Opinion

The petitions for review filed by Adamek & Dessert, Inc. and the United Farm Workers of America, AFL-CIO have been reviewed and considered by Acting Presiding Justice Staniforth, and Justices Wiener and Work. The petitions are denied. (See Tex-Cal Land Management, Inc. v. Agricultural Labor Relations Bd. (1979) 24 Cal.3d 335 [ 156 Cal.Rptr. 1 , 595 P.2d 579 ].) Although we are authorized by law to summarily deny the petitions without explanation, we elect not to do so in this case. (See Holtville Farms, Inc. v. Agricultural Labor Relations Bd. (1985) 168 Cal.App.3d 388 [ 214 Cal.Rptr. 241 ].) We believe it will be helpful to set forth our reasons for the court’s summary dismissals. We propose, inter alia, to articulate the reasons why the Board cannot average the number of employees on the payroll preceding the election, thus to insure compliance with Labor Code 1 section 1156.3 in the future.

This cause involves petitions for review of an Agricultural Labor Relations Board (hereafter ALRB or Board) decision finding that Adamek & Dessert, Inc. (hereafter Adamek) had refused to bargain with the United Farm Workers (hereafter UFW). Petitions have been filed by both Adamek and the UFW. Adamek argues that the Board erred in certifying the UFW as the bargaining representative of its agricultural employees because the representation election was not held at a time when the Company employed 50 percent of the number of employees employed during peak season, because the UFW had intimidated the employees by making home visits to employees under the supervision of one who had been convicted of arson and because the ALRB agents conducting the election had allowed drinking near the polling place and a drunk inside the voting area. The UFW contends that although the Board certified the UFW as the bargaining representative as a result of the election, it prescribed an erroneous method for determining whether, for future cases, a company currently employed 50 percent of those employed at peak employment.

*975 Peak

In an effort to assure that representation elections occur when a sufficient number of employees are working in a seasonal industry, the Agricultural Labor Relations Act (hereafter ALRA) requires that the elections take place when the company is employing 50 percent of the number of employees employed during peak season. 2

In this case, a dispute arises over the Board’s determination as to whether during the payroll period immediately preceding the filing of an election petition, the Company was employing at least 50 percent of its peak employment. The UFW petitioned for an election among Adamek’s agricultural employees on December 8, 1980. In response to the petition, Adamek provided ALRB regional staff with payroll information for both the payroll period immediately preceding the filing of the election petition (the eligibility period) and the payroll period of highest employment during that calendar year (the peak period).

The undisputed employment figures for both the peak and eligibility periods are as follows:

*976 Peak Period

[[Image here]]

Eligibility Period

[[Image here]]

The ALRB regional office determined that, due to turnover during each payroll period, it was necessary to average the daily employment levels over the number of workdays in each payroll period. Since Adamek employed some employees on a weekly basis and others on a daily basis, all the employees did not appear on the same payroll list. The payroll periods of the regular and daily employees were of different durations, so the Board agents computed separate averages for each class of employees, then added them together for comparative purposes. The office discounted, as unrepresentative, certain days on which little or no work was performed during each payroll period. This method led to the following computation: During the peak there was an average of 122 employees (83 daily plus 39 regulars). During the eligibility period, there was an average of 64 employees (30 daily and 34 regulars). 3

Since 64 is more than 50 percent of 122, the ALRB agents decided that the election petition was timely, and thereafter held an election.

On review, the Board held that the methodologies employed by the Board agents, while not unreasonable, arbitrary or capricious, were not the preferred ipethods of computing peak under the circumstances. The Board stated that, in cases where an employer had both daily employees and employees paid over a longer period, the preferred method is to add all the employees together over the longer period and obtain one average number of employee days. The Board indicated that its preferred method of computation would be prospectively applied to avoid unfairness to the UFW, which had filed its election petition in reliance on other methods approved by the Board in earlier cases.

Adamek argues that under the approach the Board deemed appropriate in this case to determine whether it was employing 50 percent of the number of employees working during peak, it did not employ 50 percent of *977 the number of employees it employed during peak season and therefore the election was untimely and could not be held pursuant to section 1156.4.

The Union argues that the averaging approach is inconsistent with the ALRA in that it diminishes the ability of seasonal employees to seek a representation election.

Because the nature of employment practices differs so greatly in the agricultural industry, determination of the numbers needed to come to a conclusion on this issue has not been uniform and simple. When enforcement of the ALRA began, the Board applied the employee count method. Under this method, the names on the employers’ payroll immediately preceding the filing of the representation petition are counted and that amount is compared with the number of names which appear on the peak payroll during the preceding year.

The method of determining peak was first modified in Mario Saikhon (1976) 2 ALRB No. 2. Under the Saikhon approach, the number of names on the current payroll was compared with the average number of employees working each day during the peak payroll period. The average number was determined by adding up the total number of employees working during the period and dividing that number by the number of days in the payroll period.

Shortly thereafter, the Board modified the Saikhon method in Ranch No. 1, Inc. (1976) 2 ALRB No. 37. The Board introduced the concept of non-representative days into its determination of the average number of employees working during the peak period and applied the Saikhon averaging approach to both the peak payroll period and the payroll period immediately preceding the filing of the election petition. Under this modification, days on which few or no employees were working were not included in determining the average number of employees working during the payroll periods. The Board did not discuss the lawfulness or why it applied the averaging method to the payroll period immediately preceding filing of the petition.

Shortly after Ranch No. 1, Inc. was decided, the Board was faced in Luis A. Scattini & Sons (1976) 2 ALRB No. 43 with a factual situation almost identical to that presently before the court. Scattini hired two groups of employees: regular employees, who were paid on a biweekly payroll, and seasonal employees, who were paid on a daily payroll. The Board considered application of the Saikhon approach and an alternate approach which determined the average for each group separately, using a five-day period for determining the average of the seasonal employees. Under the alternate *978 approach, it could then add together the two averages to use as the number employed for comparing the current number with that employed during the peak payroll period. Because under both methods of calculation over 50 percent of the number of employees who worked during peak season were working during the week preceding filing of the representation petition, the Board held it was peak without deciding whether the Saikhon approach or the alternate approach was appropriate.

The Board thus resolved the unanswered question in Scattini by following the Saikhon method for determination of peak and by applying that method to a calculation of the number of employees working during both the peak payroll period and the eligibility period.

Obviously, from the history of litigation over issues of whether a company is currently employing 50 percent of the number of employees employed during peak, there are a number of ways in which it can be determined. The question is whether each is permitted by the act.

It is well established that an administrative agency is “entitled to deference when interpreting policy in its field of expertise.” (J.R. Norton Co. v. Agricultural Labor Relations Bd. (1979) 26 Cal.3d 1, 29 [ 160 Cal.Rptr. 710 , 603 P.2d 1306 ].) However, an administrative agency cannot alter or amend the statute which it is interpreting, or enlarge or impair its scope. (J.R. Norton Co. v. Agricultural Labor Relations Bd., supra, at p. 29.)

In determining whether Adamek employed 50 percent of its peak employment during the payroll period preceding filing of the election petition, the Board determined the average daily employment during the peak period and the eligibility period. It did not determine the total number of employees who worked during each period.

Section 1156.3, subdivision (a)(1) does not say if an average of the number of current employees is not less than 50 percent of the average number who worked during the period of peak employment. Section 1156.3, subdivision (a)(1) says if the “number . . . currently employed ... as determined from his payroll immediately preceding the filing of the petition, is not less than 50 percent” of those employed during the peak. The statute does not expressly permit averaging and averaging unnecessarily complicates a simple process.

While the statute does not say how the number employed during peak is to be determined, incorporation of a concept of averaging into a determi *979 nation of the number currently employed, as determined from the payroll immediately preceding the filing of the petition, alters the express wording of the section.

When averaging is used to determine the number of employees whose names appear on the payroll for the payroll period immediately preceding the election, the average may be less than 50 percent of the work force during peak while the number on the payroll exceeds 50 percent of those employed during peak. As a result, the Board would bar an election at a time when the petitioning seasonal employees have a right to take part in an election.

As indicated, the averaging approach began with Mario Saikhon, supra, 2 ALRB No. 2. In Saikhon, the Board determined that use of the employee count method for determination of peak employment was incorrect because hypothetically a company may need only 100 employees a day and because it hires dilferent employees each day may have a five-day payroll which includes 500 names. During the week preceding filing of the petition, the company may still have 100 employees a day, but be hiring the same employees each day. If only the names on the two payrolls are counted, it will appear that the company was not at 50 percent of peak during the payroll preceding the election. However, it is not alleged by anyone that this hypothetical situation has ever occurred. If a situation does arise in which a week appears to be peak only because of turnover of employees, the approach the Board should take which is consistent with the act should be litigated at that time. It should not be litigated under hypothetical situations. This court could obviously give advice on how it believes it should be handled, if a situation arises such as that set out in the hypothetical in Saikhon. However, such an opinion would be better analyzed after briefing on the issue when the parties have real rather than hypothetical facts before them.

In the present case, the record contains no evidence of the total number employed during peak. In future cases, this number should be determined. However, the Board determined there was a six-day average employment of 122 employees during Adamek’s peak payroll period. During the payroll period preceding filing of the election petition, 85 employees were on the payroll. As a result, the number employed as determined from the payroll immediately preceding the election was not less than 50 percent of the average number employed during peak.

Home Visits and Conduct of the Election

Adamek contends the certification of the UFW was erroneous because the Union conducted home visits to potential voters under the su *980 pervision of one who had been convicted of arson and that the Board agents conducting the election allowed employees to gather and drink beer near the election site and permitted a drunk to enter and ramble on about how bad the Company was within the election site.

Two employees who had been visited at home testified. Neither employee testified that they were threatened in any way or that any loud, violent or intimidating conduct occurred. Since employees are often too busy at work to talk about the advantages and disadvantages of a union, visitation at employees’ home is often necessary. The supervisor of the UFW office at the time of the Adamek election had been convicted six years earlier of burning empty buses used to transport strikebreakers. However, the evidence fails to indicate that any employee was aware of the supervisor’s criminal record. Since no effect of the home visits on the election was shown, the Board’s refusal to withhold certification of the UFW because of the home visits is supported by substantial evidence.

It seems undisputed that a group of individuals were drinking beer near the polling place. When asked to leave, they did so. There is no evidence that the beer drinking disrupted the election or interfered with any employee’s exercise of his or her right to vote. Similarly, there is no evidence that entry of the drunk and his remarks in the voting area in any way interfered with the election. Again, the Board’s refusal to withhold certification of the Union because of the conduct of a few individuals who were drinking is supported by substantial evidence.

Make Whole

In J.R. Norton Co. v. Agricultural Labor Relations Bd., supra, 26 Cal.3d 1 , the Supreme Court concluded that employees were entitled to be made whole for losses resulting from their employer’s refusal to bargain with a union which had been certified as their bargaining representative only if failure to do so was unreasonable or not in good faith. In this case, the Board concluded Adamek did not bargain because it did not believe the UFW had been lawfully certified. The Company believed the number of employees working during the payroll period preceding the election was less than 50 percent of those who worked when the Company was at peak. The Company was incorrect. However, the Board held the refusal to bargain was reasonable and in good faith even though it found certification of the Union to have been correct. While we disagree with the Board on the peak issue, we believe it was correct on the make whole issue. Over the years, the Board has expressed a number of opinions on determination of peak. Given the lack of clarity which existed when the petition was filed and the *981 failure to bargain occurred in this case, it seems the Company’s refusal to bargain was reasonable and in good faith.

The Union argues that Adamek’s bad faith in refusing to bargain was demonstrated by the Company’s delay of nearly three months in responding to the Union’s request to begin negotiations after certification. The Union cites San Justo/Wyrick (1983) 9 ALRB No. 55. It also argues: that bad faith by Adamek is further shown by its challenging the peak determination at a time it must have known fewer employees would be hired in the future since it permanently reduced its work force by approximately 50 percent shortly after the election.

The San Justo/Wyrick case, like the present case, involved an alleged technical refusal to bargain. There was a 10-week delay between the request to negotiate and the response which indicated the company intended to challenge the certification through refusing to bargain. The Board held that delay together with a supervisor’s statement that the company was not going to sign with the Union, even if an election was held valid, and a discriminatory refusal to rehire and other interference indicated that the refusal to bargain was not in good faith. The Board noted, “The question of . . . bad faith requires a review of the totality of the circumstances raised in both the representation case and the subsequent unfair labor practice case.” (San Justo/Wyrick, supra, at p. 6, fn. 6.)

Unlike here, in San Justo/Wyrick and the recent case of Ruline Nursery Co. v. Agricultural Labor Relations Bd. (1985) 169 Cal.App.3d 247 [ 216 Cal.Rptr. 162 ], there was serious misconduct by the company, in addition to the refusal to bargain, which indicated the “technical” refusal to bargain was in fact part of an overall strategy to prevent certification of the United Farm Workers.

The UFW also argues that the Company’s bad faith was shown because it must have known at the time it first raised the lack of peak objection that it intended to reduce the size of its work force by approximately 50 percent around January 1, 1981.

Throughout their briefs the parties discuss the estimation of the size of “peak.” Section 1156.4 refers to such an estimation. However, the entire concept of determination of peak is blown all out of proportion by attempts to estimate peak when no estimation is necessary. Section 1156.3 provides that a representation petition may be filed when “the number of agricultural employees currently employed by the employer named in the petition, as determined from his payroll immediately preceding the filing of the petition, *982 is not less than 50 percent of his peak agricultural employment for the current calendar year.” The language is not difficult to understand. For the calendar year, the payroll period preceding the filing of the petition is compared with the payroll for the peak payroll period. In this case, that meant comparing the payroll for the week ending December 3, 1980, with the week ending October 8, 1980. It in no way involved the number of employees hired or intended to be hired in the year following the year in which the election occurred.

The confusion can arise because of the use of the term “estimate” in section 1156.4. In 1156.4, after discussing the intent to allow seasonal workers to take part in the choice of their union or no union and the 50 percent of peak requirement, the Legislature stated: “In this connection, the peak agricultural employment for the prior season shall alone not be a basis for such determination, but rather the board shall estimate peak employment on the basis of acreage and crop statistics . . . and upon all other relevant data.”

When section 1156.4 and section 1156.3 are read together, it seems clear that the estimate referred to in section 1156.4 was applicable to potential future peaks which may occur within the applicable calendar year. (See Ranch No. 1, Inc., supra, 2 ALRB No. 37.) It was not intended to refer to estimating the size of the work force in future calendar years. As a result, for Adamek to challenge a 1980 election on the ground it did not employ 50 percent of its peak work force for that calendar year during the payroll immediately preceding the filing of the election petition, while at the same time reducing the size of its future work force, is not an indication of bad faith.

The remedy appropriate for an unfair labor practice is a matter within the special knowledge and expertise of the ALRB and must be given special respect by the courts. (Harry Carian Sales v. Agricultural Labor Relations Bd. (1985) 39 Cal.3d 209 [ 216 Cal.Rptr. 688 , 703 P.2d 27 ].) The Board saw a difference between the San Justo/Wyrick case and the present case and the expertise and special knowledge which led to this conclusion will be respected by the court.

Extension of Certification

In its order, the Board extended certification of the UFW for one year. Adamek argues that in order to have jurisdiction to do so, it was incumbent on the UFW to have filed a petition pursuant to section 1155.2 requesting extension of certification. This contention is without mer *983 it. Under the ALRA, once a union is certified as the bargaining representative for a company’s employees, the company has a duty to bargain with the union until the union is decertified through a second election. (Montbello Rose Co. v. Agricultural Labor Relations Bd. (1981) 119 Cal.App.3d 1, 30-31 [ 173 Cal.Rptr. 856 ].) If there is no extension of the certification, a second election can be held after 12 months have passed from the date of certification. The purpose of section 1155.2 is to extend this election bar period. (Kaplan’s Fruit & Produce Co. Inc. (1977) 3 ALRB No. 28.)

As noted above, the Board has considerable discretion in determining appropriate remedies once it has concluded an unfair labor practice has occurred. (See Harry Carian Sales v. Agricultural Labor Relations Bd., supra.) The ALRB is required by section 1148 to follow applicable NLRB precedent. The NLRB has repeatedly extended certification of a union after finding that the employer has refused to bargain in good faith. (See, e.g., N.L.R.B. v. All Brand Printing Corp. (2d Cir. 1979) 594 F.2d 926, 929 ; Franks Bros. Co. v. National Labor Relations Board (1944) 321 U.S. 702, 705-706 [ 88 L.Ed. 1020, 1023-1024 , 64 S.Ct. 817 ].)

In contending that a petition requesting an extension of certification is required for the Board to extend certification in an order following a finding of refusal to bargain in an unfair labor practice proceeding, Adamek relies on Yamada Brothers v. Agricultural Labor Relations Board (1979) 99 Cal.App.3d 112 [ 159 Cal.Rptr. 905 ]. Yamada involved a petition for a writ of mandate following the ALRB’s extension of certification of the union pursuant to section 1155.2. The court held that before the ALRB could extend a union’s certification, it had to find that the union had failed to bargain in good faith. Since Yamada was not a review of a Board order following an unfair labor practice proceeding, it is inapposite. In the present case, the Board did find that Adamek had failed to bargain in good faith.

For the reasons set out, the court does summarily dismiss the petitions.

Wiener, J., and Work, J., concurred.

The application of petitioner in No. D002985 for review by the Supreme Court was denied May 21, 1986. Bird, C. J., did not participate therein.

All statutory references are to the Labor Code unless otherwise specified.

Section 1156.3 provides in pertinent part:

Section 1156.4 provides:

During the eligibility period, there were 85 employee names on the payroll lists. The record does not reflect how many actual employee names appeared on the peak payroll lists. However, the maximum number working on any one day during the peak was 134.