11 December 2006
From a forthcoming paper on legislative redistricting commissions: Iowa
has used the same scheme for the past three redistricting cycles. A commission draws three maps, and the legislature selects one of those.
The attached seats-votes graphs are for the 2000 and 2004 state house elections, before and after the 2001 cycle. As we can see, responsiveness (the slope of the curve) is high and remains high afterwards, suggesting that the fraction of contested seats is high, and justifying its reputation as a model for redistricting.
However, the curve is definitively below 50% at the median vote, meaning that an equal vote will almost always split the seats unevenly. (In this case, the Republican party gains the advantage.) This suggests that redistricting is less effective in this case.
Given Iowa's reputation as a well-run redistricter, one wonders how much it is deserved. It's also fair to wonder what would happen if this system were applied to another state where voting was racially polarized.