1 The National Labor Relations Board is authorized under § 9(c) of the National Labor Relations Act, 29 U.S.C. § 159(c) (1976), to certify representation elections. From this basis, the Board has developed its regulation of elections in the form of case law. See, e.g., General Shoe Corp., 77 N.L.R.B. 124, 126, 21 L.R.R.M. 1337, 1340 (1948) (setting of speech made to employees, who had been called to company president's office, could “prevent a free and untrammeled choice by the employees”). A party seeking to have an election overturned may file an objection with the Board, initiating an “R” (representation) case. In addition, certain kinds of election speech may constitute unfair labor practices, violating § 8 of the Act, 29 U.S.C. § 158(c) (1976). See, e.g., NLRB v. Gissel Packing Co., 395 U.S. 575 (1969) (offending employer ordered to bargain with union, despite union's having lost the election). The two kinds of cases become intertwined where an unfair labor practice complainant seeking to have an election overturned files an objection for that purpose and files an unfair labor practice charge as well. See, e.g., Sinclair Co. v. NLRB, decided with Gissel Packing Co., 395 U.S. 575, 589 (1969) (affirming cease-and-desist order and order that company bargain with union if union requests it, and overturning the election where president's numerous communications to employees had stressed that plant closings and job losses might follow upon unionization). This paper investigates the consequences of misrepresentation and coercive speech, and in doing so refers to representation cases and unfair labor practice cases without distinction. See generally Getman, Julius G., Goldberg, Stephen B., & Herman, Jeanne B., NLRB Regulation of Campaign Tactics: The Behavioral Assumptions on Which the Board Regulates [: Part I], 27 Stan. L. Rev. 1465, 1465–67 (1975) [hereinafter cited as Behavioral Assumptions] (reviewing Board rules and their development); Julius G. Getman, Labor Relations: Law, Practice and Policy 56–64 (Mineola, N.Y.: Foundation Press, 1978) (discussing Board's role in protecting right to organize). A substantial majority of objections to elections are filed by unions, objecting to employer speech or conduct. See, e.g., Samoff, Bernard, NLRB Elections: Uncertainty and Certainty, 117 U. Pa. L. Rev. 228, 253 app. table II (1968) (unions' share of all objections filed in a year fell in 1967 to low of 70.5 percent, including 3.4 percent in which employer also filed). This paper focuses on such employer speech, and does not address similar questions with respect to unions, individual employees, or other speakers, despite their potential importance in the larger context of election speech. See, e.g., NLRB v. Urban Tel. Corp., 499 F.2d 239, 241–43 (7th Cir. 1974) (failure of union to disavow an employee's threats of violence toward other employees who voted against union was basis for upsetting election, irrespective of whether employee was agent of union).Google Scholar

2 An employee's decision to vote for a union reflects many considerations in addition to that of coercive speech or conduct by the employer. These additional factors include permissible noncoercive speech—what the Board could see as “legitimate” information—as well as a variety of economic factors and other considerations. See Farber, Henry S. & Saks, Daniel H., Why Workers Want Unions: The Role of Relative Wages and Job Characteristics, 88 J. Pol. Econ. 349 (1980) (the lower an employee's wage relative to firm's average, the greater the likelihood of voting for union; other important determinants of vote include job security, nonfinancial aspects of job, and race and age of employee) (empirical research using data collected by Julius Getman, Stephen Goldberg, and Jeanne Herman); Flanagan, Robert J., The Behavioral Foundations of Union Election Regulation, 28 Stan. L. Rev. 1195, 1198–1203 (1976) (linking voting behavior to job satisfaction).Google Scholar

3 See, e.g., NLRB v. Exchange Parts Co., 375 U.S. 405, 409 (1964) (“well-timed increases in benefits … [suggest] a fist inside the velvet glove”). Moreover, the Board has at times been quick to discern employer coercion. See, e.g., Dal-Tex Optical Co., 137 N.L.R.B. 1782, 1785, 1787, 50 L.R.R.M. 1489, 1491 (1962) (interpreting employer statements “calculated to convey … the danger and futility of their designating the Union” as containing a threat not to bargain in good faith with union if the union should win the election). See Bok, Derek C., The Regulation of Campaign Tactics in Representation Elections Under the National Labor Relations Act, 78 Harv. L. Rev. 38, 74–116 (1964) (discussing various forms of prohibited employer speech, including misrepresentations, interrogation of employees, and conferral or promise of benefits); id. at 124–25 (discriminatory discharges); Getman, Goldberg, & Herman, Behavioral Assumptions, supra note 1, at 1475–78, 1488–90 (1975) (discussing threats of reprisal, promises or conferral of benefits, and possible intimidation effects). In determining whether to set aside an election, the Board considers whether a timely response by the potentially harmed party could have neutralized the impact of misrepresentations made by another party (if indeed there is sufficient time for response). Id. at 1475 n.60. See, e.g., Hollywood Ceramics Co., 140 N.L.R.B. 221, 224, 51 L.R.R.M. 1600, 1601 (1962) (last-minute handbill was distributed by union “at a time which prevents … an effective reply”); Luxuray of N.Y. v. NLRB, 447 F.2d 112, 115, 116–117 (2d Cir. 1971) (company's showing of film “And Women Must Weep,” described as “a one-sided brief against unionism,” did not violate Act because union had time to respond and did respond by showing its own film).Google Scholar

4 See Getman, Goldberg, & Herman, Behavioral Assumptions, supra note 1, at 1490 (employer promise or grant of benefits can backfire, for employees may believe benefits will increase further yet if there is a union); Getman, Julius G. & Goldberg, Stephen B., The Myth of Labor Board Expertise, 39 U. Chi. L. Rev. 681, 690 (1972) [hereinafter cited as Myth] (“Acts of reprisal … may have the effect of solidifying union support by showing the need for a union to protect employee interests, or simply by making the employees angry with the employer”). Accord, Morton Gitelman, Unionization Attempts in Small Enterprises: A Guide for Employers 70 (Mundelein, Ill.: Callaghan & Co., 1963) (“Vague general statements such as, ‘Unions are bad,’ or ‘We don't like unions,’ are worse than useless; they tend to drive the uncommitted employees into the union camp”). Alternatively, coercive speech or behavior may have occurred even before the election campaign, as a regular part of the workplace atmosphere; in that situation employee dissatisfaction may already be great, constituting a latent backlash effect. Such dissatisfaction has been found to be associated with a lower procompany vote. Getman, Julius G. & Goldberg, Stephen B., The Behavioral Assumptions Underlying NLRB Regulation of Campaign Misrepresentations: An Empirical Evaluation: Part II, 28 Stan. L. Rev. 263, 270 (1976) [hereinafter cited as Empirical Evaluation] (finding a generally negative relationship between employees' job satisfaction and proportion of employees voting pro-union). The research on which these articles are based is presented more fully in Julius G. Getman, Stephen B. Goldberg, & Jeanne B. Herman, Union Representation Elections: Law and Reality (New York: Russell Sage Foundation, 1976). For reviews of the research, in addition to the cases and articles discussed elsewhere in this paper, see Flanagan, supra note 2; Goetz, Raymond & Wike, Edward L., Book Review, 25 U. Kan. L. Rev. 375 (1977); Peck, Cornelius J., Book Review, 53 Wash. L. Rev. 197 (1977); and Shapiro, David L., Why Do Voters Vote? 86 Yale L.J. 1532 (1977). For the authors' rejoinder, see Goldberg, Stephen B., Getman, Julius G. & Brett, Jeanne M., Union Representation Elections: Law and Reality: The Authors Respond to the Critics, 79 Mich. L. Rev. 564 (1981) [hereinafter cited as Authors Respond], which also discusses many reviews of the book.Google Scholar

5 Here I use the Supreme Court's term “coercive speech” (NLRB v. Gissel Packing Co., 395 U.S. 575, 620 (1969)) to describe several kinds of speech or behavior, including reprisals or threats of reprisal and the promise or conferral of benefits. The prohibited speech and behavior includes misrepresentations but may also embrace true statements, such as well-founded predictions of plant closings. Indeed, the Board has been rather generous in construing employer statements to constitute threats of reprisal. See, e.g., Rein Co., 111 N.L.R.B. 537, 538–39, 35 L.R.R.M. 1517, 1518 (1955) (legally accurate statements threatening the loss of current benefits in retaliation for unionization); Thomas Products Co., 167 N.L.R.B. 732, 733, 66 L.R.R.M. 1147, 1148 (1967) (employer's repeated reference to strikes); Singer Co., 199 N.L.R.B. 1195 (1972), enforced, 480 F.2d 269 (1973) (employer's statements that major reason for plant's current location was low labor costs, and observation that unionization at other plants was followed by transferral of work and loss of jobs). However, the environment and relationships in which the employer speech occurs may affect the meaning and coerciveness of the speech. See Archibald Cox, Law and the National Labor Policy 44 (Los Angeles, Cal.: Institute of Industrial Relations, University of California, 1960) (“the impact of words is a function of the time and place … and past relations …. Words which may only antagonize a hard-bitten truck driver in Detroit may seriously intimidate a rural textile hand in a company village where the mill owners dominate every aspect of life. The dictionary meaning is irrelevant; the question is, what did the speaker intend and the listener understand”). The terms “number of pro-union votes” and “likelihood of pro-union vote” are used here to analyze employee decisions in the context of representation elections. Where employees are not covered by the Act, the same analysis would still apply; in that event, however, the voting terms would be replaced by terms indicating level or intensity of other pro-union activity, such as the likelihood of a strike for recognition, or the expected number of strikers.Google Scholar

6 There are, of course, other public interests to be taken into account in addition to those manifested by the union and the employer. See text at notes 39–40 and 71–75 infra.Google Scholar

7 The National Labor Relations Act of 1935, ch. 372, 49 Stat. 449 (codified as amended at 29 U.S.C. §§ 141–188 (1976 & Supp. Ill 1979)), set out the following policy:Google Scholar

to eliminate [and mitigate] … substantial obstructions to the free flow of commerce … by encouraging the practice and procedure of collective bargaining and by protecting the exercise by workers of full freedom of association, self-organization, and designation of representatives of their own choosing, for the purpose of negotiating the terms and conditions of their employment or other mutual aid or protection.Google Scholar

29. U.S.C. § 151 (1976). This list of goals was augmented by those of the Taft-Hartley amendments (Labor-Management Relations Act) of 1947, ch. 120, 61 Stat. 136 (codified as amended in scattered sections at 29 U.S.C. §§ 141(b)-188 (1976 & Supp. III 1979)). See, e.g., § 1(b), at 29 U.S.C. § 141(b) (adding goals of maintenance of “public health, safety, or interest” and protection of “rights of individual employees in their relations with labor organizations”).Google Scholar

8 These substantive and administrative goals come together in a number of ways, particularly in the Board's fear that delay in resolving representation questions may reduce incentives to obey election standards or enhance the attractiveness of various delaying tactics. See General Knit Co. of Calif., 239 N.L.R.B. 619, 622–23, 99 L.R.R.M. 1687, 1690–91 (1978) (criticism of Hollywood Ceramics rule included lack of predictability with respect to whether election would be set aside, and also the administrative unworkability of rule, which could be used as “a vehicle for delay”); Modine Mfg. Co., 203 N.L.R.B. 527, 529–30, 83 L.R.R.M. 1133, 1135–36 (1973), enforced, 500 F.2d 914 (8th Cir. 1974) (reviewing criticisms of Board's election regulation and delay). See Bok, supra note 3, at 59–65, 66 (discussing efficiency considerations in administration, and noting interrelations between these considerations and substantive decision making by Board). Cf. Getman, Goldberg, & Herman, Behavioral Assumptions, supra note 1, at 1492 (linking recent increases in Board's caseload, due in part to election campaign regulation, to lower quality of decisions).Google Scholar

9 See, e.g., NLRB v. Virginia Elec. & Power Co., 314 U.S. 469 (1941) (revising and remanding to Board, Court held that speech which is not by its own terms coercive does not violate Act unless part of a course of conduct which is coercive); NLRB v. Golub Corp., 388 F.2d 921, 926 (2d Cir. 1967) (Board's “halting response” to Court's decision in Virginia Electric Power Co. led to congressional adoption of § 8(c) of Act, providing increased speech protection). But cf. NLRB v. Gissel Packing Co., 395 U.S. 575, 617 (1969) (§ 8(c) “merely implements the First Amendment”). See NLRB v. Golub Corp., 388 F.2d 921, 926–28 (2d Cir. 1967) (historical review of Board decisions involving conflicts between First Amendment and other labor policy goals).Google Scholar

10 77 N.L.R.B. 124, 127, 21 L.R.R.M. 1337, 1341 (1948).Google Scholar

11 See, e.g., Bausch & Lomb, Inc. v. NLRB, 451 F.2d 873, 878 (2d Cir. 1971).Google Scholar

12 77 N.L.R.B. at 127, 21 L.R.R.M. at 1341.Google Scholar

13 Id. (footnote, case citation omitted).Google Scholar

14 140 N.L.R.B. 221, 224, 51 L.R.R.M. 1600, 1601 (1962) (footnote omitted), overruled, Shopping Kart Food Mkt., 228 N.L.R.B. 1311, 94 L.R.R.M. 1705 (1977), overruled, General Knit Co. of Calif., 239 N.L.R.B. 619, 620, 99 L.R.R.M. 1687 (1978), overruled, Midland Nat'l Life Ins. Co., ___N.L.R.B___, 110 L.R.R.M. 1489 (1982).Google Scholar

15 140 N.L.R.B. at 224, 51 L.R.R.M. at 1602.Google Scholar

16 See Shopping Kart Food Mkt., 228 N.L.R.B. 1311, 1314, 94 L.R.R.M. 1705, 1708 (1977) (Murphy, Chairman, concurring) (ruling of Hollywood Ceramics“has been so expanded and misapplied as to have extended far from the original intent of the Board”). See also General Knit Co. of Calif., 239 N.L.R.B. 619, 622–23, 99 L.R.R.M. 1687, 1690–91 (1978) (discussing criticism of Hollywood Ceramics rule).Google Scholar

17 See note 5 supra. The NLRB's decisions in this area have been among its more controversial ones, and at times, its more short-lived ones. See, e.g., NLRB v. Golub Corp., 388 F.2d 921, 929 (2d Cir. 1967) (denying enforcement of Board's order to bargain because, “While these fears [of employer's future inability to extend benefits to some employees] may have been unwarranted, they were not shown to have so far transcended the bounds of reason as to justify the Board in finding them to be disguised threats of reprisal”). See Getman, Goldberg, & Herman, Behavioral Assumptions, supra note 1, at 1484–86, 1485 (tendency of courts to set aside Board decisions “has been particularly pronounced in cases involving alleged union misrepresentations”); General Knit Co. of Calif., 239 N.L.R.B. 619, 626, 99 L.R.R.M. 1687, 1693 (1978) (Penello, dissenting) (in misrepresentation cases, Board loses before courts 50 percent of the time); Midland Nat'l Life Ins. Co., ___N.L.R.B___, 110 L.R.R.M. 1489, 1493 (1982), overruling General Knit, supra (Shopping Kart rule reduces the possibility of disagreement between the NLRB and the federal courts of appeal).Google Scholar

18 See, e.g., Modine Mfg. Co., 203 N.L.R.B. 527, 530 (footnote omitted), 83 L.R.R.M. 1133, 1136 (1973) (“inherent dangers make it in some degree tempting to abandon the approach taken in Hollywood Ceramics, but this we are not prepared to do”), enforced, 500 F.2d 914 (8th Cir. 1974).Google Scholar

19 228 N.L.R.B. 1311, 94 L.R.R.M. 1705 (1977).Google Scholar

20 Id. at 1313, 94 L.R.R.M. at 1707–8, citing Getman & Goldberg, Empirical Evaluation, supra note 4. But cf. Goldberg, Getman, & Brett, Authors Respond, supra note 4, at 578 (“If our study has given the Board food for thought, there is no evidence of digestion”).Google Scholar

21 228 N.L.R.B. at 1313, 94 L.R.R.M. at 1708.Google Scholar

22 General Knit Co. of Calif., 239 N.L.R.B. 619, 620, 623, 99 L.R.R.M. 1687, 1688 (1978).Google Scholar

23 Id. at 621–22, 99 L.R.R.M. at 1689–90.Google Scholar

24 Id. at 623, 99 L.R.R.M. at 1691.Google Scholar

25 Id. at 622–23, 99 L.R.R.M. at 1690–91.Google Scholar

26 ___N.L.R.B.___, 110 L.R.R.M. 1489 (1982).CrossRefGoogle Scholar

27 Id. at 1490, 1492, 1493–94.Google Scholar

28 Virginia State Bd. of Pharmacy v. Virginia Citizens Consumer Council, 425 U.S. 748, 779 n.4 (1976) (Stewart, J., concurring).Google Scholar

29 See, e.g., Samoff, supra note 1, at 245 (the Board's “commendable desire to preserve the un-corruptability of the election process has led it to pursue, indefatigably and hopelessly, the goal first enunciated in General Shoe”) (quoting Benjamin Aaron; emphasis added by Samoff). See also Raskin, A. H., Deregulation of Union Campaigns: Restoring the First Amendment Balance, 28 Stan. L. Rev. 1175 (1976).Google Scholar

30 Browne, Harry L. & Sachs, Howard F., The Suppression of Employer Free Speech—A New Ban on “Conscious Overstatements” and a Caveat Against “Brinkmanship,” 15 Vill. L. Rev. 588, 610 (1970) (Court's decision in Gissel Packing Co. is inconsistent with its concern for “chilling effect” on expression in cases outside labor field). Id. at 608–11.Google Scholar

31 Miller, Edward B., The Getman, Goldberg and Herman Questions, 28 Stan. L. Rev. 1163, 1170 (1976) (Board's standards are successful in protecting employees from “truly effective illegal speech,” despite less than total enforcement, because Board has “drawn its lines” to be more than minimally protective). From the vantage point of conceptualizing speech as the result of a single process in which production of false speech is inextricably involved in the production of protected speech, the NLRB's standards appear quite speech restrictive. From this viewpoint, any regulation of speech necessarily deters both the speech that is the target of regulation (e.g., misrepresentations) and protected speech. Cf. Gissel Packing Co., 395 U.S. 575, 620 (1969):. [A]n employer, who has control over that [employer-employee] relationship and therefore knows it best, cannot be heard to complain that he is without an adequate guide for his behavior. He can easily make his views known without engaging in “‘brinkmanship’” when it becomes all too easy to “overstep and tumble [over] the brink,”Wausau Steel Corp. v. NLRB, 377 F.2d 369, 372 (C. A. 7th Cir. 1967). At the least he can avoid coercive speech simply by avoiding conscious overstatements he has reason to believe will mislead his employees, [last brackets in original; “brinksmanship” in Wausau].Google Scholar

32 Getman, Goldberg, & Herman, Behavioral Assumptions, supra note 1, at 1467.Google Scholar

33 33 NLRB Ann. Rep. 1968, at 60 (1969).Google Scholar

34 Getman & Goldberg, Empirical Evaluation, supra note 4, at 282–84.Google Scholar

35 The level N, of pro-union support is intended to represent “the uninhibited desires of the employees,” manifested in a “free and untrammeled choice,” to use the Board's terms. General Shoe Corp., 77 N.L.R.B. 124, 127, 126, 21 L.R.R.M. 1337, 1341, 1340 (1948).Google Scholar

36 For an individual, fear and anger may be viewed as two manifestations of a single response, differing from one another primarily in degree rather than in kind. See Walter B. Cannon, Bodily Changes in Pain, Hunger, Fear and Rage 342–43 (2d ed. Boston: Charles T. Branford Co., 1929) (noting visceral changes common to both fear and rage and characterizing the few differences as “not noteworthy”); John Dollard & Neal E. Miller, Personality and Psychotherapy: An Analysis 82 (New York: McGraw-Hill Book Co., 1950) (discussing similarity of peripheral stimulus-producing responses involved in anger and fear); Jeffrey A. Gray, The Psychology of Fear and Stress 192–95 (New York: McGraw-Hill Book Co., 1971) (“a single fight/flight mechanism … [puts] the organism into a state of readiness for either fight or flight”). Id. at 210. Cf. Hirshleifer, Jack, Economics from a Biological Viewpoint, 20 J. L. & Econ. 1, 31–35, 39–43 (1977) (mathematical representation of conflict/competition versus complementarity relationships among species, and response functions in predator-prey interaction); Judith A. Lachman, A Theory of Interpersonal Conflict 7–18 (University of Michigan Institute of Public Policy Studies, Discussion Paper No. 117, Feb. 1978) (two-party model of conflict and conditions for stability of equilibrium). A similar pattern of response has been found in the context of group behavior for the relationship between level of civil strife and the coercive force used by authorities, as depicted in fig. A. Ted § Robert Gurr, A Comparative Study of Civil Strife, in 2 Hugh Davis Graham & Ted Robert Gurr, eds., Violence in America: Historical and Comparative Perspectives: A Report to the National Commission on the Causes and Prevention of Violence 443, 475 fig. 17–7 (Washington, D.C.: National Commission on the Causes and Prevention of Violence, 1969) (cross-national data on countries not facing foreign threats). See also Lewis F. Richardson, Arms and Insecurity: A Mathematical Study of the Causes and Origins of War 52–65, 253–61 (Pittsburgh, Pa.: Boxwood Press & Chicago: Quadrangle Books, 1960) (model of arms races); Kenneth E. Boulding, Conflict and Defense: A General Theory 19–40, 58–79, 105–22 (New York: Harper & Row, 1962) (dynamic conflict model). Although it is possible that an intimidation stage precedes the backlash stage (see, e.g., Getman, Goldberg, & Herman, Behavioral Assumptions, supra note 1, at 1489 & 1489 n.152), the behavioral literature appears to afford stronger support for the form used in the text. See, e.g., Aubrey J. Yates, Frustration and Conflict 70 (New York: John Wiley & Sons, 1962) (observing initial aggressive response to external instigation).Google Scholar

37 General Shoe Corp., 77 N.L.R.B. 124, 127 (1948).Google Scholar

38 Although low levels of employer coercion will work to its disadvantage, the employer should always prefer a policy allowing greater discretion to one allowing less, since the former includes the best choices available under both, and additional options as well. Under some conditions, however, it is conceivable that an employer would prefer a more speech-restrictive standard. This would occur, for example, if restrictions on employer speech were tied to restrictions on union speech, and if the employer benefited from union restriction more than it was disadvantaged by the restrictions on its own speech.Google Scholar

39 In other words, the more generous the standard, the lower the likelihood of its violation by an individual employer or union, and the fewer the violations, taken in the aggregate. The relaxation in standards pronounced in Shopping Kart Food Mkt., 228 N.L.R.B. 1311, 94 L.R.R.M. 1705 (1977) did result in fewer misrepresentation objections being filed with the Board; the number declined from 307 in the year before Shopping Kart to 180 in the year after, constituting “a stunning tribute to the effectiveness of that decision.” General Knit Co. of Calif., 239 N.L.R.B. 619, 629, 99 L.R.R.M. 1687, 1696 (Penello, dissenting). The dissenters in Shopping Kart predicted, however, that the majority's standard eventually would increase, rather than decrease, the number of misrepresentation charges, because misrepresentation would “tend to drive out the responsible statement.” 228 N.L.R.B. at 1316, 94 L.R.R.M. at 1710 (Fanning & Jenkins, dissenting in part).Google Scholar

40 See General Knit Co. of Calif., 239 N.L.R.B. 619, 621 & 621 n.13, 99 L.R.R.M. 1687, 1689 & 1689 n. 13 (1978) (“access to the Board for review … legitimizes the integrity of the electoral process,” and “a rule [such as Shopping Kart] which merely eliminates a certain classification of cases, at the expense of an important principle, is not a success”); Shopping Kart Food Mkt., 228 N.L.R.B. 1311, 1316, 94 L.R.R.M. 1705, 1708 (1977) (Fanning & Jenkins, dissenting in part) (Board consideration of misrepresentation cases is “an excellent investment in maintaining our election standards”).Google Scholar

41 Getman, Goldberg, & Herman, Union Representation Elections, supra note 4, at 119 (“campaigning viewed by the Board as potentially coercive, and thus unlawful, is not associated with a loss of union support”).Google Scholar

42 This result could also occur if the union's losses due to intimidation were always exactly equal to its gains attributable to the backlash effect. In those circumstances the line intercepts the vertical axis at level N_t. However, it would be only by coincidence that this precise offset would occur and that the average gain to the union would be equal to zero.Google Scholar

43 This diversity of empirical results possible from a single set of observations stems from differences in functional forms used in the estimation—in this example, a linear form versus a nonlinear form. See Jan Kmenta, Elements of Econometrics 399–400 (New York: Macmillan Co., 1971) (discussing use of linear functional form in estimation—when nonlinear one is appropriate—as special case of omission of relevant explanatory variable).Google Scholar

44 Getman, Goldberg, & Herman, Union Representation Elections, supra note 4, at 115 table 5–3. One should note that the definition of a “violation” affects the relationship observed between severity of violation and level of pro-union support. If, for example, the standard in force (or the lowest standard considered a violation) is located at the peak of the voter response function, or to the right of the peak, all observations of violations will necessarily fall to the right of that point, and hence within the range of the intimidation effect. Consequently, a study of voter response to employer violations of standards—as contrasted with a study of response to coercive speech more generally—will not provide information about the backlash range of the response function.Google Scholar

45 Where a majority of employees have signed authorization cards that unambiguously express support for the union, the cards may be used to determine the union's majority status. See, e.g., Sinclair Co. v. NLRB, decided with Gissel Packing Co., 395 U.S. 575, 589 (1969).Google Scholar

46 Getman, Goldberg, & Herman, Union Representation Elections, supra note 4, at 113 & n.2.Google Scholar

47 See Miller, supra note 31, at 1164 (union's support lower by 4 to 35 percent at time of vote, as compared with time of card signing) (Miller's restatement of Getman, Goldberg, and Herman results). However, it is possible that this loss of union support is attributable to attrition during this time interval rather than to a difference in the degree of commitment reflected by card signing and pro-union voting. See text at notes 60–64 infra.Google Scholar

48 See Getman & Goldberg, Myth, supra note 4, at 689 (Labor Board rules “often confuse moral outrage with impact determination”).Google Scholar

49 Getman, Goldberg, & Herman, Union Representation Elections, supra note 4, at 114–15, 115 table 5–3 (among undecided voters, and among pro-union voters in elections not leading to bargaining orders, percentage voting pro-company increased with degree of employer coercion). One might consider elections in which bargaining orders were issued as representing a variety of levels of coercion, rather than a single severe level, since “moral outrage” violations—which may or may not have a significant effect on vote outcomes—are more likely to count for a large share of cases within this category. See text at notes 45–48 supra. Consequently, it could be argued that bargaining order elections are scattered across the full range of coercion levels, rather than representing only elections in which the greatest level of coercion occurred. If one adopts this viewpoint, disregards the bargaining order elections, and focuses instead on the other two categories, the result is that higher levels of coercion are linked to greater pro-company voting within all groups delineated by Getman, Goldberg, and Herman.Google Scholar

50 Les Aspin, Legal Remedies Under the NLRA, Remedies Under 8(a)(3), in Industrial Relations Research Association Proceedings of the Twenty-Third Annual Winter Meeting, December 1970, Detroit, 264, 265–70 (Madison, Wis.: Industrial Relations Research Association, 1971).Google Scholar

51 Id. at 265–70. Cf. A Bill to Amend the National Labor Relations Act in Order to Increase Effectiveness of the Remedies: Hearings on H.R. 11725 Before the Special Subcomm. on Labor of the House Comm. on Education and Labor, 90th Cong., 1st Sess., 12, 15 (1967) [hereinafter cited as 1967 Hearings] (statement of William Kircher, Director of Organization, AFL-CIO) (although unions won in 52 percent of elections studied, unions won only 43 percent of the time in elections preceded by firings, layoffs, or demotions).Google Scholar

52 Aspin, supra note 50, at 265–70.Google Scholar

53 Leslie Aspin, A Study of Reinstatement Under the National Labor Relations Act 23–24 (Ph.D. diss., Massachusetts Institute of Technology, 1966).Google Scholar

54 See Getman & Goldberg, Empirical Evaluation, supra note 4, at 281–84. The finding that a large proportion of employees vote consistently with their precampaign attitude or intent does not imply that the campaign is uninfluential. Indeed, rather than seeking to gain the votes of undecided employees, the campaign's “more important function may be to insure that the party does not lose the votes it has.”Truesdale, John C., From General Shoe to General Knit: A Return to Hollywood Ceramics, 30 Lab. L.J. 67, 73 (1979). See, e.g., Getman & Goldberg, supra, at 281 table 8 (union lost about 11 percent more votes than company did because of vote switching during campaign period).Google Scholar

55 Getman & Goldberg, Empirical Evaluation, supra note 4, at 273 n.38. This assumption is adopted to simplify the analysis but does gloss over a significant question remaining in the authors' research: After the vote, an individual will have voted for one party or another (or not have voted); and even though the ex ante probability estimate is correct, the vote of an individual need not go in the direction of the ex ante probability estimate. For example, an employee who has actually voted pro-company may have had a .7 probability of voting for the union. The .7 probability estimate means that in 1,000 elections this employee would vote for the union an expected 700 times but would vote for the company an expected 300 times. If the particular election observed in the data is one of the 300 in which the employee would cast a pro-company vote, the original .7 probability estimate is still good: In that event, the pro-company vote does not signify a change in attitude but instead signifies simply the less probable manifestation of an attitude held consistently over time. Of course, attitude changes may occur, too, as when an employee goes from a .7 probability to a 4 probability of voting pro-company. It will be difficult, if not impossible, to distinguish such attitude changes from the votes that were simply unlikely (but possible) in the first place. For this reason, the measure of switchers may tend to overstate the number of employees whose attitude has indeed changed. However, the switcher count may also understate the extent of attitude change because it will fail to include some true switchers (e.g., the .7 probability person who would vote union then undergoes a change of attitude to .4 probability but nevertheless votes for the union). Without further investigation, one cannot know whether these counting errors offset one another and whether their net effect leads to over- or underestimation of the number that switch voting preference. Such switching behavior may, however, be critical to the outcome of the election. See, e.g., NLRB v. Savair Mfg. Co., 414 U.S. 270, 277–78 (1973) (a single vote change would have made election outcome different); Getman & Goldberg, Empirical Evaluation, supra note 4, at 279–81 (despite fact that 81 percent of employees voted in accordance with precampaign intent, votes of remaining 6 percent who were undecided and 13 percent who switched from earlier intent determined outcome in 9 out of 31 elections studied).Google Scholar

56 As represented here, the response functions of pro-union and pro-company employees differ principally in their initial level of pro-union vote probability. This difference leads to the existence of two switchover points for pro-company employees but only one for pro-union employees. The pro-union employees have, effectively, made the first switch already, to the left of the first point on the graph.Google Scholar

57 See Getman & Goldberg, Empirical Evaluation, supra note 4, at 274 table 3, 282.Google Scholar

58 This would be true also if, for example, the group expressing pro-union sentiments included previously pro-company employees already exposed to some coercive speech or behavior between elections.Google Scholar

59 See note 3 supra.Google Scholar

60 General Knit Co. of Calif., 239 N.L.R.B. 619, 627, 99 L.R.R.M. 1687, 1694 (Penello, dissenting) (1978): “Through the medium of delay, even an adverse Board or court ruling is transformed into a victory for an employer intent on thwarting the purposes and policies of the Act.” See note 8 supra; Miller, supra note 31, at 1164 (union support fell by an average of 4 percent between time of card signing and time of vote).Google Scholar

61 See Samoff, supra note 1, at 237–38. This result may reflect in part the effects of delay: “Where the union's majority is thin, the time required to process objections may erode union support.”Id. at 237. Indeed, the employer's likelihood of success improves through each successive stage of the Board's processes, from investigation to hearing to Board decision, and finally, to court resolution. Id.Google Scholar

62 Pollitt, Daniel H., NLRB Re-run Elections: A Study, 41 N.C.L. Rev. 209, 212 (1963). In this study of rerun elections, Pollitt found that objecting unions won 30 percent of the time and the objecting companies won 36 percent of the time in their respective rerun elections. Id. at 212. The longer the time interval between the original and the rerun election, the lower the percentage of elections in which the rerun reversed the original result. See id. (relationship drawn from Pollitt's figures).Google Scholar

63 See Getman & Goldberg, Myth, supra note 4, at 691–94 (Pollitt study failed to control for variables that may be related to outcome of rerun election, such as characteristics of employee unit and type of campaign conducted). For example, Pollitt relates that in one situation the employer “pulled out all the stops,” engaging in five violations of the Act, winning the first election by 46 to 36, and then winning the second one 51 to 18 (Pollitt, supra note 62, at 214 & n.17). According to Getman and Goldberg (supra, at 693), Pollitt would attribute the union's resounding defeat in the second election to the lasting coercive effect of the employer's earlier activity (which had precipitated the rerun election). This account would fail, however, “to explain the much closer vote in the first election, when the employer's conduct was fresher in the minds of the employees” (id. at 693–94). Getman and Goldberg suggest instead that both votes reflected the true preferences of employees; and that the pro-union support in the first election had been augmented by a backlash reaction to the employer's behavior (id. at 694). I would suggest an alternative though not necessarily conflicting interpretation of Pollitt's numbers, which is that the second vote reflected the true preferences of employees, as modified by the effects of employer coercive speech—positive or negative—and as further affected by the delay-related erosion of pro-union support.Google Scholar

64 Samoff's research, supra note 1, makes no attempt to carry out any form of multivariate analysis, nor does it make any claim of doing so. Instead, the author draws principally upon figures drawn from NLRB Annual Reports (see id. at 253) and upon his experience working at the Board.Google Scholar

65 Gissel Packing Co., 395 U.S. 575, 620 (1969).Google Scholar

66 With respect to the martyr effect of a low number of discharges, see Bok, supra note 3, at 41: “Although this tactic can often frustrate a union drive, any experienced organizer knows that a discriminatory discharge may rally the voters against the employer instead of frightening them into submission. A few organizers have even provoked a discharge deliberately for this reason.”.Google Scholar

67 See Getman, supra note 1, at 50 (counsel for employers favor promise of benefits as campaign tactic that is more effective and less likely to provoke backlash than are reprisals or threats of reprisal). Cf. Getman, Goldberg, & Herman, Union Representation Elections, supra note 4, at 126 (“whenever an employer discharges union supporters during an organizing campaign, employees [especially those supporting the union] tend to view that discharge as having been motivated by the employer's anti-union sentiments”).Google Scholar

68 See 1967 Hearings, supra note 51, at 15 (statement of William Kircher) (in elections preceded by wage increases, unions won 37 percent of the time, as compared with 43 percent in elections where layoffs, firings, demotions, or employer surveillance preceded the election).Google Scholar

69 Cf. Getman, Goldberg, & Herman, Union Representation Elections, supra note 4, at 124 (in interviews with employee-voters, the “greater tendency of company voters to report benefits than reprisals may … be due to their predisposition to view the employer's campaign in a favorable light”).Google Scholar

70 Indeed, differing perceptions of the location of point S may account in part for the Board's successive battles over standards. The recent Midland decision (___N.L.R.B___, 110 L.R.R.M. 1489 (1982)), for example, may reflect a renewed perception by three of the Board members that S lies sufficiently far to the left, while the dissenters may see it as being still farther to the right on the graph.Google Scholar

71 See text at notes 14–25 supra.Google Scholar

72 See Getman, Goldberg, & Herman, Behavioral Assumptions, supra note 1, at 1469–73. Suppose, for example, that a company's becoming unionized would worsen the market conditions for the company's product, because the businesses that purchase from it are rabidly anti-union and have vowed to take their orders elsewhere. Prospective voters might wish to know such information, which affects their future livelihood; but employees, even if they have access to such information, may be reluctant to bear such bad tidings. If that information is not forthcoming from the employer it is unlikely to be available to voters at all. Moreover, if information in the hands of the employer is not made available to employees during the campaign, the cost to employees of obtaining such information may be prohibitive. This might be the case, for example, if employees wished to know the likelihood of a plant's being closed, in light of planned changes in product packaging; the cost of replicating management's marketing studies and sales projections could be substantial.Google Scholar

73 Id. at 1491.Google Scholar

74 See, e.g., Sewell Mfg. Co., 138 N.L.R.B. 66, 50 L.R.R.M. 1532 (1962) (employer campaign material involving inflammatory appeals to racial prejudice).Google Scholar

75 See Samoff, supra note 1, at 235: “In the Board's view, the true purpose of its elections is not necessarily to reach undisputed results, but rather to create an atmosphere in which the ‘surrounding conditions enable employees to register a free and untrammeled choice for or against a bargaining representative’” (quoting General Shoe Corp., 77 N.L.R.B. 124, 126, 21 L.R.R.M. 1337, 1340 (1948)).Google Scholar

76 Cf. The Air Strike Is Lost. Now What N.Y. Times, Aug. 30, 1981, IV, at 16, col. 1 (editorial):. By refusing to cave in to this illegal strike, the President has set a commendable precedent that is already paying dividends in relations with other unions. Though many postal workers were said to be unsatisfied with the terms recently negotiated by their union leaders, the contracts have been ratified. Labor oratory has shifted noticeably from militancy to conciliation.Google Scholar

77 See notes 3 & 5 supra.Google Scholar

78 See text at notes 57–58 supra.Google Scholar

79 See note 4 supra; text at note 34 supra.Google Scholar

80 Of course, in order for the self-policing mechanism to come into play, it is necessary that the company perceive the coercive speech as leading to a loss of company support. The process by which such a consequence is perceived by a company—or is not perceived—is beyond the scope of this paper.Google Scholar

81 The union's net gain of votes in this range is not inconsistent with the presence of an intimidation effect. In this range of the response function the intimidation effect and backlash effect work against each other, with the former offsetting the latter only partially.Google Scholar

82 Cf. Goldberg, Getman, & Brett, Authors Respond, supra note 4, at 590:Google Scholar

[T]he reforms that we propose [the elimination of all legal restrictions on speech, and other changes] would not lead to an increase in acts of reprisal, and we think it unlikely that they would lead to an increase in threats of reprisal. They might lead to an increase in promises or grants of benefit.Google Scholar