The Labor Board petitions for enforcement of its bargaining order against respondent, Winchell Processing Corporation and Winchell Donut House. Following a representation election, the Board certified two unions, Bakery Drivers’ Local 276 (Teamsters) and the Bakery and Confectionery Workers’ International, as the joint bargaining representative of Winchell’s employees. The employer corporation asserts that false and misleading statements in the unions’ campaign literature vitiated the election. We agree and deny enforcement of the order.

Winchell makes donut mixes and distributes them to franchised retail outlets. In 1969 the unions began organizing Winchell’s production employees, ware-housemen and delivery truck drivers. Twelve employees work in these job categories, four as food blenders and eight as drivers and warehousemen. The representation election was held on the morning of July 25, 1969. The unions won, by a seven-to-four vote, with one ballot disqualified.

The Bakery Drivers’ Local mailed two leaflets to the employees shortly before the election. One, dated July 22, reached the employees on July 23. The second, dated July 23, arrived at the employees’ homes on July 24. The company first learned of both circulars on July 24. The mailings contained charges of employer conduct prejudicial to the workers, assertions about the benefits available under the union’s master contract covering bakery drivers in Southern California, assurances about strikes, and an explanation of representation procedures. The timing of the distribution gave Winchell no opportunity to reply.

The leaflets presented false or misleading information in several areas relevant to a rational voter choice. Winchell filed timely objections with the Board, asking it to set aside the election. Although he found a number of ambiguities in the leaflets, the Regional Director decided that they likely had no influence on the election outcome.

The heat of a representation campaign inevitably generates some exaggeration by both sides. The Board in most cases leaves evaluation of election propaganda to the good sense of the employees, on the theory that they are competent to sift the truth from the parties’ divergent claims. It will invalidate an election only where a material fact has been misrepresented, the opponent had no opportunity to reply, and the resulting lop-sided presentation significantly impaired the election process. Stewart-Warner Corp., 102 N.L.R.B. 1153 , 1158 (1953).

Deciding whether a misleading statement has influenced an election can be a hazardous venture. One cannot look to the voters for assistance. Taking evidence from them would be cumbersome and expensive. Their testimony would likely reveal little. They may not know why they voted as they did. Their stories may be distorted by their loyalties or by fear of reprisal.

One must turn instead to generalizations about the kinds of misinformation likely to influence a rational voter. That inquiry is itself problematic. Conduct that should move an ideal voter may have no actual impact in any given election.

Another difficulty arises when one tries to determine whether a given statement is a material misrepresentation. Many assertions are literally true, yet may still be potentially misleading because important qualifying facts have been omitted. Ambiguities and half-truths of this kind appear to be common in representation campaigns. Reasonable minds can differ over the amount of background information a party should be required to disclose.

In view of the uncertainties inherent in these legal determinations, we must consider carefully what interests we protect by regulating election propaganda. The decisions speak of preventing improper influences from inhibiting the employees’ freedom of choice.

To make a rational choice about joining a union, an employee must examine his present working conditions, try to predict what increased benefits might result from collective bargaining, and weigh the potential improvements against the likelihood of losses from strikes or other breakdowns in his relationship with management. Not an easy calculus, the decision can become hopelessly confused when one introduces at the last minute a substantial quantity of false information about important relevant elements. The absence of misleading assertion does not guarantee a reasoned voter decision, but its presence may preclude any possibility of one.

Even if it made little sense to attempt to increase the likelihood of a rational employee choice, one could still justify some policing of last minute campaign oratory on the ground of fairness to the other party. The stakes are high for both company and union.

To lose is bad enough, but to lose because the other side deluged the workers with lies at the eleventh hour would be intolerable. Rewarding the better liar, when the other side has no opportunity to reply, is hardly the way to lessen industrial strife. Prohibiting significant last minute falsehoods helps assure each side that the outcome will be determined on grounds reasonably related to the merits of its position.

II.

The July 23 mailing, a single sheet, included the following comparison:

Assertions about union benefits must be held to a fairly close standard of accuracy since a union’s statements about its own contracts sound authoritative. Employees are likely to accept them uncritically.

(1) Wage Increases. For the 30-year span as a whole this statement is roughly correct, but wages had ceased to rise at that pace in recent years. From 1964 to 1969 wages under the master contract rose 18.7 percent while the cost of living climbed 19.6 percent. To a voter, the chief value of a wage/cost-of-living comparison is what it implies for the future. Failure to reveal the trend in recent years makes this statement highly misleading.

(2) Hours. Union members now work 45 hours in five days; they formerly worked 48 hours in six days. The statement exaggerates the improvement in the work week.

(3) Fringe Benefits. The contract does provide coverage in all areas listed, but the $80 figure is speculative.

(4) Vacations. The contract allows four, not five, weeks maximum vacation.

(5) Holidays. The actual increase is from five to eight holidays. No extra compensation is paid for any holiday that falls on a weekend, with one exception.

(6) Sick Leave. The employers pay one-half a day’s wages for each full day of illness. The total compensated sick leave available is five days a year, accumulative to 15 days. The master contract expresses this as “Ten % days” and “thirty % days.” By writing these figures as “10%” and “30%” the union could have deceived the employees into believing the contract guaranteed double the actual sick leave.

The union leaflet distorts six of the seven listed benefits. Perhaps none would be significant considered alone, but together they present an exaggerated picture of what an employee could expect to gain by joining the union.

III.

The July 22 sheet states that:

The unions cannot in fact allocate the employees among themselves. And upon employer request they must bargain together as a single unit, to form one contract covering all employees. This deception could be important to an employee who liked one union but not the other.

Another passage in the July 22 leaflet says:

The constitutions and by-laws of the two unions cast doubt upon the accuracy of this statement. The two have divergent strike sanction procedures. In neither does total authority rest with the bargaining unit. For example, under the Teamsters’ constitution a vote of the membership of the entire local, not just the Winchell bargaining unit, determines whether a strike will occur.

Winchell’s final complaints are directed at this passage in the July 22 leaflet:

Winchell pays overtime for holiday work. The employees undoubtedly knew this, so the falsehood is not serious. The fryer allowance charge is another matter. Winchell drivers at one time received a base salary from the company plus an allowance from the franchise outlets for cleaning their fryer machinery. In October 1967 they became hourly employees paid only by Winchell. They suffered no loss of pay; instead, their total compensation rose.

Four of the eight drivers worked for Winchell in 1967. They certainly could evaluate the reference in the union literature. Three of the other four drivers, and two of the food blenders, had been hired within a few months of the election. They knew nothing of the circumstances of the allowance’s removal. The timing of the mailing and employees’ work schedules prevented exchange of information between the informed workers and the newer ones. A sizable proportion of them, in consequence, had no means of evaluating the union accusations. These people could have believed, as the union intended, that the company had deprived them of something of value.

IV.

The Regional Director considered each of the foregoing ambiguities, falsehoods, and half-truths individually, without bothering to think about their cumulative effect. We owe considerable deference to his expertise; he passes on hundreds of complaints about election propaganda; we see very few. But the approach that he took in this case cannot do justice to his task. A series of small misstatements could easily have a greater impact on employees than would a single bald lie. One must look at the totality of circumstances preceding the election, not examine each statement as though the employees heard it alone. See S. H. Kress & Co. v. NLRB, 430 F. 2d 1234 (5th Cir. 1970); NLRB v. Trancoa Chemical Corp., 303 F.2d 456 (1st Cir. 1962).

The misinformation in these union leaflets, by its cumulative influence, deprived the Winchell employees of the truth in many areas that would be important to anyone faced with their decision. The union’s timing of its mailings foreclosed all opportunity for rebuttal, so we cannot dismiss these utterances as normal campaign hyperbole. We cannot sanction election conduct of this kind. The Labor Board’s bargaining order will not be enforced.