Whenever I turn my thoughts to election speculation, (which is more often than is perhaps healthy), I typically assume that an MSP retiring in a certain seat will mean that the constituency is more ripe for a challenging PPC from a rival party at the next election. I have belatedly decided that this somewhat lazy assumption needs further scrutiny and, with Holyrood 2011 (as ever) in mind, I started to crunch some numbers and put the hypothesis to the test.
Of the 11 constituencies that have seen retiring MSPs during the 1999-2007 period, 10 of them resulted in the party reducing its share of the vote at the next election. The 1 constituency that saw an increase in the party’s share of the vote after one of its MSPs retired was Banff & Buchan when Alex Salmond chose Westminster over Holyrood.
The average decrease in vote-share when an MSP retires is 6.1%. The median decrease is 7.4%.
Were we to adopt the hypothesis that a party is as likely to increase its share of the vote as decrease when an incumbent retires (against an alternative hypotheis that the share of the vote should decrease), then the probability of 10 (or more) out of 11 instances all decreasing is as follows:
P (X > 10) = (0.5)^10 = 0.049%
Pretty conclusive then – at a 5%, 1% and even a 0.1% level of significance, my ‘lazy’ assumption that retiring MSPs are more likely to see their vote-share go down seems to hold true.
Of course, perhaps most seats see the incumbent lose votes and perhaps the above is tainted by the fact that the sample includes mostly Labour MSPs who will generally have seen vote shares decrease due to an increased public appetite for a new Government.

Of the 11 constituencies with retiring MSPs, 10 saw decreases in vote-share over and above any decrease for the party nationwide. This increases to 13 out of 14 when the deaths and forced retiral of Donald Dewar, Margaret Ewing and Lord Watson are included.     
So let’s delve into the detail a bit more to see what’s happening:
When looking at individual parties, there seems to be little to suggest that there is little variation across the board. (Note that the Conservatives have not yet experienced a retiring FPTP MSP so have no data to provide from a strictly Scottish Parliament population)
The retiring MSP is an event that has hitherto affected Labour much more than any other party simply because Labour holds considerably more First Past the Post seats. In each of the nine instances where an MSP has not contested the next election (including retirements, deaths and ‘other’), the party share of the vote has decreased more than the national average. The detriment for these constituencies was on average 3.7% at the 2007 election and 6.2% at the 2003 election.
Looking at individual instances, from 2003 to 2007 – vote share decreased by 5.5% on average across all 73 constituencies for Labour.
Retiring MSPs include:
Edinburgh East (Susan Deacon) – 14.8% decrease
East Lothian (John Home Robertson) – 10.4% decrease
Glasgow Rutherglen (Janis Hughes) – 7.5% decrease
Dundee West (Kate MacLean) – 7.0% decrease
Glasgow Cathcart (Lord Watson) – 6.2% decrease
So, as in the final case, even convicted politicians are popular enough to ensure that their retiral (not to be confused with retrial!) results in a more adverse movement in vote share than the national average. This, of course, is as it should be as we would hope that the known sum contribution of our politicians’ previous efforts outweighs the unknown potential of whoever is next in line.
Note that, again for Labour, the same trend was evident from 1999 to 2003 with the following retirements resulting in voteshare decreases that were below the party’s nationwide average (of -0.7%):
Fife Central (Henry McLeish) – 9.1% decrease
Strathkelvin & Bearsden (Sam Galbraith) – 7.9% decrease
Ayr (Ian Welsh) – 1.4% decrease
The Lib Dems support this trend, albeit with only two retirements to work with:
Jim Wallace’s former constituency of Orkney saw a 3.3% drop in vote share for the Lib Dems in 2007, relative to 2003, compared to a nationwide movement of only -0.3%. 
 Ian Jenkins’ former constituency of Tweeddale, Ettrick and Lauderdale saw a 7.4% drop in vote share compared to a 1999-2003 nationwide movement of +1.8%, the biggest drop for the Lib Dems across Scotland except for, interestingly, Jim Wallace’s seat of Orkney which decreased by 15.8% (the largest single decrease in any constituency for any of the parties).
 This helps highlight that the largest movements in any given constituency at any given election will not necessarily be caused by a retiring incumbent.
Airdrie & Shotts witnessed a massive 33.5% swing from Labour to the SNP between 2003 and 2007, despite Karen Whitefield being the sitting MSP in each contest. The individual change in vote-share was the largest deviation from the average for each of the two parties and is an example that may highlight what a change in challenger can potentially do to the statistics.
Ross Finnie saw his share of the vote sink by 14.2% from 2003 to 2007, 13.9% of which went to the SNP. The Labour incumbent was largely untroubled by these significant movements, holding onto the seat with relative ease.
The Lib Dems saw double digit increases in their share of the vote in Aberdeen South, Edinburgh South, Ross, Skye & Inverness West, Strathkelvin & Bearsden and West Aberdeenshire at the 2003 election. This was presumably less to do with incumbent MSPs and more to do with the targeting of the party’s resources. Needless to say, each of these five movements were way above the 1.8% increase in vote share at a national level.
So, moving back to retiring MSPs, what does this mean for next May then?
Well, boundary changes may well dilute or accentuate the occurrence but retiring MSPs include:
Jamie Stone (Caithness, Sutherland & Easter Ross) (majority 10.9% over the SNP)
John Farquhar Munro (Ross, Skye & Inverness West) (majority 12.1% over the SNP)
Cathy Jamieson (Carrick, Cumnock & Doon Valley) (majority 11.8% over the SNP)
Margaret Curran (Glasgow Shettleston) (majority 20.0% over the SNP)
Jack McConnell (Motherwell & Wishaw) (majority 26.1% over the SNP)
Labour’s existing majorities in the above seats are probably too big for a retiring MSP and the ‘hit’ of 4-7% to make much of a difference. Indeed, there is a fair chance that the reversal of Labour’s fortunes since 2007 are such that an increase in the party’s general popularity will neatly net off against the downside of a retiring MSP to result in more or less ‘nil gain, nil loss’ from the last result.
However, for the Liberal Democrats, and the considerable woes that they face, that lack of incumbency could make all the difference to their hopes of clinging on in the two seats that they currently hold.

The conclusion, if any such thing can be reached from the above, is that with numbers expected to be tight in the electoral arithmetic come May 2011, it may end up being who is stepping down and where, rather than who is standing, that makes all the difference…