“Although the legislative power under our state Constitution is vested in the Legislature, ‘the people reserve to themselves the powers of initiative and referendum.’ (Cal. Const., art. IV, § 1.) Accordingly, the initiative power must be liberally construed to promote the democratic process. (Raven v. Deukmejian [(1990) 52 Cal.3d 336,] 341 [276 *721Cal.Rptr. 326, 801 P.2d 1077].) Indeed, it is our solemn duty to jealously guard the precious initiative power, and to resolve any reasonable doubts in favor of its exercise. (Ibid., and cases cited.) As with statutes adopted by the Legislature, all presumptions favor the validity of initiative measures and mere doubts as to validity are insufficient; such measures must be upheld unless their unconstitutionality clearly, positively, and unmistakably appears. [Citation.]” (Legislature v. Eu (1991) 54 Cal.3d 492, 500-501 [286 Cal.Rptr. 283, 816 P.2d 1309], italics in original.)
While it might be possible for us to remedy the perceived constitutional defect of Proposition 73’s campaign contribution limit provisions by exercising our judicial power to reform the statutory language,1 we have not been asked to do so by the parties, the intervener or amici curiae herein. Whether or not we will be asked to consider such a course in the future remains to be seen. Accordingly, our opinion today does not attempt to effectuate the full reach of the electorate’s will, but endeavors to uphold its will by finding that Proposition 73 remains effective in substantial part.
Although I concur in the majority’s result, I write separately to express my view that Proposition 73’s ban on the public financing of election campaigns is also severable from the enjoined contribution provisions.
When provisions of an initiative statute are constitutionally or otherwise invalid, the void provisions must be stricken from the statute but the remaining valid provisions should be given effect if they are severable. (Santa Barbara Sch. Dist. v. Superior Court (1975) 13 Cal.3d 315, 330-331 [118 Cal.Rptr. 637, 530 P.2d 605]; see Calf arm Ins. Co. v. Deukmejian (1989) 48 Cal.3d 805, 821, 836 [258 Cal.Rptr. 161, 771 P.2d 1247].) “[I]n considering the issue of severability, it must be recognized that the general presumption of constitutionality, fortified by the express statement of a severability clause, normally calls for sustaining any valid portion of a statute unconstitutional in part.” (In re Blaney (1947) 30 Cal.2d 643, 655 [184 P.2d 892].) An express statement of severability generally imposes on the courts the duty of sustaining the electorate’s will as far as possible. (Cf. In re Application of Schuler (1914) 167 Cal. 282, 289 [139 P. 685].)
California cases prescribe three criteria for severability: (1) the invalid provision must be mechanically and grammatically separable; (2) it must be functionally separable; and (3) it must be volitionally separable. (Calfarm Ins. Co. v. Deukmejian, supra, 48 Cal.3d at pp. 821-822.) With regard to the third criterion, severability “ ‘depends on whether the remainder ... is complete in itself and would have been adopted by the legislative body had *722the latter foreseen the partial invalidity of the statute ... or constitutes a completely operative expression of legislative intent . . . [and is not] so connected with the rest of the statute as to be inseparable.’ [Citations.]” (Id., at p. 821.) Utilizing this test, I have no hesitancy concluding that had the electorate known that Proposition 73’s contribution limits might be held invalid, it nonetheless would have supported the proscription on public financing.
Significantly, the ballot materials accompanying Proposition 73 did not present the contribution limits as its one basic reform. Both the official title and summary prepared by the Attorney General and the analysis by the Legislative Analyst make clear that the initiative proposed to accomplish several election financing reforms, two of which consisted of limiting campaign contributions and prohibiting public funding of campaigns. These official summaries did not imply that the contribution limits comprised the measure’s central purpose, nor did they indissolubly link the limits to the public financing ban or any other proposed action.
The Attorney General’s summary of Proposition 73, located prominently at the beginning of the ballot materials concerning the proposition, informed voters in unmistakable terms that the measure proposed a number of reforms. It briefly summarized many of these reforms, listing them in the following order: campaign contribution limitations, limitations on gifts and honoraria to elected officials, prohibition of newsletters and mass mailings at public expense, prohibition of public funding for officials and candidates seeking elective office. The descriptions of the public financing ban and the campaign contribution limits were thus separated by descriptions of other reforms, and neither referred to the other.
Similarly, the Legislative Analyst’s analysis of the measure informed the electorate that Proposition 73: “[¶] • Establishes limits on campaign contributions for all candidates for state and local elective offices; [¶] • Prohibits the use of public funds for these campaign expenditures; and [¶] • Prohibits state and local officials from spending public funds on newsletters and mass mailings.” (Italics added.) The analysis subsequently described each of these proposed reforms, but nowhere promoted them as one indivisible idea.
Additionally, although some of the ballot arguments for and against Proposition 73 linked the campaign contribution limits with the public financing ban, most of the supporting arguments emphasized, without making a connection between the two, that the measure would prohibit politicians and special interest groups from using tax money to run their campaigns. The supporters also warned voters that a state matching fund *723program would be costly, and argued that “Taxpayers Should Not Be Forced to Shell Out Up to $70 Million Every Two Years for Their Extravagant Plan.” Thus, the public financing ban was promoted both as a vital part of Proposition 73 and as the principal distinguishing feature between that initiative and Proposition 68. In my view, it was this ban which resulted in the electorate giving the greater majority to Proposition 73, and the consequent defeat of Proposition 68’s alternative reforms based largely on public funding.
The dissent disagrees, concluding that the electorate would not have voted to enact the public financing ban without the contribution limits. (Dis. opn. of Arabian, J., post, at pp. 731-734.) The dissent argues first that the two provisions were indissolubly linked in the electorate’s mind. (Id., at pp. 731- 732.) It additionally points to the fact that California has no law permitting the use of public funds to finance political campaigns. (Id., at pp. 732- 733.) In the dissent’s view, because the proposed public financing ban was not a “reform” of any existing practice, it is unlikely that the electorate’s desire for reform would be satisfied by a measure that simply preserved the status quo. (Ibid.)
These are flawed arguments. First, as indicated above, the ballot materials described Proposition 73 as offering a number of separate campaign financing reforms, including contribution limits, a ban on the public financing of campaigns, restrictions on gifts and honoraria to elected officials, and the prohibition of newsletters and mass mailings at public expense. If anything, it may reasonably be inferred from the ballot arguments that it was the public financing ban, not the campaign contribution limits, that was the centerpiece of the initiative.
Second, it is incorrect to assume that the electorate would not have enacted the public financing ban merely because the public financing of campaigns had never previously been authorized under California law. The analysis by the Legislative Analyst informed voters: “California law does not generally permit any public money to be spent for campaign activities. A few local government agencies, however, have authorized the payment of public matching funds to candidates for local offices.” After providing this background, the analysis explained that Proposition 73 would prohibit the use of public funds by any candidate for public office. From this analysis and the other ballot materials, the electorate could reasonably understand that the initiative sought to reform these local campaign financing practices, and to affirmatively prevent the authorization of any future public financing at the state level (indeed, as Proposition 68 proposed to do). Thus, regardless of whether or not public financing had been authorized in the past, Proposition 73 sought to assure that such financing would not be permitted in the *724future. I have no trouble concluding that the electorate would nevertheless have voted for the ban had it foreseen the invalidity of the campaign „ contribution limits.2
Finally, I believe the passage of Proposition 68 deserves little weight in ascertaining the electorate’s will with respect to Proposition 73’s severability. As we recognized in Taxpayers to Limit Campaign Spending v. Fair Pol. Practices Com. (1990) 51 Cal.3d 744, 760 [274 Cal.Rptr. 787, 799 P.2d 1220], “[t]hat some voters would have been satisfied with the adoption of either proposition does not suggest that they wanted both, or that the same voters cast a majority of the affirmative votes for each initiative.” (Italics in original, fn. omitted.) Therefore, while campaign contribution limits were a part of both measures, that fact does not lead logically to the conclusion that the people who voted for Proposition 73 regarded such limits as the central purpose of the measure, or that they would prefer contribution limits with public funding to no limits at all.3
In sum, I believe Proposition 73’s public financing ban is mechanically, functionally and volitionally separable from the invalid contribution limits. I therefore concur in the majority’s conclusion that Proposition 68 remains inoperative.
Panelli, J., concurred.
Alternatively, the Legislature might consider amending the contribution provisions.
The dissent cites Johnson v. Bradley (1992) 4 Cal.4th 389, 410-411 [14 Cal.Rptr.2d 470, 841 P.2d 990], out of context. (Dis. opn. of Arabian, J., post, at p. 734.) That case involved a local measure, approved by a majority of the local voters of the City of Los Angeles, which provided for partial public financing of municipal elections by local taxes. There we decided that the locally enacted measure prevailed over the public financing ban contained in Proposition 73 under the home rule provision of the California Constitution (art. XI, § 5). Thus, the discussion quoted by the dissent herein was relevant to the issue whether Proposition 73’s ban was reasonably related to the statewide concern of enhancing the integrity of the electoral process. (See Johnson v. Bradley, supra, 4 Cal.4th at pp. 409-411.) In no manner, however, did our discussion in that case even purport to speak to the issue of severability regarding Proposition 73.
The dissent attempts to impeach this conclusion with evidence of an exit poll survey taken by the Los Angeles Times. (Dis. opn. of Arabian, J., post, at p. 733.) There are several good reasons, however, for rejecting this evidence as a basis for ascertaining the voters’ intent. First and foremost, the evidence is not a part of the record in this case; nor is it a proper subject for judicial notice (see Evid. Code, §§ 451, 452, 459). Second, while poll results might have some usefulness in certain contexts (see Moore v. California State Board of Accountancy (1992) 2 Cal.4th 999, 1015-1017 [9 Cal.Rptr.2d 358, 831 P.2d 798]), they can be “notoriously inaccurate” (see id., at p. 1027, dis. opn. of Mosk, J.), especially where, as here, the reliability of the particular evidence has not been tested or established.