B.C. Referendum — STV election results in a Northern Ireland election
I had a look at the results for the 2007 election to the Northern Ireland Assembly. There were 18 electoral districts, each electing six candidates. The list of those elected as well as full results of the count (in both spreadsheet or presentation format) and various election statistics are all on the web site of the Electoral Office for Northern Ireland.

I discovered that in 13 of 18 districts, the six candidates elected were the ones with the six highest numbers of first-preference votes.

In the other 5 districts, here’s how the six elected candidates ranked in terms of their first-preference votes:

• 1, 2, 3, 4, 5, 7 (Belfast West)


• 1, 2, 3, 4, 6, 7 (Fermanagh and South Tyrone)


• 1, 2, 3, 4, 5, 7 (Foyle)


• 1, 2, 3, 4, 5, 7 (South Antrim)


• 1, 2, 3, 4, 5, 8 (Upper Bann)

Thus, in this real-world example, STV produced results that were very close to what you would get by looking only at the first preferences of voters.

In all 18 districts, the top four vote-getters (on first preference) obtained seats, and in 17 out of 18, the fifth vote-getter (on first preference) did as well.

Now, what about those districts where #7 or #8 got elected but #5 or #6 did not? What is the reason for that, and is it justified?

The reason is this: although the candidate might have been #7 in first-preference votes, there were sufficient numbers of voters who listed that candidate as their second or third preference, thus allowing the #7 candidate to overcome the gap and beat #6. This might well happen when #7 and #8 belong to the same party. You would expect that most voters whose first choice was #7 would have put #8 as their second choice, and vice-versa. The first-preference vote for that party was split between #7 and #8, which is what allowed #6 to finish ahead of both of them. The STV counting system would then aggregate those votes, allowing a candidate from that party to get elected ahead of #6.