PEARSON, J.

The appellants, Charles W. Oliver and his wife, Linda L. Oliver, brought an employment discrimination suit against the respondent, Pacific Northwest Bell Telephone Company (PNB), alleging that a company-wide policy which subjected employees to disciplinary proceedings for dishonest acts committed outside of employment was discriminatory because it had a disproportionate impact upon black employees. The trial court granted a directed verdict in favor of PNB, ruling that the disparate impact analysis was inapplicable to the present facts, and even if it did apply, the Olivers had failed to establish a prima facie case. We agree with the trial court and affirm its order directing verdict in favor of PNB.

The Olivers, who are black, were employees of PNB. Linda Oliver began employment as a staff clerk in the engineering department in April 1970; Charles Oliver began employment in December 1971 as a data systems analyst.

In 1976, while employed by PNB, the Olivers' home was burglarized. After the burglary, the Olivers falsified several insurance claims. In July 1978, the Olivers were charged with filing fraudulent insurance claims. PNB took no immediate action regarding the Olivers' employment status because the details of the Olivers' conduct were unclear. In November 1978, the Olivers pleaded guilty to a gross misdemeanor (attempt to file a false insurance claim). Consequently, in December 1978, PNB terminated the Olivers for violation of a company-wide policy which requires all employees to conduct themselves honestly, on and off the job.

In September 1980, the Olivers filed a complaint against PNB in the Superior Court for King County. The Olivers contended that PNB's policy of subjecting employees who commit dishonest acts outside of employment to disciplinary proceedings was in violation of RCW 49.60, because it had a disparate impact on blacks. The matter was tried before a jury. At the conclusion of the Olivers' case, PNB moved for a directed verdict. The trial court granted PNB's motion and dismissed the Olivers' complaint. The court held that the disparate impact theory of discrimination did not apply in the Olivers' case, and even if it did apply, the Olivers failed to establish a prima facie case. The Olivers appealed to the Court of Appeals which certified the matter to this court for direct review pursuant to RAP 4.2(a).

I

[1] The rules for appellate review of an order directing verdict in favor of a party are well established: (1) evidence must be considered in favor of the nonmoving party; (2) no discretion is involved; and (3) the directed verdict will be upheld where there is no competent evidence, nor reasonable inferences arising therefrom, which would sustain a jury verdict in favor of the nonmoving party. Shelby v. Keck, 85 Wn.2d 911, 913, 541 P.2d 365 (1975). With these principles in mind, we now review the trial court's order directing verdict in favor of PNB.

[2] In Washington, the prohibition against employment discrimination is found in RCW 49.60, which prohibits an employer from discharging or barring any person from employment based on race. RCW 49.60 is patterned after Title 7 of the Civil Rights Act of 1964, as amended, 42 U.S.C. § 2000e-2 (1982). Consequently, decisions interpreting the federal act are persuasive authority for the construction of RCW 49.60. See Fahn v. Cowlitz Cy., 93 Wn.2d 368, 610 P.2d 857, 621 P.2d 1293 (1980).

Discrimination claims under RCW 49.60 may be brought under one of two theories, either "disparate impact" or "disparate treatment". See Fahn, at 378; Shannon v. Pay'n Save Corp., 104 Wn.2d 722, 726, 709 P.2d 799 (1985). Either theory may be applicable to the same set of facts. Shannon, at 732 (citing Page v. U.S. Indus., Inc., 726 F.2d 1038, 1045 (5th Cir. 1984); International Bhd. of Teamsters v. United States, 431 U.S. 324, 335 n. 15, 52 L.Ed.2d 396, 97 S.Ct. 1843 (1977)).

In the present case, the Olivers did not allege discrimination under the disparate treatment theory, but rather based their claim solely upon disparate impact. Therefore, we limit our review to whether the trial court properly granted a directed verdict against the Olivers based on their disparate impact claim.

[3] To establish a prima facie case of disparate impact, the plaintiff must prove: (1) a facially neutral employment practice, (2) falls more harshly on a protected class. Shannon, at 727; International Bhd. of Teamsters, at 349; Connecticut v. Teal, 457 U.S. 440, 446, 73 L.Ed.2d 130, 102 S.Ct. 2525 (1982); Dothard v. Rawlinson, 433 U.S. 321, 329, 53 L.Ed.2d 786, 97 S.Ct. 2720 (1977); Griggs v. Duke Power Co., 401 U.S. 424, 430-31, 28 L.Ed.2d 158, 91 S.Ct. 849 (1971). Proof of an employer's intent to discriminate in adopting a particular practice is not required. Shannon, at 727; Griggs, at 432; International Bhd. of Teamsters, at 335 n. 15. Upon establishing a prima facie case, the burden then shifts to the defendant to establish that the practice complained of has a "`manifest relationship' to the position in question", Shannon, at 727 (quoting Griggs v. Duke Power Co., supra at 432), or is justified by a business necessity. Albemarle Paper Co. v. Moody, 422 U.S. 405, 425, 45 L.Ed.2d 280, 95 S.Ct. 2362 (1975).

In this case, we must determine if the Olivers have established a prima facie case of disparate impact. Thus, initially, we must decide whether the Olivers have proven that PNB's company-wide policy is facially neutral. The Olivers contend that PNB's policy of subjecting employees to disciplinary proceedings for committing dishonest acts has a disproportionate adverse effect on blacks because a disproportionate number of blacks commit dishonest acts. Thus, according to the Olivers, PNB's policy is susceptible to a disparate impact analysis. Aside from this general allegation, the Olivers engage in very little analysis regarding whether PNB's company-wide policy is facially neutral. PNB, on the other hand, argues that the disparate impact analysis is inapplicable to PNB's company-wide policy because it does not include objective, nondiscretionary criteria and is therefore not a facially neutral employment practice. We agree.

In Washington, in order for an employment practice to be characterized as facially neutral and therefore subject to analysis under disparate impact, a plaintiff must demonstrate that he is attacking an employment practice that includes objective, nondiscretionary features. Shannon v. Pay'n Save Corp., 104 Wn.2d at 733. In this case, the trial court held that disparate impact analysis was not available to the Olivers because PNB "used a discretionary and subjective decision-making process in determining action to be taken when an employee was found to have committed a dishonest act and/or was convicted thereof." Verbatim Report of Proceedings, at 311. The evidence strongly supports the trial court's conclusion.

The record indicates the Olivers were terminated for violating PNB's company-wide policy which requires all employees to conduct themselves honestly on and off the job. Under this policy, a dishonest act, including a dishonest act resulting in a criminal conviction, potentially subjects the employee to disciplinary action, which may include dismissal. However, PNB's policy is not a flat rule which requires automatic termination resulting from commission of a dishonest act. Rather, PNB addresses each specific situation regarding a violation of the standards of conduct on a case-by-case basis, and ultimate disciplinary action depends on a variety of factors. In fact, several employees known to have committed dishonest acts remain employed with PNB.

Thus, it is clear PNB's policy is a discretionary and subjective employment practice that does not include objective or nondiscretionary practices which automatically terminate an employee for the commission of a dishonest act outside of employment. Therefore, the Olivers were not terminated by a facially neutral employment practice. Accordingly, disparate impact analysis is inapplicable to the facts of this case and the trial court's order directing verdict in favor of PNB should not be disturbed on appeal.

II

Even if we were to hold that subjective employment practices are susceptible to disparate impact analysis, see, e.g., Shannon v. Pay 'n Save Corp., 104 Wn.2d 322, 737, 739, 709 P.2d 799 (1985) (Utter, J., dissenting; Brachtenbach, J., dissenting), the Olivers have failed to establish the second element of a prima facie impact case, i.e., that the employment practice complained of falls more harshly on a protected class. As previously discussed, under disparate impact analysis the plaintiff bears the burden of proof on the issue of whether there is a facially neutral employment practice that has a substantial disproportionate impact on a protected class. Therefore, even assuming the Olivers could establish the first element of a prima facie case, they still must prove the employment practice complained of falls more harshly on a protected class.

Notwithstanding certain excluded evidence, the Olivers contend that they have established a prima facie case of disparate impact based on statistical data presented at trial by Ms. Rao and Dr. Weis. Ms. Rao concluded that PNB's policy had a disproportionate impact on blacks who committed dishonest acts outside of employment. The trial court, on the other hand, concluded that the statistical data was based on too small a sample to be reliable. The court noted that one or two additional cases would have a substantial impact on what the Olivers claim these statistics show. We agree with the trial court's conclusion.

[4] Generally, the standard of proof necessary to establish a disproportionate impact upon a protected class is whether the plaintiff has produced evidence sufficient to justify an inference that the employment practices complained of caused a substantial disproportionate exclusionary impact on the protected class, other than by mere chance. B. Schlei P. Grossman, Employment Discrimination 1370 (2d ed. 1983). The primary means of proving a substantial disproportionate impact on a protected class is through the use of statistical evidence. See EEOC v. Local 638, 401 F. Supp. 467, 478 (S.D.N.Y. 1975) (citing United States v. Wood, Wire Metal Lathers Int'l Union, 471 F.2d 408, 414 n. 11 (2d Cir.), cert. denied, 412 U.S. 939 (1973), aff'd as modified, 532 F.2d 821 (2d Cir. 1976)); see also B. Schlei P. Grossman, at 1324.

[5] Although statistics play a vital role in establishing a prima facie impact case, their usefulness may be limited. As the United States Supreme Court stated in International Bhd. of Teamsters v. United States, 431 U.S. 324, 340, 52 L.Ed.2d 396, 97 S.Ct. 1843 (1977), statistics "come in infinite variety and like any other kind of evidence, they may be rebutted. In short, their usefulness depends on all of the surrounding facts and circumstances." The Court further noted that "[c]onsiderations such as small sample size may, of course, detract from the value of such evidence". International Bhd. of Teamsters, at 339 n. 20. See also Morita v. Southern Cal. Permanente Med. Group, 541 F.2d 217, 220 (9th Cir. 1976) (statistical evidence derived from an extremely small universe has little predictive value and must be disregarded), cert. denied, 429 U.S. 1050 (1977); Contreras v. Los Angeles, 656 F.2d 1267, 1273 n. 4 (9th Cir. 1981) ("[s]tatistics are not trustworthy when minor numerical variations produce significant percentage fluctuations"), cert. denied, 455 U.S. 1021 (1982). Because statistical evidence derived from an extremely small sample has little predictive value, and is therefore unreliable, the use of such evidence must be closely scrutinized to avoid inferences of disproportionality, which are based upon conjecture, speculation, or chance, rather than discriminatory practices.

Here, substantial evidence in the record supports the trial court's conclusion that the statistical data presented by the Olivers was based on too small a sample to be reliable. Ms. Rao examined employment data involving "dishonest acts" of 100 employees of PNB over a period from 1975 through 1980. Ms. Rao found that out of 100 individuals, 82 individuals had performed dishonest acts inside employment. Further, of those 82 employees, 13, or 15.9 percent, were black. The remaining 18 individuals committed dishonest acts outside employment, and of those, 6, or 33.33 percent, were black, including the Olivers. Ms. Rao then compared the 33.33 percent of the blacks discharged for dishonest acts outside employment with 975 or 4.6 percent black PNB employees, out of a total 21,415 PNB employees. From this comparison, Ms. Rao concluded that the number of black employees found to have committed dishonest acts was significantly disproportionate with non-blacks.

Given the size of the sample, the total number of PNB employees and the significant percentage fluctuation which would occur with a minor numerical variation (for example, excluding the Olivers, the percentage drops from 33.33 percent to 25 percent), the statistical evidence presented by the Olivers is insufficient to support any inference that PNB's policy caused a substantial disproportionate exclusionary impact on a protected class, other than by chance. Accordingly, even viewing the evidence in a light most favorable to the Olivers, the statistical evidence presented is insufficient to support a prima facie impact case.

III

The Olivers also contend that the trial court erred in excluding Dr. Weis' opinion testimony which, based on Ms. Rao's statistical data, attempted to establish that discrimination was a factor in the high proportion of blacks terminated by PNB for committing dishonest acts outside of employment. The trial court excluded Dr. Weis' testimony on the ground that Dr. Weis was not qualified.

[6] Qualifications of expert witnesses are to be determined by the trial court within its sound discretion, and rulings on such matters will not be disturbed unless there is a manifest abuse of discretion. Orion Corp. v. State, 103 Wn.2d 441, 462, 693 P.2d 1369 (1985). Here, there is no manifest abuse of discretion. The trial court excluded Dr. Weis' opinion testimony because the court felt that the witness' testimony was unrelated to PNB's work force. The record supports the trial court's conclusion. The record indicates Dr. Weis dealt with the general criminal justice system and primarily with juvenile offenders; none of his research included PNB's work force. Accordingly, we will not disturb the trial court's exclusion of Dr. Weis' testimony.

Finally, the Olivers contend that the trial court improperly excluded the testimony of Ms. Kleyn. Ms. Kleyn testified that currently in King County the number of blacks introduced into the criminal justice system is substantially disproportionate to their representation in the general population in King County. Based on this conclusion, the Olivers argued that evidence of general population data versus the criminal justice system was meaningful to the experts in that it established a "data base" from which they could make predictions and forecasts regarding PNB's work force. The trial court excluded Ms. Kleyn's testimony on the ground that it was inadequate, prejudicial and misleading to the jury. We agree with the trial court's exclusion of Ms. Kleyn's testimony.

[7] The admission or refusal of evidence lies largely within the sound discretion of the trial court and will not be overturned on appeal unless abuse of discretion is shown. Himango v. Prime Time Broadcasting, Inc., 37 Wn. App. 259, 266, 680 P.2d 432, review denied, 102 Wn.2d 1004 (1984). Here, there is no abuse of discretion. Ms. Kleyn's statistics related only to November 1975, 3 years prior to the Olivers' termination date, and Ms. Kleyn expressly stated that she was not qualified to testify whether statistics for succeeding months or years would be substantially the same. Accordingly, the trial court properly excluded Ms. Kleyn's testimony.

We affirm the trial court's order directing verdict in favor of PNB.

DOLLIVER, C.J., and UTTER, BRACHTENBACH, DORE, CALLOW, GOODLOE, and DURHAM, JJ., concur.

ANDERSEN, J., concurs in the result.