[Election-Methods] Do strategic voters stay home on election day?

James Gilmour jgilmour at globalnet.co.uk
Fri Jan 11 09:41:23 PST 2008

Steve Eppley  > Sent: 10 January 2008 14:52
> In elections where strategically-minded potential voters believe the
> odds are extremely low that their vote will be pivotal, do most of them 
> conclude their best strategy is to not spend any time voting or even 
> thinking about how to vote?

There is good evidence from UK General Elections (House of Commons in the Westminster Parliament) that turnout is associated with
marginality, i.e. how close the vote was between the FPTP winner and the second-placed candidate (reported in the UK as "winner's
majority", though it is usually anything but a majority!!).  One interpretation of this is that when voters know in advance that
their votes are unlikely to change the outcome (based on past elections and current opinion polls), they stay at home.  

For example, in the 2001 General Election (data conveniently to hand), the turnout in the 100 "safest" seats (largest winners'
margins) averaged 53%, while in the 100 "most marginal" seats (smallest winners' margins) turnout averaged 63%.  Taken over all of
the 641 Great Britain constituencies (electoral districts), the negative regression of 'percentage turnout' on 'winner's margin as a
percentage of total constituency vote' was very highly significant (P <0.001).

Repeated analyses have shown that indices of "social class" are the strongest predictors of turnout in UK elections, and there is
very strong correlation between "social class" and political party support.  Thus turnout is lower in "typical" Labour Party
constituencies and higher in "typical" Conservative Party constituencies.  In the 2001 GE, the average turnout in 166 constituencies
won by Conservatives was 63%; the average turnout in 412 constituencies won by Labour was 57%.    Even allowing for the effect of
the winning political party, the regression relationship between 'percentage turnout' and 'winner's margin as percentage of total
constituency vote' remained very highly significant (P <0.001).

I am doubtful whether the data from FPTP elections in the USA could be interpreted in a similar simple way because the patterns of
voting in FPTP elections in the USA (e.g. for the House of Representatives) are very different from those in FPTP elections in the
UK and Canada (among many others).  For example, in the 2004 elections for the USA House of Representatives, taking the 370
districts that were contested by both Democrats and Republicans and usually one or more other candidates, 64% of those who voted
cast their votes for a winning candidate.  The comparable figure for typical FPTP elections in the UK and Canada (at all levels of
government) would be around 50%, sometimes just above, sometimes just below.  In the UK 2001 GE quoted, it was 51%.  For an example
from Canada (with 49.8%), see:      

It has been suggested to me in the past by a US contact that the results from primaries in the USA lead significant numbers of those
on the likely losing side to stay at home when the GE comes round, but I suspect effective incumbent gerrymandering in the USA may
have a lot to do with producing these unusual FPTP results.  So I would hesitate to interpret the USA election data in the same way
as the UK data.

Another difference between the USA elections results and those of FPTP elections elsewhere is shown in the proportion of seats won
on a minority of the votes cast in the district.    Taking the same 370 districts in the 2004 HoR elections, only 10 seats (2.7%)
were won with less than 50% of the votes in the respective districts.  In the UK 2001 GE, 320 of the 641 seats (49.9%) were won on
minority votes!  Again, I would suspect highly effective incumbent gerrymandering was involved in the USA.

James Gilmour

No virus found in this outgoing message.
Checked by AVG Free Edition. 
Version: 7.5.516 / Virus Database: 269.19.0/1218 - Release Date: 10/01/2008 13:32

More information about the Election-Methods mailing list