MARSHALL, J., dissenting 470 U. S. same as the chances that he will be struck by lightning." App. 290–291 (emphasis added; citation omitted). Similarly a memorandum from Jensen to various United States Attorney's Offices states: "Selective Service's enforcement program is presently 'passive.' Non-registrants are brought to the Service's attention either when they report themselves or when others report them. Consequently, the first prosecutions are liable to consist of a large sample of (1) persons who object on religious and moral grounds and (2) persons who publicly refuse to register." Id., at 361–362. Perhaps, by itself, this evidence would not suffice to establish the second element of a prima facie case. However, it is more than adequate to make nonfrivolous the claim that the "passive" enforcement scheme identified for possible prosecution a disproportionate number of vocal opponents of draft registration. As to the third element, the decision to implement the "passive" enforcement system was certainly a decision susceptible to abuse. "This is indeed an exceptional area of national life where conscientious opposition to government policy has been intertwined with violations of the laws which implement the policy." United States v. Falk, 479 F. 2d 616, 625 (CA7 1973) (en banc) (Fairchild, J., concurring). The correlation between vocal opposition and violations of the law makes it relatively easy to punish speech under the guise of enforcing the laws. Here, the enforcement scheme was implemented with full knowledge that its effects would be particularly harsh on vocal opponents of the Government's policies. See App. 290–291, 361-362 (quoted supra, at 627 and this page); cf. 549 F. Supp., at 1384 (Government "recognized the passive program had potentially serious first amendment problems"). Such knowledge makes the scheme directly vulnerable to the charge that its purpose was to punish individuals for the exercise of their 598 MARSHALL, J., dissenting First Amendment rights. This Court has recognized that “[a]dherence to a particular policy or practice, 'with full knowledge of the predictable effects of such adherence. . . is one factor among others which may be considered by a court"" in determining whether a decision was based on an impermissible ground. Columbus Board of Education v. Penick, 443 U. S. 449, 465 (1979); see also Personnel Administrator of Massachusetts v. Feeney, 442 U. S. 256, 279, n. 25 (1979); id., at 283 (MARSHALL, J., dissenting) ("To discern the purposes underlying facially neutral policies, this Court has . . . considered the . . . foreseeability of any disproportionate impact"); United States v. Steele, 461 F. 2d 1148, 1152 (CA9 1972). 2 Thus, Wayte has established the first and third elements of a prima facie case, and has presented a colorable claim as to the second. As a result, there can thus be no doubt that the District Court did not abuse its discretion when it found that Wayte's equal protection claim was not frivolous. The Court, of course, has not viewed this case through the same lens. Instead of focusing on the elements of a prima facie case, and on whether Wayte presented sufficient evidence as to the existence of each of these elements to earn the right to discover relevant information in the Government's possession, the Court leaps over these two issues and proceeds directly to the merits of the equal protection claim. The Court's analysis is flawed in two respects. First, as I have shown, the Court ignores the simple fact that, if Wayte is entitled to discovery, his claim cannot be rejected on the merits for lack of evidence. Second, and of equal importance, the Court errs in the manner in which it analyzes the merits of the equal protection claim. It simply focuses on the wrong problem when it states that "the Government treated all reported nonregistrants similarly" and that "those prosecuted in effect selected None of the evidence presented by the Government to the District Court places in any serious question the existence of these three elements. MARSHALL, J., dissenting 470 U. S. themselves for prosecution by refusing to register after being reported and warned by the Government." Ante, at 610. Those issues are irrelevant to the correct disposition of this case. The claim here is not that the Justice Department discriminated among known violators of the draft registration law either in its administration of the "beg" policy, which gave such individuals the option of registering to avoid prosecution, or in prosecuting only some reported nonregistrants. Instead, the claim is that the system by which the Department defined the class of possible prosecutees-the "passive" enforcement system-was designed to discriminate against those who had exercised their First Amendment rights. Such governmental action cannot stand if undertaken with discriminatory intent. As this Court has clearly stated, "for an agent of the State to pursue a course of action whose objective is to penalize a person's reliance on his legal rights is 'patently unconstitutional."" Bordenkircher v. Hayes, 434 U. S. 357, 363 (1978); see also United States v. Goodwin, 457 U. S. 368, 372 (1982). If the Government intentionally discriminated in defining the pool of potential prosecutees, it cannot immunize itself from liability merely by showing that it used permissible methods in choosing whom to prosecute from this previously tainted pool. Cf. Connecticut v. Teal, 457 U. S. 440, 450-451 (1982). Under the Court's flawed approach, there would have been no equal protection violation in Yick Wo v. Hopkins, 118 U. S. 356 (1886), this Court's seminal selective prosecution decision. In Yick Wo, the Court reversed a conviction under a municipal ordinance that prohibited the construction of wooden laundries without a license. The Court held that such a conviction could not stand because the municipal licensors had discriminatorily denied licenses to individuals of Chinese origin. If the Court then had focused only on the prosecutions themselves, as it does now, it would have found no discrimination in the choice, among violators of the ordi 598 MARSHALL, J., dissenting Indeed, all but nance, of the individuals to be prosecuted. one of these violators were of Chinese origin. Instead, the Court properly focused on the official action that led to those prosecutions. In Yick Wo, that prior action was the discriminatory denial of licenses, which affected the definition of the class from which prosecutees were chosen. In this case, the referrals made by Selective Service to the Justice Department for investigation and possible prosecution played a similar role and may also have been discriminatory. It is to that issue that the Court should have directed its attention. I do not suggest that all prosecutions undertaken pursuant to passive enforcement schemes warrant evidentiary hearings on the question of selective prosecution. But where violations of the law are so closely intertwined with political activity, where the speech at issue is so unpalatable to the Government, and where the discriminatory effect is conceded, the need for a hearing is significant and in no way opens the door to an onslaught of such hearings in less compelling contexts.3 Here, I believe that Wayte has raised sufficient questions about the Government's intentions to be entitled to obtain access to evidence in the Government's possession. I therefore dissent from the Court's outright dismissal of his equal protection claim. 'In my mind, Wayte's claim that the "passive" enforcement scheme placed a direct burden on First Amendment freedoms, ante, at 607-610, should not be addressed at this stage in the litigation. The materials that Wayte sought to discover, and that he well may be entitled to discover, could be relevant to that claim. The Court of Appeals should resolve the issue of access to evidence on remand; the resolution of the merits of Wayte's claims should await a final determination of that issue. 470 U. S. Syllabus BENNETT, SECRETARY OF EDUCATION v. CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE THIRD CIRCUIT No. 83-2064. Argued January 8, 1985-Decided March 19, 1985 In earlier proceedings in this litigation, this Court, reversing the Court of Appeals' judgment, held that the Federal Government may recover misused funds from States that provided assurances that federal grants would be spent only on eligible programs under Title I of the Elementary and Secondary Education Act of 1965, which provided for grants to support compensatory education for disadvantaged children in low-income areas. Bell v. New Jersey, 461 U. S. 773. However, the Court expressly declined to address the issue whether substantive provisions of the 1978 Amendments to the Act apply retroactively for determining if Title I funds were misused in earlier years. On remand, New Jersey argued that the 1978 Amendments, which relaxed the eligibility requirements for local schools to receive Title I funds, should be applied in determining whether funds were misused during the years 1970-1972. The Court of Appeals agreed and remanded the case to petitioner Secretary of Education to determine whether the disputed expenditures conformed to the 1978 standards. Held: The substantive standards of the 1978 Amendments do not apply retroactively for determining if Title I funds were misused under previously made grants. Pp. 638-646. (a) The Court of Appeals' reliance-based on language from Bradley v. Richmond School Board, 416 U. S. 696 on a presumption that statutory amendments apply retroactively to pending cases is inappropriate in this context. Both the nature of the obligations that arose under the Title I program and Bradley itself suggest that changes in substantive requirements for federal grants should not be presumed to operate retroactively. Moreover, practical considerations related to the administration of federal grant programs imply that obligations generally should be determined by reference to the law in effect when the grants were made. Retroactive application of changes in the substantive requirements of a federal grant program would deny both federal auditors and grant recipients fixed, predictable standards to determine if expenditures are proper. Pp. 638-641. (b) Neither the statutory language nor the legislative history indicates that Congress intended the substantive standards of the 1978 |