Instant Runoff Voting (IRV) and racial minorities

It has sometimes been stated (falsely) by IRV-propagandists, that IRV helps minorities. But in fact, the available evidence indicates it hurts them.

[Skip to conclusion]    [About the (unrelated?) fact IRV favors "extremists"]

To see this, we shall compare minority representation in federal IRV-elected and plurality-elected seats versus the minority percentages in the population at large, in the top IRV-using country (Australia) and the top two plurality-using countries (USA & UK) in years 2005-2008. We find, e.g, that there are zero racial-minority IRV seatholders... (We shall also examine some US cities and Ireland.)

The Australian House (elected via IRV)

Color photos dated 3 June 2008 of every one of the 150 Australian MPs (their MPs are elected via IRV) are available as this printable PDF document which we originally obtained from here: http://www.aph.gov.au/house/members/42p_Members_photos.pdf. Every single MP is white, despite the fact Australia's overall population is (the following figures are from the 2008 online CIA world fact book; and with slight correction from Australian 2006 census data)

White 91%Asian 7%Aboriginal and other 2%

The USA's House and Senate (elected via plurality voting)

Contrast those lily-white Australian MPs with the USA's plurality-voting-elected House and Senate:

Racial Composition of the 110th US Congress (Jan. 2007)

 

U.S. House

U.S. Senate

White

366

94

Black

40

1

Hispanic

23

3

Asian

5

2
American Indian

1

0
Total

435

100

(The 435 house members include 73 women. The original source of this data is http://www.house.gov/daily/110thdemographics.htm.) Meanwhile the US Census Bureau (2008) claims the USA population is:

White 80.1%Black 12.8%Asian 4.4% Am.Indian & Pacific Islander 1.2% Mixed race 1.6%

Also 14.8% said they had "Hispanic or Latino origin" but this is not being counted as a "race."

For a perhaps-irrelevant blast from the past, you can also see a group photograph of all Australian Labour MPs in 1901. There were zero racial minorities present then also.
We can also note that the probability that 150 randomly drawn Australians would happen to be all-white, is 0.92150≈3.7×10-6. This proves (with statistical confidence level 99.9996%) that the Australian House is not racially representative. And one could also argue the US Senate is not either. One could also note the US House & Senate are not gender-representative either [although that is not a minority problem since women are a majority of the population]. This is even more clear since, e.g, the probability 435 coin tosses would yield ≤73 heads, is 2.1×10-47, so we can have confidence

99.999999999999999999999999999999999999999999998%
that the US House is gender-biased. So all these governmental bodies have unrepresentative racial and/or gender compositions, but the Australian House with zero race-minority members clearly more-extremely underrepresents minorities than the USA.

The UK House of Commons (elected via plurality voting)

As of 25 July 2008, it contained 125 female and 521 male members (646 total – but that includes the speaker and his 3 deputies, who do not normally vote). Party breakdowns are also available, but the official House of Commons website does not give a racial breakdown (as far as I can tell).

From press reports we know this partial information: there are two black women MPs, namely Dawn Butler and Diane Abbott; the only conservative-party black MP is Adam Afriyie; Keith Vaz is "the longest-serving Asian MP" and others are Piara Khabra and Ashok Kumar; there are no Asian women MPs.

In a 19 January 2005 BBC News story titled Malik rejects all-black MP lists, Labour MP Shahid Malik was quoted saying "Just 13 of Britain's 659 MPs are from ethnic minority groups." However, that changed slightly in the May 2005 election. As of July 2008, a web site called "Operation Black Vote (www.obv.org.uk)" says there are 15 black MPs (where they appear to be counting Asians as "black"),

Date profile writtenColored MP (linked to OBV's profiles)
01-08-2006-- Adam Afriyie, MP
01-08-2006-- Ashok Kumar, Dr., MP
01-08-2006-- David Lammy, MP
01-08-2006-- Dawn Butler, MP
01-08-2006-- Diane Abbott, MP
01-08-2006-- Keith Vaz, MP
01-08-2006-- Khalid Mahmood, MP
01-08-2006-- Mark Hendrick, MP
01-08-2006-- Marsha Singh, MP
01-08-2006-- Mohammed Sarwar, MP
01-08-2006-- Parmjit Dhanda, MP
01-08-2006-- Sadiq Khan, MP
01-08-2006-- Shahid Malik, MP
01-08-2006-- Shailesh Vara, MP
15-11-2007-- Virendra Sharma, MP

Meanwhile for the population at large (not the MPs) the 2001 census says

White 92.1%Black 2%Indian 1.8% Pakistani 1.3% Mixed race 1.2% Other 1.6%

So the 2008 UK house of commons underrepresents racial minorities by a factor of 3.4 (i.e. 7.9% non-white population vs 2.3% non-white MPs).

Because the UK and Australia have the same percentage of whites (92%) in the population, their Houses are directly comparable. Assuming Malik's counts from 2005 are valid, note that the probability that 150 randomly drawn members of the 659 would be all-white, is

binomial(509, 13)/binomial(659, 13) ≈ 0.0336.
Therefore, the statistical confidence that Australia's IRV election process disfavors racial minorities (compared with the UK's plurality election process) is 97%.

If we redo the calculation using the 2008 data [646 MPs, 15 colored] then we instead find that the probability that 150 randomly drawn members of the 646 would be all-white, is

binomial(496, 15)/binomial(646, 15) ≈ 0.0181.
Based on that, the statistical confidence that Australia's IRV election process disfavors racial minorities (compared with the UK's plurality process) is 98%.

The combined confidence based on both the 2004/2005 and 2007/2008 data sets (I believe, but it has not been verified, that the previous Australian House, elected in October 2004, also was all-white) would be considerably higher.

Takoma Park, Maryland

The city of Takoma Park, Maryland adopted instant runoff voting for city council and mayoral elections in 2006. This was in large part due to Rob Richie, head of the pro-IRV organization "FairVote" who lives in Takoma Park. As of the census of 2000, Takoma Park had 17299 people in 6893 households, and 3949 families. Its racial makeup was

48.79% White, 
33.97% African American, 
4.36% Asian, 
0.44% Native American, 
0.03% Pacific Islander, 
7.44% from other races, and 
4.97% mixed from two or more races. 
Hispanic or Latino of any race were 14.42% of the population.

Since adoption of IRV (I am writing this section in June 2010), every Takoma Park election winner has been white.

Ireland

Ireland has always elected its President via IRV. So far (2010) no member of a racial (or religious?) minority has ever won.

San Francisco, CA

San Francisco started using (top-3 bastardized) instant runoff voting in 2004, and during 2004-2009 held 14 IRV elections. In all 14 cases, the same winner won as with plain-plurality voting (i.e. the IRV "first round"). Therefore, there was no change whatever in racial composition of winners using IRV versus using plain-plurality.

Conclusion

Both the US↔Australia and UK↔Australia contrasts indicate IRV does not favor racial minorities (versus plurality voting). It disfavors them. The (smaller) amount of evidence from Takoma Park, San Francisco, and Ireland also in no way supports the notion IRV favors minorities.

We have made no attempt to find out why IRV disfavors minorities; we merely focused on whether it does. But it seems plausibly logical that IRV would actually help majorities. The intent with IRV (vs plain plurality voting) was to try to help merge what otherwise would be "split" votes.
    E.g. if 3 whites run versus 1 black, and all the white votes merge instead of being split, this would seem to hurt the black relatively.
    Indeed it has been claimed that the reason, or at least a big reason, that (non-instant) runoffs were introduced heavily throughout the USA South in party primaries during early 1900s, was precisely because of the fear by whites that due to a white-candidate vote-split, some black would get nominated by the Democratic party and hence would almost surely win office. Horrors. With the top-2-runoff, that risk was averted. (If this theory was correct it might be possible to find evidence for it in all the old Southern election records.) Anyhow, whether for that or other reasons, blacks were very effectively shut out of political office throughout the 'Jim Crow era' about half a century long in the South.

Caveat (and re-examination of our logic)

In both Australian, USA, and UK society, there are forces – which have nothing to do with the voting system – causing racial minorities and/or women to be underrepresented. However (provided we agree these 3 societies are basically similar today) these forces are the same. Therefore they cancel out and would normally yield the same degrees of underrepresentation in all 3 societies. But they do not. As we've seen from the data, Australia with IRV has worse minority-underrepresentation than the USA and UK. The underlying cause of this difference, given that the 3 societies are similar, would have to be the only way those societies differ, namely, THE VOTING SYSTEM.

Now the flaw in this logic is that actually, Australia, USA, and UK are not exactly the same in every way aside from the voting system. There are other differences that may affect things somehow. Therefore, it is not possible to say for sure the voting system is the cause. It might be some other unknown cause. If instead of having data from only 3 countries, we had data from 100 countries, then we might be a lot more convincing.

However, there is no such databank. We have to be satisfied with the data that actually exists. The USA and UK are the top two plurality-using countries in the world and Australia is the top IRV-using country. No other countries currently come close. And the USA, UK, and Australia are very similar (same language, same ethnic roots mostly, shared culture, many shared aspects of governmental design, and UK & Australia have same percentage of whites in their populations and only separated about 100 years ago).

So given this, we suggest to you that based on the data that we have, and until any evidence comes along to the contrary, you should conclude that IRV disfavors racial minorities. It is irresponsible to state

"Instant Runoff" voting reform can empower communities of color.
[This is the title of a newspaper article by Isaac Peterson III in the Minnesota Spokesman-Recorder dated 7/16/2008. The Spokesman-Recorder says it is "the oldest minority owned business in the state of Minnesota... a newspaper committed to meeting the news and information needs of African Americans." If you look in the article, no evidence is given to support the claim in the headline. There is some irrelevant "evidence" like "The Public Research Institute at San Francisco State University conducted exit polls during two IRV elections in this period, as did other groups like the Asian Law Caucus. In addition, FairVote conducted studies based on precinct analysis of ballots for the 2004 and 2005 elections. The polls of San Francisco voters' opinions and the precinct analysis of turnout and use of rankings in San Francisco demonstrate that:
That all tells us absolutely nothing about whether IRV favors racial minorities. It tells us about other things, some of which sound mildly related, but the direct question of favortism of minorities is not addressed. And in addition some of those points were misleading; for example when Rob Richie, head of FairVote, got his hometown Takoma Park MD to try IRV, turnout was very poor – of the 17,299 total city population, only 1,010 people IRV-voted – and despite Takoma Park's 34% black population, zero minorities hold office as of 2008 – and in SF a large percentage of voters found IRV hard to understand and IRV countries consistently exhibit some of the highest ballot-spoilage rates in the world – including San Francisco had 7× the ballot spoilage in IRV than plurality races – if Peterson had presented that data instead, things would look different.
We repeat. My only actual direct evidence on the topic indicates that IRV disfavors minorities.


Related page on 'ethnic divisions' & IRV

Return to main page