Board noted several purported inconsistencies. Vemco gave overstaffing as one of three reasons for the layoff. Although conceding that Vemco was overstaffed by January 1989, the Board found Vemco’s failure to lay off at that time to be inconsistent with a claim of overstaffing in March. Vemco’s decision to develop and maintain a trained second shift in anticipation of an order from Ford, however, resolves this apparent inconsistency. Additionally, the Board found inconsistent Vem-co’s hiring of twelve new employees in January in the face of this overstaffing. The Board itself notes, however, that'these newly hired employees did not increase the size of Vemco’s work force, but merely replaced other employees who had terminated their employment. Vemco plausibly explains this hiring by pointing to its expec*1480tation of new business and its need to train those necessary for the work.
Vemco submitted as its second reason for the layoff its determination in early March that the anticipated Ford order would not be forthcoming. The AU found this inconsistent with Vemco’s continued contact with Ford in the form of a bid or marketing tool sent out by its sales representative two weeks after the layoff. This document or bid, however, was for other business and directed to a different division of Ford, i.e., Tempo and not Thunderbird. Vemco has never maintained that the anticipated Ford order was from the Tempo division. Vemco specifically stated that the work it was anticipating and finally concluded would not be forthcoming was work from Thunderbird. The AU’s mixing of the two sets of negotiations explains this apparent inconsistency.
In summary, substantial evidence does not support the existence of inconsistencies between Vemco’s reasons for the layoff and its other actions, and therefore this factor may not be used to determine whether causation is properly inferred.
3. Disparate treatment of certain employees compared to other employees with similar work records or offenses.
There is an underlying premise to this factor that is not clearly enunciated in the above formulation. This factor goes directly to discriminatory selection of employees for an adverse employment action. Correctly, the factor should be stated as disparate treatment of employees known to be pro-union compared to other employees with similar work records or offenses whose union sentiments are unknown or who are known to be anti-union. Generally speaking, where the GC relies on the mass layoff theory, rather than the traditional theory of section 8(a)(3) violations requiring knowledge of each individual laid-off employee’s union involvement, this factor loses some of its meaningfulness.
Additionally, any suggestion of disparate treatment is generally rebutted by the AU’s own findings. The AU stated, “Although [Vemco] asserts that it established an objective criteria [sic] for selection of persons for termination, it appears that attendance was made a singularly significant criteria [sic].” Vemco, 1991 WL 181869, at *19. Curiously, the AU’s statement suggests that attendance is not an objective criterion. An attendance report, however, generated by a computer program from records routinely kept by a business is as objective a basis for choosing employees to be laid off as a company can devise. There is no suggestion in the record that this criterion'was discriminatorily applied. Where an objective, criterion is deemed singularly significant in the selection process, substantial evidence for the existence of disparate treatment is generally lacking.
The GC did show specific employer knowledge of pro-union sentiment as to two of the sixty laid-off employees. There is ■ no evidence in the record, however, showing these two employees, in being laid off, were treated differently from other employees with similar attendance and discipline records whose union sentiments were unknown to Vemco or were known’to be anti-union. The existence of disparate treatment of these two employees is therefore not supported by substantial evidence and may not be used to support an inference of causation.
The Board also found an absence of evidence showing that the selections were based on a review of each employee’s relative qualifications and performance, noting that Vemco evaluated only the employees it laid off. There is undisputed testimony, however, that all employees were evaluated according to the five points set forth on the pre-printed termination report used by Vemco, but that the termination reports themselves, which reduced the evaluations to written form, were filled out only for those who were being terminated. The failure to fill out a termination report for employees not being terminated does not show that the retained employees were not evaluated.
Additionally, the AU’s finding that attendance was made a singularly important *1481criterion implies that all employees were evaluated relative to each other using this objective factor. Whether Vemco’s choice of primary criterion for its layoff is the wisest choice is not the question; it is certainly not inconceivable, however, that attendance might be a significant or even an overriding consideration in a plant that is organized around production lines and where timely deliveries are critical. As long as there is no suggestion that this objective criterion was used in a discriminatory manner, the absence of documentary evidence of relative evaluations based on other criteria does not support an inference of causation.
4. A company’s deviation from past practice in implementing the discharge.
Vemco, as a new company, had no established practice for plant-wide layoff implementation. This factor therefore does not come into play in this case.13
5. Proximity in time between the employees’ union activities and their discharge.
The layoff occurred and the union letter announcing its organizing campaign to Vemco arrived on the same day. This factor is therefore present and may be used to determine whether causation may reasonably be inferred.
The NLRB also considered “the haste with which the layoff was implemented.” Vemco, 1991 WL 181869, at *2. From January 1989 forward, Vemco’s upper management repeatedly discussed the need for a major reduction in work force absent an increase in business. March was targeted for this reduction, absent an increase in business, as early as January. On March 8, the record shows that Schütz, Torakis, and Winget recognized between themselves that there would be no increase in business in the near future; Schütz and Torakis testified they then settled on tbe specific date of March 17 for the layoff. Based on the AU's finding that Vemco did not schedule the layoff until it knew of the letter, however, Vemco allotted five days for implementation.
Most of this time was .used to determine the number of employees to be laid off. This determination required careful deliberation because of the danger that a miscalculation could ultimately have caused Vem-co to pay liquidated damages of $25,000 per minute of BOC downtime if Vemco failed to timely deliver. Vemco spent the remainder of the time selecting the employees to be laid off and preparing the termination notices, activities that went on simultaneously until the names and the letters were merged. These preparations took place against a backdrop of approximately $12,000 daily in unnecessary expense to the company for each day that passed before the layoff. With the cost so high of delaying the layoff any more than necessary, and an absence of evidence showing how long it would “normally” take a company of Vemco’s size and resources to implement a similar layoff, the record does not contain substantial evidence that the layoff was implemented in haste.
In summary, we conclude first that the inference of causation found by the Board is supported by substantial evidence. After giving deference to the Board, we further conclude that the GC’s prima facie case of discriminatory layoff is supported by substantial record evidence. This determination is close, however, and we therefore do not agree with the Board that the *1482GC’s prima facie showing is so strong that it in any way elevates Vemco’s burden of showing a legitimate business justification. Cf Vemco, 1991 WL 181869, at *2. We now turn to Vemco's defense that business conditions justified the layoff.
IV.
Once the GC has made out a pri-ma facie case of discriminatory employment action, the Wright Line methodology allows the employer to “show by a preponderance of the evidence that the employees would have been laid-off even if they had not engaged in protected activity.” Birch Run Welding & Fabricating, Inc. v. NLRB, 761 F.2d 1175, 1179 (6th Cir.1985). The employer’s showing referred to in Birch Run is regarded as an affirmative defense. An employer generally will attempt to meet its burden by showing a legitimate business justification for the discharge. See, e.g., Turnbull Cone Baking Co. v. NLRB, 778 F.2d 292, 296-97 (6th Cir.1985), cert. denied, 476 U.S. 1159, 106 S.Ct. 2277, 90 L.Ed.2d 720 (1986).. If the proffered business justification is deemed pretextual, the employer’s affirmative defense fails. NLRB v. Transportation Management Corp., 462 U.S. 393, 398-400, 103 S.Ct. 2469, 2472-73, 76 L.Ed.2d 667 (1983); see Turnbull, 778 F.2d at 296.
Vemco offered three business justifications for its decision to lay off permanently sixty employees on March 17, 1989: (1) overstaffing as of late January 1989; (2) failure to obtain an anticipated contract from Ford; and (3) the high cost (approximately $60,000 per week in direct and indirect expenses) of keeping 60 unnecessary employees on the payroll, absent the prospect of additional work.14 Although these three reasons are interrelated, Vemco’s most critical business reason for the layoff, which would operate to justify both the decision to implement the layoff and its timing, is the contention that, by failing to obtain the Ford contract on which it had bid, it had no use for a number of employees it had trained and retained in anticipation of such work.
Consideration of Vemco’s affirmative defense has been clouded throughout these proceedings because of the NLRB’s disinclination to distinguish carefully between two different and independent dealings between Vemco and Ford. In fall of 1988, Vemco began negotiating with two separate Ford divisions, Tempo/Topaz (Tempo) and Thunderbird, to paint fascias. Although the Board and the GC mix events from the Tempo and Thunderbird negotiations in their discussion of Vemco’s affirmative defense, Vemco did not base its affirmative defense on the outcome of its negotiations with Tempo because it did not consider the Tempo work to be a serious possibility at the time the layoff decision was made. In a broad sense, however, Vemco’s affirmative defense involves the general issue of a lack of work. We will therefore analyze and dispose of the Tempo negotiations before examining Vemco’s affirmative defense based on the Thunderbird negotiations.
A. TEMPO
In early November 1988, Tempo representatives, including Charles Baker, a parts buyer for Ford, met with Vemco personnel at the Vemco facility. Tempo reviewed Vemco’s facility to determine if Vemco was capable of painting and assembling certain fascias produced in the Tempo plant for the model year in progress. Tempo sought an outside source for these services because it projected that by spring of 1989 its production demands for the 1989 model would exceed its own in-house capacity to paint the fascias in question.
At the meeting, Tempo and Vemco discussed Tempo’s delivery requirements, Vemco’s manpower requirements, and Vemco’s current inability to run the type of paint necessary for the Tempo parts. Tempo was looking to an April 1 product availability date, which would have required *1483Vemeo to start preparing the work by-March 1. Vemeo stated that it would need to establish a second shift if it obtained the contract and that a lead time of 12 to 24 weeks above and beyond the 30-day production start-up period would be required to acquire and install new equipment to handle the type of paint application Tempo required.
After this meeting, Tempo shipped test-run parts to Vemeo later in November.. These were used only for estimating purposes because Vemeo did not then have the-equipment necessary to run the parts. Vemco’s sales representative submitted an estimate to Baker by the end of that same month. Tempo reviewed this estimate in early December. In the end, Tempo never awarded this contract because a sales downturn reduced demand for the car models in question. Thus, Tempo’s projected production shortfall never actually materialized.
By the time of the layoff, Vemeo was justified in its determination that the Tempo work was not a serious possibility. Counting backward from Tempo’s April 1 product availability date, and taking the shortest lead time Vemeo told Tempo it needed to obtain and install the necessary new equipment, Vemeo would have needed an order from Tempo by December 1, 1988, to assure on-time delivery. (If the longer equipment lead time is factored in, the project was out of the question before the meeting even took place.)
Tempo was still reviewing Vemco’s bid in early December, so this latest date for placing the order to assure timely delivery came and went. There is no evidence in the record that Tempo ever communicated with Vemeo concerning this bid before Vemco’s layoff decision, other than to inform Vemeo that its pricing was out of line. Tempo never informed Vemeo of any revised delivery date for the parts involved. At any time past mid-December, therefore, Vemeo would have been justified in deciding that the Tempo work was not a serious possibility under the terms discussed during the November meeting, especially if it were not willing or able to adjust its pricing to meet Tempo’s cost objections. By March 1, the date Vemeo would have had to Ijave started production to meet the only delivery date it was aware of, the absence of the Tempo work would have been a legitimate reason for Schütz, Torakis, and Winget to realize that Vemeo had not obtained the anticipated order.
Baker testified that in late March 1989 (after the layoff), he received a document from Vemeo supplying more comprehensive and detailed pricing of the operations required. The GC refers to this document as a bid and implies that this bid shows Vemeo had not lost hope for the Tempo work it bid on in 1988. From this implication, coupled.with Baker’s testimony that Tempo did not make a decision against giving the work to Vemeo, he infers Vem-co’s determination that it would not receive the work was pretextual. Vemeo, on the other hand, calls this document a marketing tool and states it was intended to show Tempo what kinds of information Vemeo would make, available to Tempo for,review.
The document itself is labelled a “Cost Analysis”; it does not refer to a Ford part number, a number Baker testified would normally appear on a price quotation related to a Ford manufactured part. It is also labelled “Original Quote/Price Request— Supplier Price Breakdown.” If indeed this was a bid, there is no evidence that it was in any way updated or revised to meet Tempo’s objection that Vemco’s pricing was out of line. Further, the GC’s unartic-ulated premise that Vemeo could not pursue contact with Tempo as a potential customer without proving by such contact that the particular work quoted four months earlier was still a live project ignores the realities of the marketplace. A vendor’s failure to obtain the first order bid on does not necessarily terminate the vendor’s courtship of the buyer. Just as a rejected suitor may still have high hopes of capturing the bride by altering his appearance, personality, or whatever it takes, so it may be with a determined supplier.
Baker testified that he was still looking for a supplier for the work in question in March 1989. He also testified that he re*1484ceived a telephone call the first week of April from Vemco’s sales representative inquiring “what was going on.”' Baker responded that he had passed everything on to the appropriate parties at Tempo (in December) and had heard nothing since then.
Baker’s statement that he was still actively looking for a supplier in March is inconsistent with the total lack of interest he displayed between December and April in following up Vemco’s bid; his lack of interest is more consistent with an awareness that the bid had been rejected by the “appropriate parties” and a realization there was no possibility of reaching an agreement on whatever had caused the rejection. Further, that Baker was still looking for a supplier in March when Vemco implemented its layoff does not mean that Vemco, after four months had passed without any word from Tempo except a rejection of its price, was even aware of his continued search. Even assuming such awareness, Vemco had no sensible basis under these circumstances to believe that it still had serious prospects for the business. The Board’s determination and the GC’s insistence that Vemco’s failure to obtain Tempo work was pretextual simply does not follow from the evidence.
B. THUNDERBIRD
Early in October 1988, Thunderbird representatives, including Spencer Teller, the engineering manager of the paint operation at the Thunderbird plant, met with Vemco personnel at the Vemco facility to establish a plan to evaluate Vemco’s capability of painting Thunderbird fascias for the 1989 model year in progress. In October or November, Thunderbird sent Vemco test parts. The first test parts painted by Vem-co failed Thunderbird’s test for paint adhesion. Schütz wrote to Thunderbird regarding this failure in January of 1989. A second meeting was held with Vemco in early February to evaluate its progress on painting the Thunderbird fascias. At that time, Teller found that paint quality and color appearance were acceptable on parts produced during a subsequent test run, but that masking needed improvement. In mid-February, another meeting was held to re-evaluate' Vemco’s masking capabilities, and Teller determined at that time that Vemco was capable of producing parts meeting Thunderbird’s quality standards.
At this meeting, Teller and Schütz also discussed the possibility of having Vemco run the remaining Thunderbird parts on hand without an agreement on price, but with an understanding that once a price was finally agreed upon, it would retroactively apply to these parts. This idea was rejected later the same day; Teller then instructed Schütz to return the parts, which Vemco did. Thunderbird did not proceed to evaluate Vemco’s quantity- and first time through capabilities.
While Teller conducted Thunderbird’s quality evaluations, Vemco’s sales representative dealt with Richard Berggren, an industrial engineering manager for Ford, on pricing matters. An estimate was submitted to Berggren in November after test parts were first received. Thunderbird considered Vemco’s prices to be out of line. In mid-January, Berggren requested an update on the estimate. In late April and in June, Teller also requested a requote. No update was ever submitted, and Vemco never did any Thunderbird work beyond the trial parts it had run. At the hearing, Ford testified that it never told Vemco that Vemco had lost the order.
There is general agreement that during the summer and fall of 1988, Vemco was forced to hire many employees above its originally forecast labor needs to compensate for unexpected technical difficulties at plant start-up. Some of these employees were hired for a second shift in December to relieve Vemco’s existing work force from extremely long workweeks and to train for the possibility of receiving work from Ford, which would have required a second shift. It is also agreed that by January 1989 Vemco experienced productivity gains resulting from resolution of some of its technical difficulties. This resulted in a shortage of work for its employees; they were cut to a four-day workweek in mid-January. Moreover, because of the *1485overlap of first and second shifts, they often had nothing to do for several hours a day even with a short workweek. There were employee rumors in March that Vem-co was going to implement either a three-day workweek or a layoff.
Corporate officers testified that they maintained this level of staffing until it finally became clear to them in early March that the anticipated Thunderbird work would not be forthcoming. At that point, Vemco immediately moved to reduce the work force pursuant to groundwork laid for such a reduction between December 1988 and early March 1989. This groundwork is evidenced by, inter alia, a document submitted to BOC by Vemco at the end of January promising a reduction in its work force and in price resulting from productivity gains.
The GC contends Vemco testified that Ford informed Vemco it would not be receiving the order in question, whereas Ford testified it never told Vemco it was out of the running for the order. The ALJ resolved this apparent conflict in testimony by specifically crediting Ford’s testimony where it seemed to conflict with Vemco’s. The AU stated:
I conclude that [Vemco’s] assertion that it learned that it would not receive the Ford business late in February or early March, because of a decision by Ford, is false. Accordingly, its assertion that the loss of the Ford business precipitated the economic necessity of the layoff is also false.
Vemco, 1991 WL 181869, at *18.
A review of the record as a whole, however, suggests that the conflict in testimony is more apparent than real. The testimony was readily reconcilable and the AU’s credibility determination unnecessary. The GC states in his brief that “Company official Schütz testified that he received a phone call in late February from Ford informing him that the Company [Vemco] had not been chosen, and that the Company would receive official notice thereafter.” The GC cites to several pages in the record of Schutz’s testimony to support this statement. The record on these pages reveals only the following:
On direct examination by counsel for Vemco—
Q When did you personally become aware that Vemco would not be receiving the Ford business?
A Around the end of February.
Q And you said you learned this, that you would not be receiving it, about the end of February?
A Yes.
On cross examination by the NLRB— Q You stated it was around the end of February, the beginning of March when you found out that you were not going to get this contract, is that true?
A That’s true. ■
Q You testified that you were told by Ford Motor Company that you weren’t going to get the contract?
A Yes.
Q Were you told that with regard to this particular bid?
A In regard to ultimately, all of them.
Q All of them. Isn’t it true, Mr. Schütz, that what you were actually told is that your bids were out of line? A Yes.
Q And isn’t it also true, Mr. Schütz, that you were told that from the beginning of March, end of February, 1989?
A The beginning of March, the end of February, yeah, that’s when we found out that we wasn’t going to have anymore dealing.
(Emphasis added).
The record divulges no specific reference in Schutz’s testimony to a telephone call or any subsequent official notice. What it does show is that the NLRB pressed Schütz to clarify his testimony regarding what Ford actually told Vemco until it became clear that Vemco considered Ford’s rejection of its pricing as the equivalent of official word that it would not be getting the order. Thus, Vemco’s testimony is reconcilable with Ford’s..
Taken in context with other record evidence regarding (1) the March decision made, not by Ford, but by Winget, Torakis, *1486and Schütz, that Vemco was not going to obtain the Thunderbird work and must therefore immediately move to remedy its overstaffing problem; (2) the mid-February telephone call from Teller instructing Vem-co to return all parts to Thunderbird; (3) the apparent inability to resolve the pricing problem; and (4) Thunderbird’s failure to complete its evaluations, the testimony on which the GC relies does not constitute substantial evidence that Vemco’s testimony conflicted with Ford’s. Because the record does not support the purported conflict in testimony on which the AU relied, we conclude accordingly that Vemco's business justification for the layoff cannot be found to be pretextual on the basis of this nonexistent conflict in testimony.
The remaining question is whether substantial record evidence supports a finding of pretext on some other basis. The AU acknowledged Vemco was overstaffed at some point in January, but seemed to reject overstaffing as a legitimate basis for the March layoff simply because the layoff did not take place in January. This leads to the implication that the AU rejected as pretextual the possibility of obtaining work from Thunderbird, because otherwise this possibility, with its attendant need for a trained second shift, would explain the delay in the layoff and, at the same time, preserve the reality of overstaffing as a legitimate business justification for a layoff. The AU, however, implicitly acknowledged that the possibility of obtaining the Thunderbird work was real when he found pretextual Vemco’s early March determination that the possibility of getting the' Thunderbird work ho longer existed. Finally, the AU never addressed Vemco’s claim that the high cost of retaining employees who were unnecessary to meet current production requirements was a reason for the layoff.
. Vemco basically contends'that when Win-get, S'chutz, and Torakis concluded in early March that Vemco would not obtain the Thunderbird work, this triggered their subsequent decision that Vemco’s overstaff-ing, with its associated drain on production and finances, mandated a reduction in work force. Tt appears that once the AU rejected as pretextual Vemco’s asserted failure to obtain the Thunderbird work, he viewed the overstaffing and associated costs as chronic, rather than critical, problems that could not legitimately support the layoff.
We focus our remaining inquiry into Vemco’s affirmative defense of lack of work based on its failure to obtain the Thunderbird orders. If this failure is not a pretext for Vemco’s early March decision to implement the layoff, it legitimizes not only the decision to lay off employees, but the timing of the layoff as well. At that point there would be no need to analyze Vemco’s other business reasons independently.
Because we believe Vemco’s determination that it had lost the order is plausible given the state of negotiations in early March, we conclude the Board’s finding of pretext is generally not supported by substantial evidence. When Ford said Vem-co’s pricing was out of line, Vemco’s management reasonably understood that it could not secure the order at its quoted price. Vemco knew it was not the only bidder on the job; because of this, Vemco would presumably have submitted what it viewed as an attractive price to diminish the risk of not being taken seriously as a contender for Ford’s business, or worse, of summárily losing the order to a lower bidder. In other words, Vemco would have been justified in rationalizing from its perspective that if Ford indicated Vemco’s price was not simply somewhat high, but was “out of line,” and if the quoted price did not leave Vemco much room for adjustment, then terms could not be reached on price and Vemco could not obtain the work.
The AU credited Teller’s testimony that since Vemco was capable of producing a quality part by mid-February, “it was simply a matter of working out the economics:” Vemco, 1991 WL 181869, at *17. He assumed that the ability to work out the economics was a given and disregarded the evidence that Ford and Vemco could not come to an agreement on price as a reason for Vemco’s decision not to pursue the elusive work order. The AU found instead that Vemco made its decision not to pursue *1487this work for “otherwise unexplained reasons.” Id. at *18. Finally, he found that Vemco walked away from 10 to 18 million dollars in business, without regard to its invested costs of setting up and running numerous test runs and maintaining an overstaffed plant trained and ready to work, not because it could not reach an agreement on price with Ford, but because of the Union’s organizational activity.
It makes no sense whatsoever that a sophisticated business management team would, in the aftermath of a financially difficult plant start-up, turn its back on an enormous amount of business from a major automobile manufacturer because of union organizing activity, especially after Vemco had made a substantial investment of time and money quoting, preparing for, and demonstrating it could do the work. Accordingly, we conclude that the NLRB’s finding of pretext in Vemco’s early March decision that the Thunderbird work would not be forthcoming, in conjunction with its overstaffing and the attendant costs, is generally not supported by substantial record evidence. Vemco met its burden to establish its affirmative defense of legitimate business justification; the mass layoff was therefore not in violation of section 8(a)(3) of the Act.
V.
A. THE BARGAINING ORDER
The NLRB has broad discretion to devise remedies to undo the effects of NLRA violations. Detroit Edison Co. v. NLRB, 440 U.S. 301, 316, 99 S.Ct. 1123, 1131, 59 L.Ed.2d 333 (1979); Shaw College at Detroit, Inc. v. NLRB, 623 F.2d 488, 489 (6th Cir.1980). Nonetheless, “[t]he issuance of a bargaining order is a severe remedy which is properly utilized in only two situations.” NLRB v. Arrow Molded Plastics, Inc., 653 F.2d 280, 284 (6th Cir.1981). In NLRB v. Gissel Packing Co., 395 U.S. 575, 89 S.Ct. 1918, 23 L.Ed.2d 547 (1969), the Supreme Court discussed the bargaining order as a remedy in cases involving violations of the Act. The Court set forth three categories into which such cases fall and determined whether a bargaining order is appropriate for each category. 395 U.S. at 613-15, 89 S.Ct. at 1939-41.
First, there are “exceptional” eases marked by “outrageous” and “pervasive” unfair labor practices in which the NLRB may impose a bargaining order without inquiring into whether the union has acquired majority status on the basis of authorization cards or by any other means. Id. at 613, 89 S.Ct. at 1939. The unfair labor practices in question must be of “such a nature that their coercive effects cannot be eliminated by the application of traditional remedies, with the result that a fair and reliable election cannot be had.” Id. at 614, 89 S.Ct. at 1940.
Second, the Court approved “the Board’s use of a bargaining order in less extraordinary cases marked by less pervasive practices which nonetheless still have the tendency to undermine majority strength and impede the election processes.” Id. Where the union’s majority status can be shown to have existed at one time, e.g., by employee sentiment expressed through authorization cards, a bargaining order should issue “[i]f the Board finds that the possibility of erasing the effects of past practices and of ensuring a fair election (or a fair rerun) by the use of traditional remedies, though present, is slight.” Id. Thus, the Board clearly has the authority to order an employer to bargain with a union even if the union has lost a representation election. NLRB v. C.J.R. Transfer, Inc., 936 F.2d 279, 282 (6th Cir.1991):
Third, the Court stressed that there exists a final category of “minor or less extensive unfair labor practices, which, because of their minimal impact on the election machinery, will not sustain a bargaining order.” Gissel, 395 U.S. at 615, 89 S.Ct. at 1940.
The ALJ, relying in large part on the mass layoff as an unfair labor practice, imposed a category two bargaining order on Vemco, Vemco, Inc., 304 NLRB No. 118, 1991 WL 181869, at *35 (NLRB Aug. 27, 1991), which Vemco challenges. As set *1488forth above, to justify a category two order, the GC must first show a one-time union majority. As of April 29, 1989, when the AU found the count of total employees in the bargaining unit to stand at 292, the record offers the following evidence: (1) 131 union authorization cards signed by current employees; (2) 56 union authorization cards signed by employees who had been laid off on March 17.15
Authorization cards signed by employees who were legally terminated do not count in the determination of a card majority. See Florsheim Shoe Store Co. v. NLRB, 565 F.2d 1240, 1245 (2d Cir.1977) (concluding employer’s contention that cards of legally dismissed employees may not be counted toward a union majority fails because employees’ dismissals in fact were illegal). Therefore, the 56 cards of Vemco employees permanently laid off on March 17 cannot be counted to determine majority status.
In summary, the ALJ determined that on the date the Union purportedly attained majority status, it had authorization cards from 187 of 292 unit employees. However, because the cards of the 56 laid-off employees may not be counted, the Union actually had only 131 authorization cards of 292 unit employees and therefore did not achieve majority status. The ALJ’s finding that the Union had majority status as of this date is obviously not supported by substantial evidence now that the cards signed by the legally laid-off employees cannot be counted. Moreover, the Board may not rely on the layoff of March 17 as a basis for the bargaining order. As a direct result, the Board abused its remedial discretion in issuing a category two bargaining order.
B. A RERUN ELECTION
The final issue in this case is whether Vemco’s unfair labor practices support the traditional remedy of a rerun election. In addition to the undisputed section 8(a)(1) violations discussed in the context of the layoff, the Board found Vemco to have committed a variety of other section 8(a)(1) violations between the layoff and the election. As with the undisputed pre-layoff violations, Vemco acknowledges these violations except insofar as they encompass alleged protected speech under section 8(c). We therefore examine the violations that may rely on protected speech before reviewing the evidence as a whole to determine whether a bargaining order is appropriate.
The violations that Vemco claims rely on protected speech related to predictions made to employees and the media, both verbal and written, that if the company were required to reinstate employees laid off on March 17, it might have to cut workhours or suffer another layoff because of a lack of work.16 In discussing employer protected speech, the Court set *1489forth its views on employer conduct in the course of an organizing campaign:
[A]n employer is free to communicate to his employees any of his general views about unionism or any of his specific views about a particular union, so long as the communications do not contain a “threat of reprisal or force or promise of benefit.” He may even make a prediction as to the precise effects he believes unionization will have on his company. In such a case, however, the prediction must be carefully phrased on the basis of objective fact to convey an employer’s belief as to demonstrably probable consequences beyond his control.... If there is any implication that an employer may or may not take action solely on his own initiative for reasons unrelated to economic necessities and known only to him, the statement is no longer a reasonable prediction based on available facts but a threat of retaliation based on misrepresentation and coercion, and as such is without the protection of the First Amendment [as implemented by section 8(e)].
Gissel, 395 U.S. at 617-19, 89 S.Ct. at 1941-43 (citations omitted).
First, we note that it is not at all clear whether the Vemco prediction at issue is a prediction of the consequences of company unionization or, instead, a prediction of the consequences of an impending NLRB decision that might mandate reinstatement of the laid-off employees, regardless of the outcome of the organizing campaign.17 If the latter, it is not a violation of the Act; if the former, we believe it satisfies the Gissel prediction standard for the reasons discussed below. Thus, it is protected speech which must be excluded in a review of the record to determine the propriety of a rerun election. .
Vemco could have objectively and reasonably assumed, even absent an NLRB order, that the UAW would seek reinstatement of the terminated employees. The Union had obtained authorization cards from fifty-six of sixty terminated-employees, with all but one card signed after the layoff.. The terminated employees were allowed to vote in the representation election, although their ballots were challenged. It is evident that the UAW continued to consider these people as members of the bargaining unit. The obvious resolution of the anomaly of considering terminated employees as members of the bargaining unit is to seek to have those employees reinstated.
Vemco claimed in its prediction that reinstatement would result in a shortage of work, necessitating a cutback in workhours or another layoff. The record shows that Vemco employed temporary help in the fall of 1989 and expected to open a plant expansion in January 1990 to handle new work obtained for a 1990V2 year model. The record also contains evidence that the temporary employees were used to cover workhours related to temporary start-up difficulties on lines running parts for the new model year and that the plant expansion would not result in a net increase in employees because additional productivity gains on the existing lines were projected to allow existing employees to be transferred to the new lines.
There was .nothing, to suggest that the predicted- shortage of work was within Vemco’s control, and there was no implication that Vemco would implement a cutback in hours or a layoff solely on its own initiative for reasons unrelated to the economic necessity of adjusting to a shortage of work.- Given the severe consequences of a failure to deliver according to JIT, it hardly seems likely Vemco would cut back if it had the work. Vemco therefore made a reasonable prediction based, on available facts, and not a threat of retaliation based on misrepresentation and coercion. This prediction is protected employer speech under section 8(c) and will not be used to determined whether a rerun election should be ordered.
The remaining violations are summed up as follows:
*14901. Threat to close plant for non-economic reasons in the face of an organizing campaign and to later reopening ■ with new employees; ■
2. Threat of harsher personnel decisions;
3. Promise of benefits relating to the institution of recall rights and advance notice of weekend overtime;
4. Confiscation of union literature;
5. Solicitation of grievances; and
6. Overly broad no-solicitation rule and threats of' discriminatory discharge involving employee Hall.
These undisputed violations are sufficient to support the traditional remedy of a rerun election. They do not, however, warrant a bargaining order under either category set forth in Gissel.
VI. CONCLUSION
The Board’s finding that Vemco’s March 17 layoff constitutes a violation of section 8(a)(1) and (3) of the Act is not supported by substantial evidence on the record taken as a whole because Vemco’s business justifications for the layoff are not pretextual. Enforcement of provisions 2(a) and (b) of the NLRB’s order relating to the laid-off employees is therefore denied. Additionally, the Board abused its discretion in issuing a Gissel category two bargaining order because the existence of the necessary union majority status is not supported by substantial record evidence. Enforcement of the bargaining order in provision 2(d) of the NLRB’s order is therefore denied.
Assuming the UAW had sufficient card support to file a petition for election in July 1989, the NLRB’s order is modified to include the traditional remedy of a rerun election. The petition to enforce the remaining provisions of the NLRB’s order, as modified, is granted. If the challenged ballots cast by 52 of the laid-off employees have been counted and have led to certification of the Union as bargaining representative, this certification is to be set aside.
The case is REMANDED to the NLRB for action consistent with this opinion.
. A bargaining order is an extraordinary remedy utilized by the NLRB when an employer has engaged either in outrageous and pervasive conduct or in less pervasive practices which have the tendency to undermine the majority strength of the alleged bargaining unit, which so impede the election process that the Board finds itself compelled to direct the parties to bargain without an election determination of the majority status of the union, or at times even when a union has lost the election. See NLRB v. Gissel Packing Co., 395 U.S. 575, 613-15, 89 S.Ct. 1918, 1939-41, 23 L.Ed.2d 547 (1969).
. We granted the petition of the UAW for leave to intervene.
. "Fascia" is an industry term for bumper. "Cladding” is the ground effects package that is used on the lower 13 inches of an automobile to create a decorator look in conjunction with the fascias. Masking takes place before parts are painted to insure paint application only in the appropriate areas.
.JIT is a system under which suppliers are given short turn-around times from release to delivery, forcing the supplier, rather than the buyer, to maintain the inventory necessary to satisfy the buyer's assembly requirements.
. NLRA section 8(c), 29 U.S.C. § 158(c), provides as follows:
The expressing of any views, arguments, or opinion, or the dissemination thereof, whether in written, printed, graphic, or visual form, shall not constitute or be evidence of an unfair labor practice under any of the provisions of this subchapter, if such expression contains no threat of reprisal or force or promise of benefit.
. These positions are carefully analyzed in Rebecca H. White, The Statutory and Constitutional Limits of Using Protected Speech and Evidence of Unlawful Motive Under the National Labor Relations Act, 53:1 Ohio St.L.J. 1 (1992). In an unpublished opinion, Active Transportation Co. v. NLRB, 1991 WL 9979, 1991 U.S.App. LEXIS 1603 (6th Cir. Jan. 31, 1991) (per cu-riam), this court ordered enforcement of the Board’s decision in Active Transportation Co., 296 NLRB No. 58, 1989 WL 224295 (NLRB Aug. 31, 1989), where the Board relied in part on its own interpretation of section 8(c) in Holo-Krome I for the proposition that "statements not independently violative of the Act may serve as a basis for finding animus,” at *4, and we thereby rejected the Second Circuit’s approach in Holo-Krome II.
. In this case, it is difficult to ascertain if the panel of the Board that reviewed the- ALJ’s decision was aware' of the Board’s policy in this area. To illustrate, the ALJ erred in using Vem-co’s protected expression of its non-union philosophy as it appeared in its personnel handbook as substantial evidence of several violations. To rectify this error without upsetting the ALJ’s findings, the panel simply disavowed any use of Vemco’s non-union philosophy in .affirming the ALJ’s findings. The panel did not directly discuss the evidentiary value of this protected speech.
. Section 7 of the NLRA, 29 U.S.C. § 157, provides:
Employees shall have the right to self-organization, to form, join, or assist labor organizations, to bargain collectively through representatives of their own choosing, and to engage in other concerted activities for the purpose of collective bargaining or other mutual aid or protection....
. Section 8 of the NLRA, 29 U.S.C. § 158(a)(1), (3), provides in pertinent part:
(a) Unfair labor practices by employer It shall be an unfair labor practice for an employer—
(1) to interfere with, restrain, or coerce employees in the exercise of the rights guaranteed in section 157 of this title;
(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. ...
.Henceforth, discussion of the layoff as a breach of section 8(a)(3) will encompass the automatic section 8(a)(1) violation.
. The failure to do so, however, would preclude a finding on the traditional claim that the employer selected the employees for layoff in a discriminatory fashion. Therefore, the ALJ’s rather vague finding that "some of the persons chosen for termination were selected because of their support for the Union,” Vemco, Inc., 304 NLRB No. 118, 1991 WL 181869 (NLRB Aug. 27, 1991), is supported by substántial record evidence only where the GC showed that Vemco had specific knowledge of an employee’s pro-union sentiments at the time the selection was made. Thus, the selection process appears to have been tainted only as to the two employees whose names were not submitted until after the UAW’s letter had arrived and who were identified in that letter as members of the organizing committee. However, a bona fide business justification for employment termination is an adequate defense to a prima facie case of discriminatory discharge, whether the GC has proceeded on the mass layoff theory or on the more traditional individual basis. See infra part IV.
. A real question arises here as to the appropriateness of this factor. Technically, the GC’s prima facie case should be evaluated before evidence is heard on Vemco’s affirmative defense. This means Vemco’s reasons for the layoff should not even be in evidence when the existence of the inference of causation is being considered.
. In its personnel handbook, however, Vemco alerts its employees to the absence of a recall policy: It specifically states that no promise or guarantee of recall is or can be made to anyone ■ laid off. In the face of this policy, the AU inferred that the layoffs were made permanent rather than temporary for discriminatory reasons. Vemco, 1991 WL 181869 at *20. There is no evidence that at the time of the layoff Vemco anticipated being able to recall the laid-off employees. Further, given the personnel policy, Vemco’s employees had no reason to expect a layoff to be anything but permanent. Substantial evidence does not support the ALJ’s determination that the permanent character of the layoff was discriminatory. See NLRB v. Hovey Elec., Inc., 964 F.2d 543, 548 (6th Cir.1992) ("Because the Company viewed the employees as permanently laid off, it committed no unfair labor practice by noting this permanent status on the layoff slips.”).
. Despite the layoff, Vemco was able to meet BOC's delivery requirements. This is an important indicator that Vemco was indeed overstaffed by at least 60 production employees before the layoff.
. Fifty-five of 56 cards signed by laid-off employees were signed after the March 17 layoff. The ALJ found a total of 187 of 292 employees had signed authorization cards as of April 29, 1989, including the employees he deemed illegally laid off. This is misleading: If the authorization cards of the laid-off employees are to be counted toward majority status, the size of the bargaining unit also must be increased by the 60 laid-off employees. Therefore, the Union either had authorization cards from 131 of 292 members of the bargaining unit (if cards signed by laid-off employees do not count) or it had authorization cards from 187 of 352 members of the bargaining unit (if the cards signed by laid-off employees count). Regardless of any error by the ALJ, if the cards signed by the laid-off employees were not to be counted, the Union never would have attained majority status; if they were to be counted, it would have attained majority status.
. Representative of these statements is the version found in a letter from Schütz to all employees dated July 27, 1989:
The lay-off last March was permanent. If the Union gets its way and Vemco is required to hire back 60 of those people, the answer seems to be one of two alternatives. Since we do not have enough work for an additional 60 people on a full-time basis either (1) we could try to keep everyone employed—working 3 or maybe 4 days per week, or (2) another lay-off would be made out of our present work force.
Similar statements were made in a question and answer sheet distributed to employees, a press release that was quoted in a local newspaper, and during a media conference to a television reporter.
. Most of the statements were made immediately after the first hearings in July 1989 and generally referred to the "trial,” the proceeding, or possible NLRB orders stemming from the hearings.