[EM] piling on against IRV

[Abd ul-Rahman Lomax]

At 03:30 PM 5/9/2010, Jameson Quinn wrote:

>2010/5/9 robert bristow-johnson 
><<mailto:rbj at audioimagination.com>rbj at audioimagination.com>
>in comparison, i have seen 3 different TTR elections for City 
>Council in Burlington.  none had more than 55% turnout on runoff day 
>(in comparison to the number of voters that came on the first 
>election day).  the IRV election had 93% of the voters participating 
>in the final round.   93% turnout is a lot better turnout than 
>55%.  every voter that expressed an opinion of at least one of the 
>two candidates that made it to that final round participated in the 
>actual choice of the elected candidate.

What IRV election and what City Council elections are RBJ talking about?

http://www.ci.burlington.vt.us/ct/elections/docs/ElectionResultsFinal030309.pdf 
shows an election in 7 wards. 6 found a majority in the primary. 
These are all city elections, Burlington does not schedule its city 
elections with the November general elections, apparently, which is 
what jurisdictions do that want to increase participation. Total 
votes in the primary, March 3, 2009, with the votes for each 
candidate (I = Independent, G = Green, P = Progressive, R = 
Republican, write-in count is last):

Ward 1: 737.                            G:94    I:635   / 8
Ward 2: 651.  D:173     P:447           G:28            / 3
Ward 3: 996.  D:461     P:498           G:34            / 3
Ward 4: 1884. D:957             R:923                   / 4
Ward 5: 1368. D:1115                    G:245           / 8
Ward 6: 1139. D:979                     G:105           / 55
Ward 7: 1696. D:619             R:612           I:461   / 4

Wards 3, 4, and 7 were close races. Notice that the parties seem to 
not tend to run candidates in wards where they can't win. The Greens, 
however, in spite of very low support, were running in 5 out of 7 
wards. The Democrats run in all wards except for Ward 1, where I 
would guess there is a popular independent candidate. The 
Progressives and Republicans are only runing in two wards, never 
against each other, always against a Democrat. The composition from 
this election was I:1, D:3, P:2, and Ward 7 was close enough to 
require a runoff. (Are the rules 40%? If so, we can see how that's a 
Bad Idea from Ward 3, where the election was really close, it's not 
at all clear who might have won a runoff. Or perhaps it's considered 
that you win some and lose some and that, in the end, it evens out.)

In Ward 7, the runoff results:
Ward 7: 940. D:425              R:515

940/1696 is 55% of the first day turnout. However, if we look at 
those who voted on the first day for the R and the D only, it's 76%. 
The vote for the Democrat in the runoff was 68% of the vote in the 
primary; for the Republican, it was 84%. This was a comeback 
election; comeback elections are rare in IRV in nonpartisan 
elections, but Burlington is partisan, the IRV was comeback also. In 
Top-Two Runoff, nonpartisan comeback elections are roughly one-third 
of runoffs. (With IRV, by "comeback," I mean that the first rank 
leader ends up losing to the runner-up -- or perhaps to an even lower 
first-rank rated candidate, but I've never seen that with IRV. It 
happened in the first Bucklin election held in the U.S.)

What happened in Ward 7 could be explained by differential turnout, 
Republicans, perhaps, being more likely to vote, perhaps because this 
election was important to them -- it was their only shot at a seat on 
the city council, whereas it wasn't so important to the Democrats. Or 
it could be explained by some preference for the Republican from the 
Independent candidate, I don't know.

Participation in the final round in top two runoff, though, is 
clearly a different matter from participation in IRV. It actually 
involves a separate voting process. In the primary, all registered 
voters have one vote to cast, and they have an independent vote to 
cast in the runoff. In IRV, they cast additional votes only if their 
first vote is eliminated. If they are unlucky enough to have their 
first choice survive to the last round, their additional votes don't 
count. In other words, they do worse by doing better, having been 
"protected" by IRV from "harming" their first choice.

>This is a good point. But:
>
>at least 38% fewer voters come to the polls on runoff day and get 
>counted with TTR.  that is what our experience is.
>
>
>You mean "on average" or "so far, at least" or something. There's no 
>"at least" here; in a contentious runoff, it's possible for turnout 
>to increase.

Yes. It depends on the importance of the race to the voters, and on 
their perception of how much difference it makes. High turnout in a 
runoff may be a sign of a serious race between two strong candidates 
with serious power hanging on it, or it may be a sign that a 
seriously bad candidate made it into the runoff because of 
vote-splitting in the primary. Low turnout generally indicates 
relative voter indifference to the outcome. (Which can mean 
satisfaction with the likely outcome, or dissatisfaction.)

When runoffs are special elections, not combined with other 
elections, fewer votes will have enough at stake to vote. As I've 
mentioned, in Cary, North Carolina, the mayoral election was held in 
October, with a November runoff if needed. Thus the runoff turnout 
was, in fact, slightly higher than the primary. It's remarkable, in 
fact, that it wasn't much higher, but the mayoral election was 
scheduled in off-years, not the Presidential election year. You want 
high turnout, schedule an election with the U.S. Presidential election.

But high turnout is not productive of higher social utility, it adds 
noise, as voters with less interest -- and very likely less knowledge 
-- vote. I favor, greatly, improving access to voting so that it is 
equal for all voters, but if the *difficulties* of access are equal, 
some level of difficulty probably improves results. What's harmful is 
inequitable difficulty, with convenient access for, say, affluent 
voters, and inadequate facilities, long lines, etc., for less 
well-off voters. With special elections, though, because of lower 
turnout, when that happens, voting may be very easy, comparatively, 
so low turnout really does indicate low interest.