On Tuesday on this blog, Fabian Richter wrote that only fifteen percent of the population lived in safe seats based on the assumption that a seat is marginal if there is a gap of 5,000 votes between the winning candidate and the runner up. Glenn Gottfried takes issue with this assumption and explains how a distance of 10 per cent between candidates may be more useful.

In our report, Worst of Both Worlds: Why First Past the Post no longer works, Guy Lodge and I examined the arguments in defence of First-Past-the-Post (FPTP) and concluded that the system is no longer fit for contemporary British politics. Part of our analysis focuses on the reduction of marginal seats over time. In an article published on Tuesday on this blog, Fabian Richter attacks our definition of marginal seats then sets out to redefine it in rather perplexing and unconvincing terms.

Richter argues that the report, “strangely”, uses two definitions interchangeably of what constitutes a marginal seat. We use the same definition of a marginal seat, however, and apply it to two separate analyses.

First, we define marginals in simple terms – where the distance between the winning candidate and the runner-up is less than 10 per cent. Richter makes the accusation that this figure was pulled out of thin-air. It is true that there may be some justification in arguing that 10 per cent is a subjective number. After all, is a victory margin of 10.5 percent (a safe seat) really that much safer than one with 9.5 per cent (a marginal seat)? The point of the report was not to redefine a definition used consistently by academics and political commentators alike. If we had chosen to do so there may be no comparative value to any of our claims.

Using the 10 per cent threshold, we explored the justification of FPTP. Proponents of FPTP claim the system guarantees strong single-party rule because it awards the winner an added bonus of seats. A small swing in votes from one party to the other would see a significant number of seats change hands. This assertion rests on two assumptions: first, the same two parties are competing for the most seats and second, these seats must be highly competitive – in other words marginal. “The existence of a winner’s bonus depends on the prevalence of large numbers of marginal seats” (see page 7 of our report). This was certainly the case in the 1950s, the glory days of FPTP, when Labour and the Conservatives received nearly 90 per cent of the total vote.

We show however that over time the inclusion of third parties has decreased the vote share between the two main parties. We provide evidence that the number of marginal seats between Labour and the Conservatives have also decreased over the past 50 years. Richter explains that this analysis is unfounded in an era of third party competition. But this is the very essence of our argument. The fact that Lab-Con marginals are decreasing shows that the claim of FPTP providing a winner’s bonus is seriously being diminished. Strong single-party governments will likely become harder to achieve for both Labour and the Conservatives in future elections.

Richter believes that when deciding what is marginal it should be taken from raw vote counts rather than proportionality. He uses the threshold of 5,000 votes to define this, explaining that in the average constituency of 75,000 with a 65 per cent turnout it would take a 4.8 per cent swing to reduce the winner’s majority to zero. Assuming a uniform swing, the winner would lose roughly 5 per cent while the runner-up gains this 5 per cent. This brings us to the magic number of 10 per cent. So far this makes perfect sense. Richter goes on however to explain that it is the raw vote numbers that matter. 10 per cent in the average constituency equals nearly 5,000 votes.

What is uncertain is why the raw vote count of 5,000 is applied to all constituencies rather than 10 per cent. Electorate sizes and turnout levels greatly vary by constituency so certainly these would be better compared by percentage proportions. Take Na h-Eileanan an Iar in Scotland for example, with the smallest vote count in the 2010 election with 14,717 votes. Here, 5,000 votes would equate to 33 per cent. SNP candidate Angus MacNeil won on 6,723 votes (45.7 per cent) while Labour’s Donald John MacSween finished second with 4,838 (32.9 per cent). Under the traditional definition this seat would be considered safe for the SNP, however under Richter’s definition MacNeil would need to win by a margin larger than Labour’s total votes altogether to consider his seat safe.

On the other end of the spectrum there is the Isle of Wight, the constituency with the largest electorate and nearly 70,000 voters in the last election. In this seat 5,000 votes is approximately 7 per cent – a difference of 26 per cent to Na h-Eileanan an Iar. To go a step further, turnout levels change each election thus one cannot compare marginals to any prior results accurately, even in the same constituency, if not using a proportional comparison.

Richter starts his marginal definition using a good basis, but on further investigation it loses all value and sense of comparability.

Please read our comments policy before commenting.