NLRB v. Beverly Health

UNPUBLISHED UNITED STATES COURT OF APPEALS FOR THE FOURTH CIRCUIT NATIONAL LABOR RELATIONS BOARD, Petitioner, v. BEVERLY HEALTH AND REHABILITATION No. 96-2195 SERVICES, INCORPORATED, d/b/a Morgan Manor Nursing and Rehabilitation Center, Respondent. On Application for Enforcement of an Order of the National Labor Relations Board. (6-CA-27750) Argued: May 9, 1997 Decided: August 12, 1997 Before HAMILTON and MOTZ, Circuit Judges, and CURRIE, United States District Judge for the District of South Carolina, sitting by designation. _________________________________________________________________ Enforcement denied by unpublished per curiam opinion. Judge Motz wrote a dissenting opinion. _________________________________________________________________ COUNSEL ARGUED: Daniel Josef Michalski, NATIONAL LABOR RELA- TIONS BOARD, Washington, D.C., for Petitioner. Martin J. Saunders, JACKSON, LEWIS, SCHNITZLER & KRUPMAN, Pitts- burgh, Pennsylvania, for Respondent. ON BRIEF: Frederick L. Fein- stein, General Counsel, Linda Sher, Associate General Counsel, Aileen A. Armstrong, Deputy Associate General Counsel, David Fleischer, Senior Attorney, NATIONAL LABOR RELATIONS BOARD, Washington, D.C., for Petitioner. Terri Imbarlina Patak, JACKSON, LEWIS, SCHNITZLER & KRUPMAN, Pittsburgh, Pennsylvania, for Respondent. _________________________________________________________________ Unpublished opinions are not binding precedent in this circuit. See Local Rule 36(c). _________________________________________________________________ OPINION PER CURIAM: The National Labor Relations Board (the Board) seeks enforcement of its bargaining order against Beverly Health and Rehabilitative Ser- vices, Inc. (Beverly). Beverly asserts that the Board improperly certi- fied election results in favor of representation by the Health Care and Social Service Union, SEIU, AFL-CIO (the Union), after modifying the bargaining unit post-election to exclude all licensed practical nurses (LPNs). Because the Board's election procedures denied the employees an opportunity to vote for representation in the unit certi- fied by the Board, we deny the Board's application for enforcement of its order. I. Beverly, d/b/a Morgan Manor Nursing and Rehabilitation Center, operates a 100-bed long-term nursing care facility in Morgantown, West Virginia. On June 16, 1992, the Union filed a petition for certifi- cation of representation with Region Six of the Board. The petition included a proposed bargaining unit consisting of all full-time and regular part-time service and maintenance employees at the nursing home, including LPNs and nine other job classifications. The petition 2 excluded, among others, all supervisors as defined in § 2(11) of the National Labor Relations Act (the Act).1 See 29 U.S.C. § 152(11). On July 13, 1992, a Board hearing officer conducted a representa- tion hearing, the primary issue of which was whether the LPNs at the Beverly facility were "supervisors" within the meaning of § 2(11) of the Act. See id. Beverly argued that the LPNs should be excluded from the bargaining unit as supervisors. Of the approximately eighty- two employees in the proposed bargaining unit, sixteen were LPNs. These LPNs had a certain amount of authority over another forty-six employees included in the proposed bargaining unit, including the authority to arrange work schedules and to evaluate the performance of these employees. On July 31, 1992, the Regional Director for Region Six issued a decision and direction of election in which he found that the LPNs were not supervisors under § 2(11) of the Act. Beverly filed a timely request for review of the Regional Director's decision, and on August 27, 1992, the Board denied Beverly's request for review. On that same day, an election was conducted at the Beverly facility. In con- ducting the election, the Regional Director permitted the LPNs to vote, and he did not segregate the ballots of the LPNs from the ballots of the other employees. In addition, Beverly was not permitted to challenge the ballots of the LPNs. Of the eighty-two eligible employ- ees, seventy-five employees voted. Fifty-five employees voted in favor of Union representation, while twenty employees voted against Union representation. Three ballots were challenged. On September 9, 1992, the Regional Director, noting that no objections to the elec- _________________________________________________________________ 1 The unit described on the official notice of election included: All full-time and regular part-time service and maintenance employees, including licensed practical nurses, certified nurse's aides, nurse's aides, dietary employees, environmental service employees, medical records coordinator, central supply clerk, physical therapy aides, activities director's assistants and mainte- nance employees employed by [Beverly] at its Morgantown, West Virginia facility; excluding all office clerical employees, registered nurses and guards, professional employees and supervisors as defined in the Act. (J.A. 21 (emphasis added)). 3 tion had been filed, certified the Union as the exclusive bargaining representative of the unit employees. Two days later, on September 11, 1992, we denied enforcement of the Board's order in a case in which the issue was whether LPNs at a different health care facility were statutory supervisors under § 2(11) of the Act. See Beverly Calif. Corp. v. NLRB, 1992 WL 223815 (4th Cir. Sept. 11, 1992) (unpublished opinion). Following this decision, Beverly filed a request for reconsideration of the Board's certification of the election in this case. On May 26, 1993, the Board granted Beverly's request for reconsideration, finding that it raised substantial issues warranting review. Almost one year later, on May 11, 1994, the Board affirmed the Regional Director's decision and direction of election. Two weeks later, on May 23, 1994, the Supreme Court issued its decision in NLRB v. Health Care & Retirement Corp., 511 U.S. 571 (1994), in which it rejected the "patient care analysis" on which both the Regional Director and the Board had relied in finding that the LPNs in this case were employees and not "supervisors" under § 2(11) of the Act. In light of this decision, Beverly filed a second motion for reconsideration. On August 4, 1994, the Board granted Beverly's motion and remanded the case to the Regional Director for reconsideration in light of the Supreme Court's decision. On October 24, 1994, a hearing officer conducted a hearing pursu- ant to the Board's August 4 order. At the hearing, the parties stipu- lated that the LPNs at the Beverly facility were, at all times relevant to the petition, "supervisors" within the meaning of § 2(11) of the Act. However, the parties disputed whether the election should be set aside and the certification revoked or, alternatively, whether the certifica- tion could be merely amended to exclude the LPNs from the bargain- ing unit. Additionally, Beverly argued that the entire election process, including the Union's showing of interest, was tainted by the LPNs' active participation in the election campaign. In support of this posi- tion, Beverly attempted, but was not permitted, to put on evidence with regard to the LPNs' conduct prior to the election. Specifically, Beverly proffered the testimony of LPNs who were employed during the 1992 campaign, the content of which was that a number of LPNs attended Union meetings, signed Union authorization cards, solicited 4 authorization cards from non-supervisory employees, actively cam- paigned for the Union with employees over whom the LPNs exercised supervisory authority, and encouraged such employees to vote for the Union. On November 16, 1994, the Regional Director issued a supplemen- tal decision revoking the certification of the Union and directing a new election. The Regional Director found that the exclusion of the LPNs represented a major change in the scope and character of the bargaining unit, thereby invalidating the election and requiring that a new election be held. The Regional Director rejected, however, Bev- erly's contention that the representation petition should be dismissed because of the participation of the LPNs in the organizational cam- paign, noting that the validity of the showing of interest is not subject to litigation and that Beverly's challenge thereto, coming two years after the filing of the petition, was untimely. On December 2, 1994, the Union filed a request for review of the Regional Director's decision; Beverly did not file a request for review. On October 31, 1995, the Board issued a decision in which it reinstated the Union's certification, amending the certified bargain- ing unit to exclude LPNs and vacating the Regional Director's direc- tion of a new election. Contrary to the Regional Director, the Board found that the exclusion of the LPNs, one of ten job classifications and sixteen of eighty-two employees, from the pre-election unit did not change the basic character and scope of the unit. In addition, the Board noted that the election was not close and that, even without the votes of the LPNs, the Union had won by at least nineteen votes. Not- ing that Beverly had not requested review of the Regional Director's decision, the Board declined to address the Regional Director's rejec- tion of Beverly's argument that the LPNs' pre-election conduct tainted both the showing of interest and the election and required the dismissal of the representation petition. On November 15, 1995, the Union requested that Beverly meet for negotiations. On December 1, 1995, two days after the issuance of the amended certification, the Union renewed its bargaining request. Bev- erly did not respond, and on December 6, 1995, the Union filed an unfair labor practice charge against Beverly, alleging violations of §§ 8(a)(1) and (5) of the Act. See 29 U.S.C. §§ 158(a)(1) and (5). The 5 Union filed an amended charge on January 23, 1996, and on February 1, 1996, the Regional Director issued a complaint, to which Beverly timely filed an answer. On April 2, 1996, the Board's general counsel filed a motion for summary judgment with the Board. On April 10, 1996, the Board issued a notice to show cause why the motion for summary judgment should not be granted. On April 23, 1996, Beverly filed its response, reiterating its position that the exclusion of the LPNs from the bar- gaining unit changed its character and scope so as to render the elec- tion invalid. In addition, Beverly argued that the alleged participation of the LPNs in the organizational campaign raised a serious question as to the validity of the election. On May 15, 1996, the Board granted the general counsel's motion for summary judgment and ordered Beverly to bargain with the Union. In its decision, the Board found that all representation issues raised by Beverly had been or could have been litigated in the prior representation proceeding. On August 28, 1996, the Board filed an application in this court for enforcement of the Board's order. II. The Board enjoys broad latitude to develop and implement the pro- cedures for certification elections. See Sears, Roebuck & Co. v. NLRB, 957 F.2d 52, 55 (2d Cir. 1992); Nightingale Oil Co. v. NLRB, 905 F.2d 528, 531 (1st Cir. 1990); NLRB v. Lorimar Productions, Inc., 771 F.2d 1294, 1298 (9th Cir. 1985). As a general rule, the Board's findings are to be accepted if they are supported by substan- tial evidence on the record as a whole. See Monongahela Power Co. v. NLRB, 657 F.2d 608, 611 (4th Cir. 1981). We review the Board's decision to uphold an election for abuse of discretion. See NLRB v. VSA, Inc., 24 F.3d 588, 592 (4th Cir. 1994); NLRB v. Manufacturers Packaging Co., Inc., 645 F.2d 223, 225 (4th Cir. 1981). III. As a general rule, the burden is on the party challenging the valid- ity of the election to prove that its fairness has been compromised. 6 See Manufacturers Packaging Co., 645 F.2d at 225. The test is not whether optimum practices were followed, but whether the manner in which the election was held raises a reasonable doubt as to its valid- ity. See Nightingale Oil Co., 905 F.2d at 531; see also NLRB v. Her- bert Halperin Distrib. Corp., 826 F.2d 287, 290 (4th Cir. 1987) (party challenging election "must show by specific evidence `not only that improprieties occurred, but also that . . . they materially affected the election results'" (citation omitted)). Nevertheless, "when the Board has effectively denied employees the right to make an informed choice in a representation election," an order of the Board will not be enforced. See Hamilton Test Sys. v. NLRB, 743 F.2d 136, 142 (2d Cir. 1984). Where employees are led to believe that they are voting on a partic- ular bargaining unit and that bargaining unit is subsequently modified post-election, such that the bargaining unit, as modified, is fundamen- tally different in scope or character from the proposed bargaining unit, the employees have effectively been denied the right to make an informed choice in the representation election. See NLRB v. Parsons Sch. of Design, 793 F.2d 503, 506-08 (2d Cir. 1986); Lorimar Productions, 771 F.2d at 1301-02; Hamilton Test Sys., 743 F.2d at 140-42. Thus, the Board may not "inform employees that they are voting for representation in [one] unit and later . . . consider the ballot as a vote for representation in a [different] unit." Hamilton Test Sys., 743 F.2d at 140; see also Lorimar Productions , 771 F.2d at 1301 (quoting Hamilton Test Sys.). In determining whether an employee's freedom to make an informed choice has been compromised by the subsequent modification of the bargaining unit for which he voted, we consider several factors: (1) the difference in size between the bar- gaining unit proposed to the employees before the election and the final size of the unit; (2) the character and scope of the pre- and post- election units; and (3) the closeness of the election results. See Sears, Roebuck & Co., 957 F.2d at 57. The first factor to be considered in determining whether a post- election modification of the bargaining unit necessitates the invalida- tion of the election is the difference in size between the bargaining unit proposed to the employees before the election and the final size of the unit. In this case, the exclusion of the LPNs from the bargain- ing unit decreased its size by approximately 20%. Although the Board 7 argues that the exclusion of one out of ten employee classifications and 20% of the number of eligible voters is not a significant change in the size of the bargaining unit, it is certainly true, as the Second Circuit Court of Appeals has recognized, that "a smaller bargaining unit may be less attractive to potential union members because of reduced bargaining power." Sears, Roebuck & Co., 957 F.2d at 57. Thus, while a 20% increase in the size of the bargaining unit may not be likely to affect the other employees' votes, a 20% decrease may well have such an effect. Compare Parsons Sch. of Design, 793 F.2d at 507-08 (10% decrease in size of unit) with Nightingale Oil Co., 905 F.2d at 57 (10% increase in size of unit). Accordingly, this factor weighs in favor of denying the Board's application for enforcement. The second factor to be considered is the character and scope of the pre- and post-election units. This factor requires the court to consider the similarity or dissimilarity of job classifications between the origi- nal and final units and the possibility that the final unit split an other- wise unified work force. See Sears, Roebuck & Co., 957 F.2d at 57- 58. For example, in Parsons School of Design, the Second Circuit held that an election in favor of union representation should be set aside because of the post-election modification of the bargaining unit to exclude all full-time faculty members. See Parsons Sch. of Design, 793 F.2d at 507-08. As originally proposed and voted upon, the bar- gaining unit included both part-time and full-time faculty members. See id. at 504-05. Although the exclusion of the full-time faculty members resulted in a numerical reduction of only 10% of the bar- gaining unit, the Second Circuit recognized that full-time faculty members were "particularly important to the school and carr[ied] great weight in all matters affecting it," making their exclusion from the bargaining unit particularly significant. Id. at 507-08. In this case, although the LPNs numbered only sixteen of eighty- two eligible employees in the bargaining unit, they supervised forty- six of the eligible employees. As supervisors, they had the responsi- bility for scheduling the daily assignments for their subordinates and for evaluating their performance. Because the LPNs wielded such a significant amount of influence, it is likely that a number of the employees voted in favor of Union representation in this case with the understanding that the bargaining unit would, if successful, include their supervisors. It is also quite possible that had the proposed bar- 8 gaining unit not included the LPNs, many of their forty-six subordi- nates may not have voted in favor of representation, fearing a division between them and their supervisors. Therefore, the influence wielded by the LPNs may have been an important factor in the decision of other employees to vote in favor of representation. The impact of their exclusion from the bargaining unit, then, may go well beyond their sixteen votes. In short, because of the importance of the LPNs to the proposed bargaining unit as a whole, their subsequent exclusion post- election fundamentally changed the character of the unit, and this fac- tor also weighs in favor of denying the Board's application for enforcement. The third and final factor to be considered when determining whether to invalidate an election based on a post-election modifica- tion of the bargaining unit is the closeness of the election results. See Sears, Roebuck & Co., 957 F.2d at 57. In this case, the Union victory was decisive with fifty-five votes cast in favor of Union representa- tion and twenty votes cast against Union representation. Even if every LPN who voted in the election voted in favor of Union representation, the exclusion of the LPNs as part of the bargaining unit would result in an election in favor of Union representation by nineteen votes. This factor, then, weighs in favor of granting the Board's application for enforcement. Considering all three factors in this case, we conclude that the Board abused its discretion in certifying a bargaining unit that dif- fered significantly both in the number of employees included in the bargaining unit and in the unit's character and scope from the bar- gaining unit voted upon by the employees. Although the strength of the Union's victory clearly weighs against invalidation of the election, this factor must be considered in light of the fact that the excluded LPNs had direct supervisory authority over forty-six of the sixty-six employees remaining in the bargaining unit after the LPNs' exclusion. If only ten of those employees who voted in favor of Union represen- tation did so because of their understanding that their supervisors would also be part of the bargaining unit, the election results might have been different had the LPNs been excluded prior to the vote. To the extent that the third factor weighs against the invalidation of the election, then, it is outweighed by the first two factors, both of which weigh heavily in favor of invalidation of the election. Because the 9 post-election exclusion of the LPNs from the bargaining unit so sig- nificantly changed the character and scope of the bargaining unit, the Beverly employees were effectively denied the right to make an informed choice in the representation election. As a result, the Board abused its discretion in certifying the election results, and its order compelling Beverly to bargain with the Union based on the validity of the election may not be enforced. We reject the Board's suggestion that our decision is inconsistent with this court's decision in Prudential Ins. Co. of Am. v. NLRB, 832 F.2d 857 (4th Cir. 1987). In Prudential, one employee was errone- ously included in a proposed bargaining unit because she was a "con- fidential employee." See id. at 860 (recognizing that certain confidential employees are excluded from collective bargaining units). Including the confidential employee, the bargaining unit con- tained seven members. The vote in the representation election was 4- 1 in favor of union representation, with two challenged votes sealed and not counted in the election tally. On appeal, we rejected Pruden- tial's argument that the exclusion of the confidential employee neces- sarily required the invalidation of the representation election because the other employees' votes may have been different had they known that the confidential employee was not a member of the bargaining unit. See id. at 861. Instead, we remanded the case for the Board to consider whether, based on the Board's resolution of the two chal- lenged ballots, the exclusion of the confidential employee's vote could have affected the union's victory. See id. The Board argues that Prudential requires us to uphold its decision validating the representation election in this case, suggesting, in essence, that Prudential stands for the proposition that a new election is unnecessary where a proposed bargaining unit has been modified post-election, unless we conclude, based on a numerical analysis alone, that the election results would have been different had the employees voted on the appropriate bargaining unit. We do not believe that Prudential stands for such a broad proposition. Instead, Prudential stands for the narrower proposition that a new election is not always required where a bargaining unit is modified post-election; rather, the court must assess the potential impact of the modification on several levels. In that decision, for example, when distinguishing Hamilton Test Systems, we specifically noted that the modified bar- 10 gaining unit in Hamilton Test Systems was less than half the size and "`considerably different in character'" from the proposed bargaining unit, while the modified bargaining unit in Prudential was only one employee smaller than the proposed bargaining unit. Id. at 861 (cita- tion omitted). Our discussion in Prudential about Hamilton Test Systems, then, suggests that, in addition to the closeness of the elec- tion results, both the size and the character and scope of the modified bargaining unit are appropriate considerations when deciding whether an election must be invalidated. When all relevant factors are considered, it is clear that the circum- stances in this case differ significantly from those presented to the court in Prudential. First, the modified bargaining unit in Prudential contained only one less employee than the proposed bargaining unit, while the modified bargaining unit in this case contains sixteen fewer employees. Second, and more importantly, in this case, there was a supervisory relationship between the LPNs and the other employees, and it affected over two-thirds of the employees remaining in the bar- gaining unit after the LPNs were excluded. Therefore, while the exclusion of only one employee who does not have supervisory authority over other members of the proposed bargaining unit may not change the character and scope of the bargaining unit, as the court implicitly held in Prudential, the exclusion of the LPNs in this case who supervised most of the other employees in the bargaining unit fundamentally changed the character and scope of the bargaining unit. Because the circumstances in this case are fundamentally different from those presented to the court in Prudential , a different conclusion is warranted. IV. For the foregoing reasons, we deny the Board's application for enforcement of its bargaining order against Beverly. 2 _________________________________________________________________ 2 In addition to arguing that the Board abused its discretion in certifying the election results in light of the post-election modification of the bar- gaining unit, Beverly argues that the election should be set aside because of inappropriate supervisor behavior during the election campaign. Because we conclude that the Board's application for enforcement of its bargaining order should be denied on the former ground, we decline to consider the latter. 11 ENFORCEMENT DENIED.3 DIANA GRIBBON MOTZ, Circuit Judge, dissenting: I would enforce the bargaining order of the National Labor Rela- tions Board certifying the election. In formulating procedures for the conduct of an election, the Board is entitled to make "justifiable and reasonable adjustment[s] of the democratic process," to accommodate interests such as finality and minimizing delay. NLRB v. A.J. Tower Co., 329 U.S. 324, 333 (1946). Furthermore, the Board's factual find- ings are conclusive if supported by substantial evidence in the record as a whole, even if "we might have decided the case differently de novo." National Poster, Inc. v. NLRB, 885 F.2d 175, 178 (4th Cir. 1989). The Board's determination that the election was fairly con- ducted can be set aside only for an abuse of discretion. See Nightingale Oil Co. v. NLRB, 905 F.2d 528, 531 (1st Cir. 1990); see also NLRB v. VSA, Inc., 24 F.3d 588, 592 (4th Cir. 1994). After a careful review of the record, I cannot conclude that the Board abused its discretion. Accordingly, I respectfully dissent from the majority's holding to the contrary. I agree with the majority that the non-exclusive, three factor test stated in Sears, Roebuck & Co. v. NLRB, 957 F.2d 52, 57 (2d Cir. 1992) governs our analysis. Under Sears, we look to: "(1) the differ- ence in size between the bargaining unit proposed to the employees before the election and the final size of the unit; (2) the character and scope of the pre- and post-election unit; and (3) the closeness of the election results." Id. _________________________________________________________________ 3 Because over five years have elapsed since the representation petition in this case was filed and because of the particular factual circumstances of this case, we simply deny the Board's application for enforcement and choose not to remand this case for further proceedings by the Board. See NLRB v. Lundy Packing Co., 68 F.3d 1577, 1583 (4th Cir. 1995) (deny- ing petition for enforcement of bargaining order where the Board certi- fied an inappropriate bargaining unit). In accordance with our decision in NLRB v. Lundy Packing Co., 81 F.3d 25 (4th Cir. 1996), the Board no longer has jurisdiction over this case and may not revive the represen- tation petition that is the subject of this decision. See id. at 25-26. 12 Unlike the majority, however, I believe, as the Board found, that the facts of this case do not indicate "a significant change in unit size to warrant setting aside the election" under the first prong of the Sears test. In this case the pre-election unit included 82 eligible voters, including 16 LPNs. The exclusion of the LPNs reduced the size of the bargaining unit by approximately twenty percent. In upholding the election, the Board compared this modification with those in another Board decision, Toledo Hospital, 315 NLRB 594 (1994), in which the Board upheld an election after the addition of a number of employees expanded the unit by 19.5 percent. The Board then distinguished Hamilton Test Sys. v. NLRB, 743 F.2d 136 (2d Cir. 1984) and NLRB v. Lorimar Productions, Inc., 771 F.2d 1294 (9th Cir. 1985), where the unit differentials exceeded 40 percent, by the size of the numerical change in the unit. The Board's conclusion maintains consistency with this court's application of the Sears test in Prudential Ins. Co. v. NLRB, 832 F.2d 857, 859 (4th Cir. 1987). In Prudential, the employees had voted for the union 4-1, with two votes disputed. We held that one of the voters was a confidential employee who could not be a member of the bar- gaining unit. Id. at 860. Despite this twenty percent change in the size of the unit, the same differential at stake in the present case, we did not order a new election, as the employer requested. Id. at 861. Instead, we remanded the case to the NLRB "to consider the two votes challenged by the Union and not included in the tally" to estab- lish whether the disputed votes and the removal of the confidential employee "would affect the Union's majority status." Id. If the Union lost its majority it would "be necessary to hold a new election in a properly constituted unit. Otherwise, the representation result must stand." Id. Thus, despite the removal of the confidential employee from a five member unit, a potential loss of 20 percent, we held that a new election was not automatically required. The majority, citing Sears, places significant weight on the fact that the unit size approved by the Board was smaller than that voted on by the employees, just as it was in NLRB v. Parsons School of Design, 793 F.2d 503 (2d Cir. 1986)(invalidating the election procedures). In Parsons, however, the election was extremely close and the court emphasized that the character and scope of the unit had been signifi- cantly altered. Parsons, 793 F.2d at 504. Neither of these factors are 13 present here. Furthermore, though it may be true that "a smaller bar- gaining unit may be less attractive to potential union members because of reduced bargaining power," Sears , 957 F.2d at 57, I see no reason why the opposite may not be true as well. Some union members may prefer smaller units that better represent their specific interests, or they may prefer to separate themselves from other employees who do not perform similar work, earn equivalent pay, or otherwise share their community of interests. The second Sears factor focuses on the change in character and scope of the unit before and after modification: This factor requires consideration of the similarity or dis- similarity of job classifications between the original and final units and the chance that the ultimate unit split an oth- erwise unified workforce. . . . If the new composition of the unit excludes workers performing under conditions similar to those already included in the unit, employees are more likely to have voted differently because they may feel that their individual interests cannot be best represented by a group consisting of diverse, potentially adverse interests. Sears, 957 F.2d at 57-58 (citations omitted). In the present case, the Board found that "the scope and character of the unit has not changed to any significant extent, and the unit remains the service and maintenance unit petitioned-for originally." In my view, although I recognize that it is a close question, the fact that there is no evidence that the LPNs had greater bargaining power than the other unit members as well as the large margin of victory provide sufficient evidentiary support for these findings. Beverly argues to the contrary, asserting that because the LPNs supervised approximately half of the members of the bargaining unit, the LPNs' removal is very significant and would alter the vote. In making this contention, the company relies heavily on the Second Circuit's decision in Parsons. In Parsons, both full-time and part-time faculty had originally voted for the union but the court subsequently removed the full-time faculty from the unit. This action was held to have significantly altered the unit because the"full-time faculty 14 employees excluded from the unit, while comparatively few in num- ber, are particularly important to the school and carry great weight in all matters affecting it." Parsons, 793 F.2d at 507. Because "the elec- tion was a very close one" and "the part-time instructors may have feared insufficient strength in a unit comprising less than all of the faculty," the Second Circuit ordered a new election. Id. at 508. Although there are similarities between this case and Parsons, there are two critical differences. First, the remaining employees in the bargaining unit at issue in the case at hand are all classified as full-time. There is no evidence of a bargaining power differential between the LPNs and the other Beverly employees like that between the full-time and part-time faculty at issue in Parsons. Second, the election in the present case was not nearly as close as that in Parsons, where a vote change by "only four of the 99 part-time instructors who voted for the Union" would have changed the result of the election. Id. at 508. Here, even assuming that all 16 LPNs voted for the union, the margin of victory would still be 39-22. The final factor addressed under the Sears test is the margin of vic- tory in the election. "A narrow victory heightens the need to scruti- nize the election process to ensure that votes would not change with a more fully informed electorate. [Parsons , 793 F.2d at 507]. This factor must be observed in combination with the two other factors addressed above in order to see if the concerns uncovered in examin- ing the other factors would have led to a vote change sufficient to alter the election outcome." Sears, 957 F.2d at 58 (emphasis added). The purpose of this third factor is to weigh the potential votes lost through the change in the size, scope, and character of the unit against the margin of victory for the union. In instances where the margin of victory is small, any change in the size of the unit or character may necessitate a new election. See Parsons, 793 F.2d at 508 (noting that if "four of the 99 part-time instructors" changed their vote, union would lose); Lorimar Productions, 771 F.2d at 1302 ("The vote was so close that the union would have lost had one employee voted dif- ferently."); Hamilton Test Systems, 743 F.2d at 141 ("A change of one vote from union to non-union would have altered the outcome of the election."); Monongahela Power Co. v. NLRB , 657 F.2d 608, 609 (4th Cir. 1981)(noting that the union won vote 58-56). 15 In the present case, as the Board found and the majority acknowl- edges, the election was not close: 55 employees voted for the union and 22 against. Even assuming that all 16 LPNs voted for the union, an assumption with no support in the record, the margin would still be 39-22. In weighing this margin of victory with the other two fac- tors, it seems to me that the Board did not abuse its discretion by con- cluding that the exclusion of the LPNs would not change the votes of the employees in a new election. Beverly is unable to cite a single case in which a court ordered a new election or refused to abide by the Board's decision not to do so when the margin of the original union victory was as great as that in this case. As the majority recog- nizes, the National Labor Relations Board "enjoys broad latitude to develop and implement the procedure for certification of elections." Ante at 9. Given the margin of victory, I cannot agree with the major- ity that the Board exceeded that broad authority and abused its discre- tion. 16