Medical Center of Beaver County, Inc. ("the Hospital") has petitioned for review of an order of the National Labor Relations Board ("the Board"). The Board ordered the Hospital to cease and desist from violating sections 8(a)(1) and 8(a)(5) of the National Labor Relations Act, 29 U.S.C. Sec. 158(a)(1), (a)(5) (1976), by refusing to bargain with District 1199P, National Union of Hospital and Health Care Employees, RWDSU, AFL-CIO ("the Union").1 The Board has filed a cross-application for enforcement of its order. We have jurisdiction pursuant to 29 U.S.C. Sec. 160(e), (f) (1976). Because we conclude that the Union was not properly certified as the exclusive bargaining representative of a designated unit of hospital employees, we will grant the Hospital's petition for review, and we will deny the Board's cross-application for enforcement of its order.

On March 10, 1981, the Union filed a petition with the Board's regional office seeking a representation election among the licensed practical nurses and certain technical employees of the Hospital. Pursuant to the parties' stipulation for certification upon consent election, the regional director conducted an election by secret ballot on May 15, 1981. Of the 285 unit employees who voted, 143 employees voted for the Union, and 133 employees voted against the Union. Thus the Union received a bare majority of the 285 votes cast. App. at 9, 633. Nine of the 285 ballots were challenged ballots. Because the number of challenged ballots was insufficient to affect the result of the election, however, the nine challenged ballots remained uncounted. See 29 C.F.R. Sec. 102.69 (1982).

On May 22, 1981, the Hospital filed timely objections to the election because of the conduct of members of the Union's organizing committee on election day. The Hospital alleged, inter alia, that the election was tainted because the Union had kept a voter eligibility list, in violation of long-established Board policy, in order to record the names of unit employees who voted. The regional director ordered a hearing on the Hospital's objections. After a four-day hearing the hearing officer recommended that, although the case was close, the Hospital's objections should be overruled.

On April 30, 1982, the Board adopted the hearing officer's recommendation. The Board found that the Union had in fact maintained an unauthorized voter list, but concluded that such list-keeping had only a de minimis impact on the election. App. at 660-61. The Board accordingly certified the Union as the exclusive bargaining representative of the designated unit of licensed practical nurses and technical employees. Member Hunter dissented, reasoning that direct evidence and "compelling circumstantial evidence," app. at 665, precluded a finding of de minimis impact. He therefore concluded that the election should be set aside. App. at 663.

Following its certification by the Board, the Union asked the Hospital to bargain with it. When the Hospital refused, the Union filed the section 8(a)(1) and 8(a)(5) charge at issue against the Hospital. The Hospital admitted that it had refused to bargain with the Union, but it asserted that it did so in order to challenge the Board's decision to overrule the Hospital's objections and to certify the Union.2 The Board found no grounds which would warrant a reexamination of its certification decision and accordingly granted summary judgment in favor of the Union. It ordered the Hospital to cease and desist from refusing to bargain with the Union. Member Hunter dissented for the same reasons expressed in his dissent from the Board's certification decision.

The Hospital filed this petition for review, and the Board filed a cross-application for enforcement of its order. The only issue which we address in this case is whether "the Board's determination regarding the impact of the [list-keeping] is supported by substantial evidence on the record considered as a whole." Jamesway Corp. v. NLRB, 676 F.2d 63, 69 (3d Cir.1982).3 To determine whether the Board's de minimis determination is supported by substantial evidence, we must consider the facts surrounding the election in closer detail.

The facts in this case are largely undisputed. The election was held in a voting room located inside the hospital complex. The actual voting took place during two sessions, the morning session which ran from 6:00 a.m. to 9:30 a.m. and the afternoon session which ran from 2:00 p.m. to 5:30 p.m. There are only two entrances through which employees could have gained access to the voting area. One is the doctors' entrance which is located approximately forty-five feet down a corridor from the voting room. A sidewalk connects the doctors' entrance to a parking lot, and that entrance was more convenient for those employees who were not scheduled to work during the polling period and who came to the hospital only to vote. The other entrance, the employees' entrance, is located on the floor above the voting room. It is generally used by employees who are scheduled to work because it is located near the time clock.

It is undisputed that members of the Union's organizing committee decided to station themselves on election day at those two entrances in order to provide "moral support" for the Union and in order "to instill confidence" in employees who were coming to vote. App. at 77, 636. During the morning voting session two organizing committee members, Shirley Olshesky and Madeline Berninger, stationed themselves at the employees' entrance. During the afternoon voting session, two organizing committee members and a unit employee stationed themselves at the employees' entrance, and four organizing committee members and the same employee stationed themselves at the doctors' entrance for varying periods of time. Also during the afternoon session, Union Vice-President Kay Tillow and Union Secretary-Treasurer Jack Hustwit drove up to the doctors' entrance on several occasions to pick up committee members and to check up on voter turnout. App. at 211-12, 639.

Sometime between 7:30 a.m. and 8:00 a.m., Hustwit admittedly furnished a clipboard containing a copy of the voter eligibility list to Olshesky and Berninger, who were stationed at the employees' entrance. App. at 98-99, 636. Hustwit wanted Olshesky and Berninger to check off names of unit employees as they entered in the morning so that he could contact those employees who did not vote in the morning to encourage them to vote in the afternoon. App. at 94-96, 99, 636-37. Olshesky and Berninger admitted that they checked off ten to twelve names on the list and that they returned the list to Hustwit after the close of the morning polls. App. at 358, 637.

Hustwit testified that he instructed Olshesky and Berninger to conceal the clipboard, app. at 99-100, and they testified that they did not mark the clipboard in the presence of other employees. App. at 328-29, 340, 357. Numerous employees, however, testified that they saw Olshesky and Berninger making marks on the clipboard. Earline Clark, a unit employee, testified that as she was approaching the employee's entrance she saw two women with a clipboard stopping two other unidentified women who were entering the hospital. App. at 487-88, 637-38. Clark testified that she saw them speak to the approaching individuals and then that she saw one of the women make marks on the clipboard. App. at 489, 637-38. Mary Fedeyski, another unit employee, testified that at around 11:00 a.m. she saw two women whom she did not recognize at the employees' entrance. App. at 459. They told Fedeyski that she was too late to vote at the morning session, and that that was the time when her vote really counted. After she convinced them that she was going to vote in the afternoon anyway, Fedeyski testified that one of the women made a mark on a clipboard and commented, "that makes seven." App. at 459-60.

Arlene Verdier, a nonunit employee, testified that she saw Olshesky and Berninger speak with individuals going through the employees' entrance and then make check marks or "X's" on the clipboard. App. at 165-68, 637. She testified that Berninger was trying to hide the clipboard under her trench coat, but that while Verdier was there, "it was very easy to see the clipboard." App. at 166-67. Verdier testified that she saw four individuals enter the employees' entrance at that time, but that she did not know whether they were unit employees. App. at 182-83, 190-91. Ruth Sutherin, another nonunit employee, testified that she saw Olshesky and another woman making marks on a clipboard at the employees' entrance. App. at 150-52, 637. She testified that she overheard them make comments such as, "did you get that one, has she voted, do you know if we've gotten her." App. at 151. Sutherin also testified that six or seven employees passed through the employees' entrance at that time, but that she did not know whether they were unit employees. App. at 152, 160, 637 n. 6. Frances Cerjak, another nonunit employee, testified that, as she passed through the employees' entrance, she saw Olshesky seated on a bench just inside the entrance. App. at 433. She testified that Olshesky had a clipboard that contained a list of names and check marks. Id. Cerjak testified that Olshesky was making no attempt to conceal the clipboard. App. at 433-34.

In addition to the list-keeping activity at the employees' entrance, there is also credited evidence of list-keeping at the doctors' entrance. Personnel Director Fred Delvaux testified that at approximately 4:15 p.m. he saw Mara Delfiacco, another member of the organizing committee, making marks on a clipboard as individuals passed through the doctors' entrance. App. at 213-14, 639 n. 9, 644. The hearing officer also found that Kay Tillow, who made numerous trips to the doctors' entrance by car to pick up union organizers and to check on voter turnout, had a voter eligibility list in her car. App. at 414, 448-50, 639 n. 9.

The Board has stated that "[i]n election proceedings, it is the Board's function to provide a laboratory in which an experiment may be conducted, under conditions as nearly ideal as possible, to determine the uninhibited desires of the employees." General Shoe Corp., 77 N.L.R.B. 124, 127 (1948). As one means of preserving those laboratory conditions, the Board has long maintained a policy prohibiting anyone from keeping a list, aside from the official eligibility list, of employees voting in a representation election. Masonic Homes of California, Inc., 258 N.L.R.B. 41, 41 n. 4, 47-48 (1981); Piggly-Wiggly # 011, 168 N.L.R.B. 792, 792-93 (1967); International Stamping Co., 97 N.L.R.B. 921, 922-23 (1951). In the interest of ensuring free, non-coerced elections, the Board has set aside elections "if employee voters know, or reasonably can infer, that their names are being recorded" on unauthorized lists. Masonic Homes, 258 N.L.R.B. at 48; see also Marathon Le Tourneau Co., 208 N.L.R.B. 213, 223-24 (1974); Belk's Department Store of Savannah, Ga., Inc., 98 N.L.R.B. 280 (1952). Absent such knowledge or inference on the part of voters, any list-keeping activity, although technically prohibited, obviously could not interfere with the exercise of voter free choice and would not warrant setting aside an election. The party objecting to an election can prove voter knowledge, however, both by direct evidence and by circumstantial evidence. Piggly-Wiggly, 168 N.L.R.B. at 792-93; A.D. Juillaird and Co., 110 N.L.R.B. 2197, 2199 (1954). Once voter knowledge has been proven, the Board has not required a showing that the list-keeping actually interfered with the free choice of any voter. International Stamping, 97 N.L.R.B. at 923.

In some cases the Board has refused to set aside elections where it has concluded that the list-keeping had only a de minimis impact on the election. See Robert's Tours, Inc., 244 N.L.R.B. 818, 824-25 (1979), enforced, 633 F.2d 223 (9th Cir.1980) (where only one voter had seen the list-keeping and where 15 of 26 eligible voters had voted for the union); Tom Brown Drilling Co., 172 N.L.R.B. 1267 (1968) (no evidence that any voters had seen the list-keeping); Juilliard, 110 N.L.R.B. 2197 (same); see also N.L.R.B. v. South Mississippi Electric Power Association, 616 F.2d 837, 839 (5th Cir.1980); N.L.R.B. v. Fenway Cambridge Motor Hotel, 601 F.2d 33, 38 (1st Cir.1979).

In this case the Board noted that the Hospital had offered direct evidence establishing that only one voter, Earline Clark, was actually aware of the list-keeping. It further concluded that the circumstantial evidence in this case was inadequate to support the inference that any other voters were similarly aware. It commented "that the union supporters attempted to hide their unauthorized voting lists and that these efforts were largely successful." App. at 661. The Board then concluded that in a unit of over 300 employees, only one voter knew of the Union's conduct. Thus the Board determined that the list-keeping had only a de minimis impact on the election.

Examining the facts in this case, we find that the Board's determination of de minimis impact is not supported by substantial evidence on the record. This case does not involve isolated, short-lived incidents of list-keeping. Rather it concerns public voter tallying by members of the Union's organizing committee at locations proximate to the polling area. The Board concluded that only one unit employee, Earline Clark, was aware of the list-keeping. There was testimony, however, that another unit employee, Mary Fedeyski, also was aware of the prohibited activity. The hearing officer failed to consider that testimony as it relates to the Hospital's list-keeping objection. Rather he considered Fedeyski's testimony only as it related to another of the Hospital's objections, not renewed here, that the two women with the clipboard attempted to convince Fedeyski, an eligible voter, not to vote in the afternoon session.4 We find Fedeyski's testimony obviously relevant to the list-keeping objection, and we find that Fedeyski's testimony renders unwarranted the Board's conclusion that only one unit employee was aware of the prohibited conduct.

In addition to the direct testimony of unit employees Clark and Fedeyski that they were aware of the list-keeping, we find abundant circumstantial evidence in the record to support the inference that many other unit employees must have been aware of the Union's list-keeping, particularly at the employees' entrance. It is undisputed that Olshesky and Berninger checked off names for well over an hour at the employees' entrance. Hustwit testified that he instructed them to keep the clipboard hidden. However, employees Earline Clark, Mary Fedeyski, Arlene Verdier, Ruth Sutherin and Frances Cerjak all testified that they saw the women making marks on the clipboard. Verdier testified that "it was very easy to see the clipboard," app. at 167, as she passed through the employees' entrance and no finding of the hearing officer or Board contradicts that. Verdier and Sutherin could not be sure whether the numerous individuals also passing through at that time were unit employees. However, at least ten to twelve unit employees had to have passed through the entrance while Olshesky and Berninger were stationed there because that is the number of unit employees whose names were checked off on their list. Thus, the testimony of at least five witnesses belies the Board's conclusion that the union adherents were successful in their efforts to conceal their improper lists.

Furthermore, the election in this case was a close one. The Union received a bare majority of the votes cast. If one voter who voted for the Union had instead voted against the Union, the Union would not have received the requisite majority without the resolution of the challenged ballots. We have often indicated that where an election is close, we will require closer scrutiny by the Board of election objections. Jamesway, 676 F.2d at 71; Monmouth Medical Center v. NLRB, 604 F.2d 820, 823 n. 4 (3d Cir.1979); Aircraft Radio Corp. v. NLRB, 519 F.2d 590, 594 (3d Cir.1975) (misrepresentation cases). The Board focused only on the size of the employee unit in making its de minimis determination. The factor of the closeness of the election, however, is important where the Union has admittedly engaged in prohibited conduct and where the only question is whether that conduct had more than a de minimis impact on the outcome of the election.

We therefore conclude that the Board's determination that the Union's list-keeping had only a de minimis impact on the election is not supported by substantial evidence on the record. The Board's long-standing policy against list-keeping represents an exercise of the Board's wide discretion to establish safeguards for conducting representation elections. See NLRB v. A.J. Tower Co., 329 U.S. 324, 330, 67 S. Ct. 324, 327, 91 L. Ed. 322 (1946); Monmouth Medical Center, 604 F.2d at 823. However, the Board's "[e]lection rules which are designed to guarantee free choice must be strictly enforced against material breaches in every case, or they may as well be abandoned." International Stamping, 97 N.L.R.B. at 923.

We accordingly will grant the Hospital's petition for review, order the representation election to be set aside, and deny the Board's cross-application for enforcement of its order.

A. LEON HIGGINBOTHAM, Jr., Circuit Judge, dissenting.

I respectfully dissent from the majority's conclusion that there was not "substantial evidence on the record" to support "the Board's determination that the Union's list-keeping had only a de minimis impact on the election", p. 1001.

This is indeed a classic case concerning the degree of judicial restraint a court must exercise when reviewing an agency's factual determination that we may have decided differently if we had had the obligation to consider the matter ab initio as the original factfinders. Thus, if my esteemed colleagues were writing as hearing officers or as members of the National Labor Relations Board, their findings might be appropriate and permissible; accordingly, a reviewing court could conclude that even the quite different findings now made by the majority are supported by substantial evidence. However, the vice of the majority opinion is its failure to appreciate that, on appeal, one record may contain substantial evidence to support two diametrically opposed conclusions--just as in a negligence case there can be substantial evidence which supports the diametrically opposed conclusions that the plaintiff was negligent or that the plaintiff was not negligent.

The hearing officer stressed that "only one voter ..., Earline Clark, whose testimony I credit, was shown to have known that the unauthorized list of voters was being maintained." App. at 644 (Emphasis added). He found "that the list-keeping [was] de minimis ... where it [was] shown that only one employee in a bargaining unit of over 300 employees knew of the unauthorized list's maintenance and use." Id. (Emphasis added) He ruled that the one proven sporadic incident of list-keeping did "not warrant setting aside the election." Id. On the basis of the record, I think that the hearing officer's findings were permissible and based on substantial evidence.

The view which the employer asserts before us now was argued vigorously before the Board by Board Member Hunter in his dissent and answered by the Board. The Board stated "[o]ur dissenting colleague would sustain this objection and set aside the election. Although he concedes that there is no direct evidence that any voter other than Clark saw the list, he would infer from the circumstances that knowledge of the list was more widespread. We find that such an inference is unwarranted." J.A. 660. (Emphasis added)

In light of our limited scope of review, the fundamental question before us is whether as a matter of law on the de minimis issue the facts of this case required the fact finders to construe this record only as the dissenting Board member did and as the majority of this Court now does. Or does the record permit the different inferences that accord with the findings of the hearing examiner and the majority of the Board?

Despite all that is written in the majority's opinion there is still no direct evidence that any voter other than Clark saw the list checking. Of course, there is evidence which could cause one to make broader inferences, but the point is that no one need necessarily make those broader inferences.

From my review of this record, the Board could have made either of two contradictory inferences from the circumstantial evidence as to the number of persons who saw the list checking. It could have adopted the view pressed by dissenting Board Member Hunter or the view taken by the majority Board members.

This summer, as my mind occasionally wanders beyond the confines of the courthouse, I think that the members of the NLRB sometimes are somewhat like a baseball umpire behind home plate making judgment calls on pitches that are on or near the edge of the strike zone. When a fast ball comes in over the plate, quite close to the knees, some persons might think it was a strike, others a ball, but whatever way the umpire calls it, there was probably substantial evidence to justify his judgment. Despite the seeming serenity in an appellate courtroom, the black robes and the historic formalism of the adjudicatory process, sometimes fact finding in the court or before administrative agencies is not much more of a precise science than umpiring in a ball park.1 In this case, with commendable candor, the hearing officer said he had to decide a factual question that "is a close one." App. at 644. When there is substantial evidence to support the Board's conclusion, I do not believe that simply because a case is close the Board must lose, even though others might have "called" it differently.