¶1with whom COMPTON, Justice, joins, dissenting.
¶2Given this court’s underlying assumption that positional bias exists, I must dissent from its conclusion that, as a matter of law, AS 15.15.030(6) does not impermissibly burden the right to vote. Op. at 640. I cannot agree that the modest financial interest the State identifies is important enough to justify the burden the statute may impose.
¶3For purposes of reviewing the superior court’s grant of summary judgment to the State, this court forthrightly assumes that positional bias can affect at least five percent of the votes cast. Op. at 639 n. 7. This assumption, or at least an assumption that some positional bias exists, is mandated by the procedural posture of this ease. Because the State submitted no admissible evidence supporting a finding that positional bias does not exist or cannot be measured, our starting point is Sonneman’s complaint. It alleged that positional bias benefits whichever candidate is at the top of the ballot. It asserted that studies confirmed the existence of positional advantage, affecting as much as six to seven percent of the candidates’ votes.
¶4Despite the absence of admissible evidence rebutting the complaint’s allegations of positional bias, the court concludes that the State is entitled to judgment as a matter of law; it necessarily reasons that Sonneman has raised no genuine fact that is material. Op. at 635-636, 641. In my view, the materiality of the factual dispute turns on whether the statute imposes an impermissible burden on the right to vote. If a five percent shift constitutes an impermissible burden, then the existence — and extent — of positional bias is a fact dispute that is necessarily material.
¶5Positional bias could impose a serious burden on the right to vote. Many elections in *642this state have been decided by margins smaller than five percent. Our opinions reflect some of those contests. See, e.g., Cissna v. Stout, 931 P.2d 363, 364 (Alaska 1996) (affirming in election recount decided by one vote). Assuming the facts alleged by the complaint, positional bias alone could alter the outcome of all single-district races decided by margins of less than six percent.
¶6The potentially significant effects of positional bias cannot be justified by the weak regulatory interests identified by the court. Op. at 639-640. On appeal, the State articulates only one interest: cost reduction. According to the court, the State saved an estimated $64,024 per election cycle by eliminating rotation in single-district races. Op. at 640. To me, the burden positional bias potentially imposes far outweighs savings of this magnitude.
¶7The State also argued in the superior court that rotation would confuse voters by creating discrepancies between sample ballots and actual ballots. This court relies on that argument, even though the State has chosen not to assert it on appeal. Op. at 639-640. It seems unlikely that this is of significant continuing concern, given that sample ballots also reflect statewide races, for which rotation is still practiced. If confusion were of real concern, a clear disclaimer on sample ballots would cure it.
¶8The interests the court identifies are minor in comparison with the burden we must assume the statute imposes on the right to vote. I would therefore remand for determination of whether positional bias exists, and the extent of its effect. Only then can a court decide whether the legislature’s choice unreasonably burdened the right to vote.