Posted on

Safe seats revisited (II)

Peter W. Wickham

Safe seats revisited (II)

Social Share

Five of the main factors impacting on seat safety are summarized in the appended table and the most important of these relates to historic support in the constituency.
In this regard, historic is assessed based on the previous election and not an average of a series of elections.  
This represents a modification of the CADRES approach which benefited from the critique provided by the late Wendell A.W.A. McClean. He argued (and I belatedly agreed) that the most relevant aspect of a constituency’s history is reflected in the previous election, therefore, reference to several previous elections can be distracting and misleading.
It is challenging to identify a precise level of support that triggers safety. However, the range appears to be the mid to upper 60 per cent region which means that a candidate who achieves more than 65 per cent in any election is likely to be “safe” in the next one. If one looks at the comparative cases of St Michael West Central versus Christ Church West (2008), it supports this contention since the Barbados Labour Party (BLP) went into the former in 2008 with 61 per cent and failed to retain the seat, while the latter’s support level was 65 per cent and remained with the BLP.
Over the years, commentators have identified “electoral surprises” in the shape of Evelyn Greaves’ loss in 1981, which occurred in a year that the Democratic Labour Party (DLP) enjoyed a favourable swing, and Branford Taitt’s loss in 1999 which previously appeared to have been impossible, along with Sir Harold St John’s defeat as a prime minister in 1986.
No surprise defeats
In reality, none of these defeats was a “surprise” if one uses this approach to seat-safety since Greaves went into 1981 with 51 per cent of the popular vote, while Taitt went into 1999 with 56 per cent and St John faced the polls in 1986 with 58 per cent.  Although the personalities of these candidates influenced the perception that they were “unbeatable”, the objective reality is that none of these seats was realistically “safe” at the time they were lost.
Related to the previous point is the extent to which the candidate performed well in the previous election, but in this instance, the perspective relates to “improvement” and not simple party support. The typical CADRES swing analysis is employed here which suggests that a candidate who performs at least as well as his party in the previous election can be said to be a good candidate and one who is more likely to be “safe” in the next election.  
It is here that Sir Branford’s case presents itself as a better example since the keys to his lack of safety rest in an analysis of his performance in 1994. On that occasion, his personal swing was – 12.6 per cent while the DLP lost 12 per cent on average. This performance was not a bad one, but it does not place Sir Branford among the DLP’s top performers in 1994. With the reduction in his support to 56 per cent, it meant that this seat was no longer safe as the 1999 election demonstrated.
Reference to the previous election’s performance could also have helped to demonstrate the extent to which St Lucy was also a safe seat in 1999 since the 1994 performance of Denis Kellman in that seat was in reality quite remarkable. On that occasion, Kellman actually improved his performance by two per cent while his party lost 12 per cent. It should be noted that he did this notwithstanding the distraction of a National Democratic Party candidate there in 1994.
Upon reflection, this is perhaps the strongest evidence that St Lucy would have been a safe seat in 1999 and, moreover, that seat safety can often be related more to the candidate than the party.
The next factor also highlights the role of the personality in politics and suggests that the safety of a seat can be impacted by the change of candidate in the incumbent party, or the change of candidate in the challenging party. In the same way it is observable that rural candidates tend to be more stable than urban ones, it would also appear that once a candidate is elected, that person’s stability is enhanced once they continue to contest the seat.
Similarly, where that candidate leaves the party, one cannot assume that the next candidate starts with a similar level of support. Correspondingly, if the opposing party’s candidate is changed, it makes the incumbent’s life easier.
It is probably best to demonstrate the impact of this phenomenon in St Michael North East, which has become safe for the BLP’s candidate in more recent times. No attempt is being made to discount the impact of Mia Mottley’s individual candidacy, but the candidate’s success has been “helped” considerably by the instability of the DLP’s candidate presentation there.
Unfavourable swing
Initially, Mottley lost that race as a new candidate in 1991 to a person considered to be a DLP “lightweight” who did not perform well in that election. If one were to balance the pro-BLP factors of a favourable national swing added to the Mottley factor, against the anti-DLP factors of an unfavourable swing, along with a candidate with little national appeal whose individual performance left much to be desired, it would become clear that his incumbency was perhaps the deciding factor.
Interestingly, in the next election, Mottley won over a DLP newcomer (Leroy McClean) and here also it could be argued that her “incumbency” was a major factor. In the next election (1999), both candidates were familiar to the voters there since Mottley faced McClean a second time; so it was arguable that there was no “incumbency” advantage and the result was one of the most resounding defeats ever recorded in Barbados with 80 per cent of the votes cast and a 20 per cent individual swing.  
McClean did not return in 2003, and as such, the DLP’s candidate was new (again) and this allowed Mottley to increase her majority, albeit marginally. However, a “new” candidate (Patricia Inniss) was able to achieve a personal swing of 13 per cent, which was better than the DLP’s national performance and resulted in the first ever reduction of Mottley’s support since she entered politics in 1991.
The lesson in all of this is that incumbency can help, but is by no means the deciding factor. The change of candidates in 1994 and 2003 appeared to have helped Mottley, while the DLP’s performance in 1999 with the same candidate did not mitigate against an outstanding performance for Mottley; nor did the change in 2008 prevent the DLP’s Inniss from registering the best DLP performance there since 1991.
The national support or national performance of the political party seems an obvious contributor to the safety of any seat. The party’s performance is assessed based on the likely support level and likely swing, both of which CADRES can easily predict.
Moreover, the CADRES swing analysis tool, which was developed after 1999, can effectively determine which party is more likely to have safe seats and which of the losing party’s seats are likely to be safe based on the highest possible individual swing.
Since it has already been established that seats become safe when they attain a support level in the high 60s, we can correspondingly assume that seats in the mid to low 60s become vulnerable in a political environment where there is an unfavourable national swing that is sufficiently high to make individual races achievable. Specifically, the St James South seat held by Liz Thompson in 2008 would perhaps have been safe with 61 per cent of the popular support, were it not for the fact that the prevailing swing in 2008 was nine per cent in favour of the DLP.
Certainly, Donville Inniss’ outstanding 14 per cent personal swing was due largely to his personal effort. However, the prevailing environment made his opponent unsafe at that time, while DLP seats with smaller majorities like Christ Church East Central (DLP 53 per cent) remained “safe”.
The final factor is perhaps the most important, but one which is often not within the control of CADRES. The poll of public opinion either nationally or within the constituency can give an accurate reflection of the political opinion “on the ground” that would directly indicate whether a seat is safe or not. Although CADRES has found national polls to be more reliable, it is also true that CADRES has seldom been commissioned to execute constituency polls.
Candidates often assume that the data attributable to constituencies in national polls is indicative, but we have repeatedly argued that this is a dangerous misuse of data. It can therefore be argued that where constituency polls are available, then these data would be the final major factor which could be employed in the effort to identify a safe seat.
• Peter W. Wickham ([email protected]) is a political consultant and a director of Caribbean Development Research Services (CADRES).