PER CURIAM:

Treating the petition for rehearing en banc as a petition for panel rehearing, the petition for panel rehearing is DENIED. The court having been polled at the request of one of its members, and a majority of the judges who are in regular active service and not disqualified not having voted in favor (FED. R. APP. P. 35 and 5TH CIR. R. 35), the petition for rehearing en banc is DENIED.

In the en banc poll, 6 judges voted in favor of rehearing (Judges Jolly, Jones, Smith, Clement, Owen, and Elrod), and 9 judges voted against rehearing (Chief Judge Stewart and Judges Davis, Dennis, Prado, Southwick, Haynes, Graves, Higginson, and Costa).

E. GRADY JOLLY, Circuit Judge, joined by JONES, SMITH, CLEMENT, OWEN, and ELROD, Circuit Judges, dissenting from the denial of rehearing en banc:

This appeal presents another example of the current National Labor Relations Board's ("NLRB") determination to disregard established principles of labor law. The NLRB certified a small bargaining unit consisting of only the cosmetics and fragrances employees at a Macy's department store in Saugus, Massachusetts. On appeal, the panel denied Macy's' petition for review and granted the NLRB's application for enforcement of its unfair labor practices order, which ordered Macy's to bargain with the Union.1 Macy's, Inc. v. NLRB, 824 F.3d 557, 560-61 (5th Cir. 2016). On petition for en banc review, the en banc Court, in a split vote, denied further review. I respectfully dissent from that denial.

As an initial matter, the panel erred by allowing the NLRB's decision to stand when it and its underlying foundations are marred by the misapplication of the NLRA and its historical interpretation. As the NLRB acknowledges, it has long held that, in the retail industry, storewide units of salesforce employees are the presumptively appropriate collective bargaining unit. Macy's & Local 1445, 361 N.L.R.B. 4, at *17-19 (2014); see also, e.g., I. Magnin & Co., 119 N.L.R.B. 642 (1957); May Dep't Stores, 97 N.L.R.B. 1007 (1952). Even if this presumption has been overcome on infrequent occasions, the NLRB has only authorized smaller units where a petitioned-for unit of employees has "a `mutuality of interests' not shared by all other selling employees ... and are `sufficiently different' from the other selling employees so as to justify representation on a separate basis." Macy's & Local 1445, 361 N.L.R.B. 4, at *20. Such cases have been rare for an obvious reason: no matter the titular differences, such as employees' assignment to different departments, all salesforce workers have the same basic employment, skills, interests, function, and working conditions.

Here, there are no circumstances that isolate the cosmetics and fragrances employees from the presumptive bargaining unit of all salesforce employees. The NLRB nonetheless applied, inaptly, the two-prong standard from Specialty Healthcare & Rehab. Ctr. of Mobile, 357 N.L.R.B. 83 (2011), to allow the smaller and select unit that the Union had successfully organized.

There are statutory constraints on the NLRB's evaluation of a union's requested collective bargaining unit. Section 9(c)(5) of the NLRA expressly provides that "the extent to which the employees have organized shall not be controlling." 29 U.S.C. § 159(c)(5). Courts have interpreted this to mean that the extent of union organization may only be "`consider[ed] ... as one factor' in determining whether a proposed unit is appropriate." Blue Man Vegas, LLC v. N.L.R.B., 529 F.3d 417, 421 (D.C. Cir. 2008) (quoting NLRB v. Metro. Life Ins. Co., 380 U.S. 438, 442, 85 S.Ct. 1061, 13 L.Ed.2d 951 (1965)). But, here, the only justification for a unit of only cosmetics and fragrances employees is that it reflects the apex of the Union's organizational strength. Indeed, the Union failed in two efforts to organize larger bargaining units at this store. The Union was only successful on its third try: this time with a micro-unit of cosmetics and fragrances employees that evidently reflected its greatest strength. But the en banc Court must acknowledge that the Supreme Court has explained that "the enforcing court should not overlook or ignore an evasion of the § 9(c)(5) command." Metro. Life Ins. Co., 380 U.S. at 442, 85 S.Ct. 1061. In short, the NLRB's decision here challenges this admonition.

Furthermore, the panel decision pays little respect to one of the underlying policies of the NLRA: the promotion of labor peace and stability.2 Peace and stability are weakened by the balkanization of bargaining units in a single, coordinated workplace. NLRB v. R. C. Can Co., 328 F.2d 974, 978-79 (5th Cir. 1964). In this case, the NLRB sacrificed considerations of promoting labor peace by using a rationale that approved a small, carved-out bargaining unit that contains no real limiting principle in future cases. For example, nothing in the NLRB's rationale prevents a dozen micro-units within a retail store's salesforce — all fraught with mini-bargaining at multiple times and the possibility of disputes and mini-strikes occurring continually over the working year. One is led to assume, as the amici suggest, that three bowtie salesman would be an appropriate bargaining unit if they sold bowties at a separate counter from other merchandise. So much for promoting labor peace and stability.

On a different level, the panel has effectively disregarded our own precedent in NLRB v. Purnell's Pride, Inc., 609 F.2d 1153 (5th Cir. 1980). When the NLRB "exercises the discretion given to it by Congress, it must `disclose the basis of its order' and `give clear indication that it has exercised the discretion with which Congress has empowered it.'" Id. at 1161 (quoting Metro. Life Ins. Co., 380 U.S. at 443, 85 S.Ct. 1061). Although the panel acknowledged Purnell's Pride, it gave the NLRB a pass on its requirements: in the words of one panel member, the NLRB's decision reads like "a bad law school exam."

And now, from the broad strokes, to the analysis.

I.

"This court ... reviews unit determinations only to determine `whether the decision is arbitrary, capricious, an abuse of discretion, or lacking in evidentiary support.'" Macy's, 824 F.3d at 563 (citation omitted). But this deference does not require the Court to "bow to the mysteries of administrative expertise." E.g., Envtl. Def. Fund, Inc. v. Ruckelshaus, 439 F.2d 584, 597 (D.C. Cir. 1971). This Court has "refused to enforce [NLRB] orders where they have no reasonable basis in law" because they "fail[ed] to apply the community of interest standard" and where "the reasons supporting the Decision ... [were] not sufficiently articulated to permit proper judicial review." NLRB v. Magna Corp., 734 F.2d 1057, 1061, 1064 (5th Cir. 1984) (citation and internal quotation marks omitted); Purnell's Pride, 609 F.2d at 1161. The panel opinion is a troublesome decision that permits the NLRB's decision to stand despite the fact that it contains both of these critical flaws; troublesome especially when we have precedent that rejects the breezy analysis employed by the threesome.

A.

The NLRB abused its discretion by applying an incorrect standard for analyzing the first prong of the Specialty Healthcare framework: whether the petitioned-for employees share a community of interest. Moreover, the flawed analysis demonstrates that the NLRB's determination was controlled by the extent of union organization, which NLRA § 9(c)(5) explicitly prohibits.

1.

The NLRA constrains the NLRB's evaluation of a union's proffered collective bargaining unit. As noted, NLRA § 9(c)(5) provides that "the extent to which the employees have organized shall not be controlling," although it "may be `consider[ed]... as one factor' in determining whether a proposed unit is appropriate." 29 U.S.C. § 159(c)(5); Blue Man Vegas, 529 F.3d at 421 (citation omitted). Thus, "while still taking into account the petitioner's preference," the NLRB "must proceed to determine, based on additional grounds," whether "the proposed unit is ... appropriate." Specialty Healthcare, 357 N.L.R.B. 83, at *13.

"To guide its discretion, and to avoid giving controlling weight to the extent of organization," the NLRB traditionally uses a multi-factor community of interest analysis to determine whether a petitioned-for unit is appropriate. Nestle Dreyer's Ice Cream Co. v. NLRB, 821 F.3d 489, 495 (4th Cir. 2016) (citation omitted). In Specialty Healthcare, the NLRB clarified3 the traditional principles of unit determination and explained how the traditional standard applies when an employer contends that the appropriate unit contains more employees than those in the petitioned-for unit, as Macy's does here. 357 N.L.R.B. 83, at *12-20. In such cases, the NLRB applies a two-step test.

In the first step, the NLRB decides whether the petitioned-for unit is prima facie appropriate. It begins by determining whether the employees in the petitioned-for unit "are readily identifiable as a group (based on job classifications, departments, functions, work locations, skills, or similar factors)." Id. at *17. This first step is completed by examining whether "the employees in the group share a community of interest." Id. In making this decision, the NLRB examines:

Macy's, 824 F.3d at 568-69 (emphasis in original) (quoting Specialty Healthcare, 357 N.L.R.B. 83, at *14). If it finds that a petitioned-for unit is "readily identifiable as a group" and "that the employees in the group share a community of interest after considering the traditional criteria," the NLRB proceeds to the second step. Specialty Healthcare, 357 N.L.R.B. 83, at *17.

In the second step, the burden shifts from the petitioner to the employer to show that "employees in [a] larger unit share an overwhelming community of interest with those in the petitioned-for unit." Id. An employer satisfies this burden if it shows that "there `is no legitimate basis upon which to exclude certain employees from it'" so that the community of interest "factors `overlap almost completely.'" Id. at *16 (quoting Blue Man Vegas, 529 F.3d at 421-22). If the employer cannot make this showing, the NLRB "will find the petitioned-for unit to be an appropriate unit." Id. at *17.

But the NLRB itself has more than a perfunctory obligation when analyzing the community of interest factors in the first step: the NLRB must compare and contrast the employees in the group with each other and with employees outside of the group. The NLRB has repeatedly recognized the importance of this comparison. It has stated, for example, that:

Wheeling Island Gaming, Inc., 355 N.L.R.B. 637, at *1 n.2 (2010) (emphasis in original) (citations and internal quotation marks omitted). The NLRB maintained this requirement in Specialty Healthcare. It formulated the community of interest test detailed above, which emphasizes this comparison, and applied the test using an analysis replete with distinctions between the employees in the petitioned-for unit and excluded employees. Specialty Healthcare, 357 N.L.R.B. 83, at *14. Moreover, the Fifth Circuit and its sister circuits have recognized that employees in a petitioned-for unit must be compared with employees who share common interests but have nonetheless been excluded from the petitioned-for unit.4

Ultimately, in applying Specialty Healthcare, the NLRB must guard against violating NLRA § 9(c) by making "arbitrary exclusions." Nestle, 821 F.3d at 499. If it does not compare employees in the petitioned-for group with excluded employees in the first step or if it only identifies "meager differences" between these employees, the NLRB "conduct[s] a deficient community-of-interest analysis" that "fails to guard against arbitrary exclusions" and creates an "apparent union gerrymander." Id.; see also Blue Man Vegas, 529 F.3d at 425-26; NLRB v. Lundy Packing Co., 68 F.3d 1577, 1580, 1580-81 (4th Cir. 1995), supplemented, 81 F.3d 25 (4th Cir. 1996); see Macy's, 824 F.3d at 568-69. This conduct violates § 9(c) because, "[b]y rubber-stamping [a union's petitioned-for unit] and then applying the overwhelming-community-of-interest test, `the [NLRB would] effectively accord[] controlling weight to the extent of union organization.'" Nestle, 821 F.3d at 499 (citation omitted).

2.

Here, the NLRB conducted a deficient analysis of whether the petitioned-for unit of cosmetics and fragrances employees was prima facie appropriate. The NLRB began by incorrectly phrasing step one of the Specialty Healthcare analysis as being concerned only with "whether employees in a proposed unit share a community of interest." Macy's & Local 1445, 361 N.L.R.B. 4, at *10 (emphasis added). Then, in conducting this community of interest analysis, the NLRB barely noticed how the employees in the petitioned-for group differed from excluded employees and made no effort to explain how the admittedly questionable difference it identified was not, in fact, "meager."

The NLRB discussed similarities between employees within the petitioned-for group, but it did not discuss similarities between the included employees and the excluded employees. Id. at *10-11. For example, it addressed Macy's' arguments as to why employees within the petitioned-for group did not share similar interests. Id. at *11. But it only acknowledged Macy's' contention that the cosmetics and fragrances employees' interests did not meaningfully differ from those of other sales employees once it advanced to step two of the Specialty Healthcare analysis. Id. at *11-17.

The NLRB also cited only one5 distinction between cosmetics and fragrances employees and Macy's' other selling employees: only cosmetics and fragrances employees sell fragrance and cosmetic products to customers. Id. at *10. This distinction is, however, hollow and just plain meaningless. The NLRB had to admit that there was "evidence regarding cosmetics and fragrances products being rung up in other departments." Id. at *13 n.41. And the NLRB did not explain why this purported difference had contextual substance or was not "meager" — an explanation that was particularly necessary because the NLRB later conceded that "the petitioned-for employees and other selling employees perform similar, related duties." Id. at *14.

3.

Regrettably, the panel has failed properly to grasp and to apply the principles that guide step one of the Specialty Healthcare analysis. It is clear to any reasonable reader that the panel did not require the NLRB actually to engage the crucial step of rigorously weighing the community of interest factors by comparing the employees in a petitioned-for unit with employees outside of that unit. Instead, in a blow-by treatment of whether the NLRB applied the correct standard, the panel stated without further explanation "[t]hat [rigorously weighing the factors] is precisely what the [NLRB] has done in the instant case. As a result, the test and its application do not violate Section 9(c)." Macy's, 824 F.3d at 568.

This conclusionary expression does not reconcile the NLRB's analysis with the NLRA's and Specialty Healthcare's requirements. The fact remains that, in its analysis under Specialty Healthcare's first prong, the NLRB articulated and applied the wrong standard. The NLRB failed to consider any of the similarities between included and excluded employees, only identified one questionable distinction between them, and did not explain how that distinction was meaningful. Because the NLRB did not apply the correct community of interest standard, its decision, in the final analysis, had "no reasonable basis in law" and was therefore an abuse of discretion. Magna Corp., 734 F.2d at 1061, 1064 (citation omitted).

Moreover, and crucially, this case is a picture-perfect example of the NLRB violating the NLRA by approving a bargaining unit defined by the limited success of a union's organizational efforts in the larger and appropriate unit. This bypassing of statutory barriers has been achieved by avoiding an analysis of the guiding precedents in shaping bargaining units. After the Union was stymied from organizing a storewide unit to join a multi-store unit and lost an election for a stand-alone storewide unit, the Union cherry-picked a unit of only cosmetics and fragrances employees — the group apparently most favorable to the Union's organization efforts.6 And the NLRB allowed it to hobble across the finish line as a survivor "substitute" bargaining unit. The NLRB has long used a thorough community of interest test, which compares employees within and outside of the proposed unit, "to avoid giving controlling weight to the extent" to which employees have organized, which NLRA § 9(c)(5) unequivocally prohibits. Nestle, 821 F.3d at 495 (citation omitted). But, here, the NLRB rubber-stamped the Union's proffered unit by engaging in a callow community of interest analysis. It then improperly forced Macy's to satisfy an overwhelming community of interest standard. Thus, the NLRB gave excessive deference to the composition of the requested unit and arbitrarily disregarded the collective bargaining interests of other salesforce employees to be in the same unit, "`effectively accord[ing] controlling weight to the extent of union organization.'" See id. at 499 (citation omitted); see also, e.g., Blue Man Vegas, 529 F.3d at 425-26; Lundy Packing Co., 68 F.3d at 1580, 1580-81.

B.

The NLRB not only abused its discretion and violated the NLRA as noted, but it also inadequately explained the reasons for its decision, thereby disregarding our circuit precedent and preventing proper judicial review.

1.

While "a bargaining unit designation by the [NLRB] is not lightly to be overturned," "it was manifestly not the congressional intent that appellate scrutiny of [NLRB] decisions be relegated to a formalistic ritual of stamping an appellate imprimatur on administrative determinations without having undertaken a careful examination of the basis of the [NLRB]'s action." Amalgamated Clothing Workers of Am. v. NLRB, 491 F.2d 595, 597 (5th Cir. 1974). Rather, courts must carefully review the record "to determine whether the [NLRB]'s decision is a rational one supported by the evidence." Id. This "translates into a duty by the [NLRB] ... to articulate `substantial reasons' for its unit determinations." Id. (citation omitted); see also Metro. Life, 380 U.S. at 443, 85 S.Ct. 1061.

To satisfy this requirement, the NLRB must "do more than simply tally the factors on either side of a proposition." Purnell's Pride, 609 F.2d at 1156.7 Because "[t]he crucial consideration is the weight or significance ... of factors relevant to a particular case," the NLRB "must assign a relative weight to each of the competing factors it considers" in order "to permit proper judicial review." Id. Thus, "unit determination[s] will be upheld only if the [NLRB] has indicated clearly how the facts of the case, analyzed in light of the policies underlying the community of interest test, support its appraisal of the significance of each factor." Id. at 1156-57 (citing Metro. Life, 380 U.S. at 442-43, 85 S.Ct. 1061 (remanding a unit determination case to the NLRB because its "lack of articulated reasons for the decisions in and distinctions among [unit determination] cases" frustrated judicial review)).

2.

Here, the NLRB has determined that Macy's' cosmetics and fragrances employees share a community of interest using a remarkably similar analysis to one this Court rejected in Purnell's Pride. In Purnell's Pride, the NLRB8: (1) made findings of fact; (2) discussed the traditional community of interest factors; (3) stated the employer's objections; (4) addressed them by applying the facts to the factors and citing four times to precedent; (5) "concluded that evidence bearing on [some factors] supported approval of the proposed unit while evidence [on other factors] militated against the proposed unit"; and (6) found that the evidence supported approval of the proposed unit. 609 F.2d at 1159-60. This Court held that the analysis contained a crucial flaw: the NLRB "d[id] not adequately explain ... the weight ... assigned to each individual factor." Id. at 1160. The reasons supporting the NLRB's decision were therefore "not sufficiently articulated to permit proper judicial review." Id. at 1161-62.

Here, like the decision we overruled in Purnell's Pride, the NLRB: (1) made findings of fact; (2) discussed the traditional community of interest factors; (3) applied the facts to the factors; (4) stated the employer's objections; (5) addressed them by applying the facts to the factors and citing to four cases; (6) concluded that "differences among the petitioned-for employees... are insignificant compared to the strong evidence of community of interest that they share"; and (7) found that the evidence supported approval of the proposed unit. Macy's & Local 1445, 361 N.L.R.B. 4, *10-11. But, as in Purnell's Pride, the NLRB did not address the weight it assigned to each competing factor.

3.

The panel fails to acknowledge that, just as in Purnell's Pride, the NLRB committed a "fatal" error by not weighing the community of interest factors and explaining why the differences between the cosmetics and fragrances employees and other selling employees outweighed the similarities. Macy's, 824 F.3d at 565. The panel summarily dismissed Macy's argument in three sentences:

Macy's, 824 F.3d at 565-66 (citations omitted).

Respectfully, the panel's analysis is obviously flawed. First, as discussed above, in Purnell's Pride, the NLRB patently did not, as the panel asserts, simply list the factors that guide the unit determination. Second, in Purnell's Pride, we required more of the NLRB than, as the panel asserts, identifying some factors that could weigh against a petitioned-for unit because the NLRB did precisely that in Purnell's Pride. Instead, Purnell's Pride required the NLRB to assign a weight to each community of interest factor and weigh the factors. 609 F.2d at 1156-57. Third, the NLRB neither weighed the community of interest factors here nor explained why the differences between the cosmetics and fragrances employees and other selling employees outweighed the similarities. Consequently, "the reasons supporting the Decision ... [were] not sufficiently articulated to permit proper judicial review." See id. at 1161.

In this light, the panel's decision to nevertheless uphold the NLRB's decision contravenes circuit precedent. The next panel that addresses the question of whether the NLRB, or another agency, has sufficiently articulated the reasons for its decision may not be bound by the panel opinion in this case because Purnell's Pride predates the panel's decision and remains cognizable law in this circuit. Rios v. City of Del Rio, Tex., 444 F.3d 417, 425 n.8 (5th Cir. 2006) ("The rule in this circuit is that where two previous holdings or lines of precedent conflict the earlier opinion controls and is the binding precedent in this circuit (absent an intervening holding to the contrary by the Supreme Court or this court en banc)."). In short, when we fail to follow clearly applicable precedent, we send confusing signals to the litigants and to the district courts. They deserve better.

II.

For these reasons, I respectfully dissent from the failure of the Court to vote this case en banc.