'Electoral commentators sometimes like to play god. Rex Boyd says that ‘a more evenly distributed pattern of voting would ideally be more desirable’, but no specialist can or should specify what pattern of voting is or is not ‘more desirable’. Nor was the 2014 outcome, as so many claim, the product of some brilliant strategy. With 59% of the vote nationally, Fiji First would likely have won the election under any of the best known electoral systems' - Professor Jon Fraenkel
Fijileaks Editor: Someone calling himself or herself "Rex Boyd" sent us the following for publication
Fiji’s 2014 Voting Results – Misplaced Analysis, Faulty Conclusions
By "Rex Boyd"
Dear Sir,
Allow me to make some observations about the voting system and the results from the last election under the current Fiji constitution. There appears to be misplaced analysis and conclusions drawn by you from the last election. In one of your blog pieces you made the following comments in relation to the election of one of the FPP candidates where you said “he is entering Parliament under the d’Hondt Electoral System and under the coat tail of Bainimarama with a derision 895 votes only – a tenth of 8,000 plus visitors to Fijileaks website from inside Fiji everyday of the week, not to mention hundreds globally.
The leader of SODELPA Mrs Kepa expressed similar view when in response to Mr Bainimarama’s UN human rights remarks she said “there is nothing democratic about a system of voting that allows a candidate with 830 votes to enter Parliament and deny 32 others who polled more a seat in the House”.
It appears that people who are looking at the results based on the number of votes a candidate got fail to realise that Fiji’s current voting system is based on proportional representation (PR) which is different from the first-past-the-post (FPP) voting system of the past. Under the PR system the number of seats a party gets is based on the proportion of votes its gets rather than on the number of votes a candidate gets (though individual votes add up to the party’s total). Hence the higher the proportion of votes a party gets (above the 5% threshold) the more seats it will have in parliament vis a vis other parties.
Thus the Fiji First Party with 64% of the total votes cast (revised proportion after excluding the votes of parties/independents that got less than 5%) has 32 seats in parliament, SODELPA with 30% has 15 seats while NFP with 6% has 3 seats. However, if the seats were based on the FPP system where candidates with the largest number of votes would be elected to parliament then the results would be vastly different from this (if we can assume that the votes each candidate got in the last election in some way reflected this FPP system). This can be better explained using Table 1 below.
Fiji’s 2014 Voting Results – Misplaced Analysis, Faulty Conclusions
By "Rex Boyd"
Dear Sir,
Allow me to make some observations about the voting system and the results from the last election under the current Fiji constitution. There appears to be misplaced analysis and conclusions drawn by you from the last election. In one of your blog pieces you made the following comments in relation to the election of one of the FPP candidates where you said “he is entering Parliament under the d’Hondt Electoral System and under the coat tail of Bainimarama with a derision 895 votes only – a tenth of 8,000 plus visitors to Fijileaks website from inside Fiji everyday of the week, not to mention hundreds globally.
The leader of SODELPA Mrs Kepa expressed similar view when in response to Mr Bainimarama’s UN human rights remarks she said “there is nothing democratic about a system of voting that allows a candidate with 830 votes to enter Parliament and deny 32 others who polled more a seat in the House”.
It appears that people who are looking at the results based on the number of votes a candidate got fail to realise that Fiji’s current voting system is based on proportional representation (PR) which is different from the first-past-the-post (FPP) voting system of the past. Under the PR system the number of seats a party gets is based on the proportion of votes its gets rather than on the number of votes a candidate gets (though individual votes add up to the party’s total). Hence the higher the proportion of votes a party gets (above the 5% threshold) the more seats it will have in parliament vis a vis other parties.
Thus the Fiji First Party with 64% of the total votes cast (revised proportion after excluding the votes of parties/independents that got less than 5%) has 32 seats in parliament, SODELPA with 30% has 15 seats while NFP with 6% has 3 seats. However, if the seats were based on the FPP system where candidates with the largest number of votes would be elected to parliament then the results would be vastly different from this (if we can assume that the votes each candidate got in the last election in some way reflected this FPP system). This can be better explained using Table 1 below.
Column 1 in the table lists the main parties that took part in the last election. Column 2 shows the seats the parties hold in the current parliament. Colum 3 shows the percentage of votes each party received at the last election. Column 4 shows the seats each party would have received if results were based on the FPP system. This column is derived by first ranking all the candidates in terms of the votes they received from the highest to the lowest (based on the final results posted on the Elections Fiji website), next selecting the top 50 candidates from this list and finally allocating these candidates to the parties they belonged to. On this basis SODELPA would have got 24 seats (compared to its current 15 seats), Fiji First 21 seats (compared to 32 seats), NFP 2 seats, PDP 2 seats and One Fiji 1 seat. Column 5 shows the corresponding percentage of seats for each party based on column 4 results.
So here lies the punchline. SODELPA with just 28% of the popular votes cast would have had 48% of the seats in parliament while Fiji First with almost 60% of the popular votes would have got just 42% of the seats in parliament. And interestingly enough One Fiji with just 1.2% of the votes cast would have got 1 seat in parliament, while FLP with twice the proportion of One Fiji votes would have got no seats in parliament. If we were to pursue the above arguments to its logical conclusion then SODELPA with the largest number of seats in parliament (under the FPP system) would most likely have been asked to form government.
Most probably it would have cobbled a coalition with the other three parties, rather than with Fiji Fist. This would have given it a total of 29 seats enabling it to form a majority government and rule for the next four years. So here you would have a situation where parties with a combined total of 41% of the total votes cast would have been allowed to form government while a party with almost 60% of the votes would have been deprived of such an opportunity. So one could ask where is the democracy in this?
It may also be important to point out from the above discussion is that SODELPA leader conveniently forgets to mention that some of her party candidates with votes lower than the votes of some candidates from other parties also got into parliament so this problem is not confined to the Fiji First Party alone. Five of SODELPA candidates with votes ranging between 3268-4532 got into parliament ahead of the 4956 votes got by one of the PDP candidate while five of its candidates with votes ranging between 2105-2523 got into parliament ahead of the 2788 votes got by one of the One Fiji Party candidate (source – Elections Fiji website). The reason why SODELPA got ahead of these candidates is because of the PR voting system which the party appears to be deriding. So much for hypocrisy !
Now getting back to the issue of the votes each individual candidate got and who ended up in parliament and who didn’t. Note as mentioned earlier the number of seats a party gets is proportional to the votes it gets and not on the vote each individual gets. Under the present PR system Fiji First with 64% of the votes is rightly entitled to 32 seats in parliament, irrespective of the fact that some of its candidates got less votes than some other candidates who got more but missed out. It may be noted that under the present system there is nothing stopping the entire votes that a party gets goes to one or just a few of its candidates rather than being more evenly spread across all its candidates. For example the entire Fiji First vote could have gone to party leader Bainimarama while all the other party candidates could have got zero votes. The party would still be entitled to 32 seats even though the remaining 31 candidates who enter parliament on behalf of the party got zero votes. If such a situation were to eventuate then this would have given you a field day where you gleefully would have pointed out to the fact that Fiji First candidates with zero votes got into parliament while some other candidates with many, many more votes did not!.
People who compare individual votes are either ignorant of how the current voting system works or are trying to be mischievous. Under the current PR voting system Fiji has chosen the open list system as opposed to the alternative closed list system. Under the open list system the voters are allowed to vote for particular candidates and not just the parties. Under the closed list system (which is practised in several European countries) the party fixes the order in which the candidates are to be elected, the voter casts a vote for the party as a whole and the winning candidates for the party (the number of which is proportional to the votes the party received) are selected in the exact order they appear on the original list. If such an alternative system were adopted in Fiji then this would save some of the winning party candidates the difficult task of defending themselves that they got into parliament with fewer votes than some other candidates from other parties with more votes. And it would also prevent some politicians, journalists, bloggers and academics making unwarranted comparisons and drawing faulty and erroneous conclusions from the results based on individual votes rather than on total votes received by the party.
Note under the PR system voting is not a popularity contest. Although a less skewed and more evenly distributed pattern of voting at the party level would ideally be more desirable there is no such requirement for this under the current PR voting system. The strategy for each party is to maximise the votes it gets which in turn determines the number of seats it gets in parliament. It appears that the Fiji First Party used this strategy to its fullest advantage and reaped its just rewards. Parties and independents that failed to utilise this strategy and failed to see the big picture cannot really blame the system or the winning party. Comparing individual votes is irrelevant.
So here lies the punchline. SODELPA with just 28% of the popular votes cast would have had 48% of the seats in parliament while Fiji First with almost 60% of the popular votes would have got just 42% of the seats in parliament. And interestingly enough One Fiji with just 1.2% of the votes cast would have got 1 seat in parliament, while FLP with twice the proportion of One Fiji votes would have got no seats in parliament. If we were to pursue the above arguments to its logical conclusion then SODELPA with the largest number of seats in parliament (under the FPP system) would most likely have been asked to form government.
Most probably it would have cobbled a coalition with the other three parties, rather than with Fiji Fist. This would have given it a total of 29 seats enabling it to form a majority government and rule for the next four years. So here you would have a situation where parties with a combined total of 41% of the total votes cast would have been allowed to form government while a party with almost 60% of the votes would have been deprived of such an opportunity. So one could ask where is the democracy in this?
It may also be important to point out from the above discussion is that SODELPA leader conveniently forgets to mention that some of her party candidates with votes lower than the votes of some candidates from other parties also got into parliament so this problem is not confined to the Fiji First Party alone. Five of SODELPA candidates with votes ranging between 3268-4532 got into parliament ahead of the 4956 votes got by one of the PDP candidate while five of its candidates with votes ranging between 2105-2523 got into parliament ahead of the 2788 votes got by one of the One Fiji Party candidate (source – Elections Fiji website). The reason why SODELPA got ahead of these candidates is because of the PR voting system which the party appears to be deriding. So much for hypocrisy !
Now getting back to the issue of the votes each individual candidate got and who ended up in parliament and who didn’t. Note as mentioned earlier the number of seats a party gets is proportional to the votes it gets and not on the vote each individual gets. Under the present PR system Fiji First with 64% of the votes is rightly entitled to 32 seats in parliament, irrespective of the fact that some of its candidates got less votes than some other candidates who got more but missed out. It may be noted that under the present system there is nothing stopping the entire votes that a party gets goes to one or just a few of its candidates rather than being more evenly spread across all its candidates. For example the entire Fiji First vote could have gone to party leader Bainimarama while all the other party candidates could have got zero votes. The party would still be entitled to 32 seats even though the remaining 31 candidates who enter parliament on behalf of the party got zero votes. If such a situation were to eventuate then this would have given you a field day where you gleefully would have pointed out to the fact that Fiji First candidates with zero votes got into parliament while some other candidates with many, many more votes did not!.
People who compare individual votes are either ignorant of how the current voting system works or are trying to be mischievous. Under the current PR voting system Fiji has chosen the open list system as opposed to the alternative closed list system. Under the open list system the voters are allowed to vote for particular candidates and not just the parties. Under the closed list system (which is practised in several European countries) the party fixes the order in which the candidates are to be elected, the voter casts a vote for the party as a whole and the winning candidates for the party (the number of which is proportional to the votes the party received) are selected in the exact order they appear on the original list. If such an alternative system were adopted in Fiji then this would save some of the winning party candidates the difficult task of defending themselves that they got into parliament with fewer votes than some other candidates from other parties with more votes. And it would also prevent some politicians, journalists, bloggers and academics making unwarranted comparisons and drawing faulty and erroneous conclusions from the results based on individual votes rather than on total votes received by the party.
Note under the PR system voting is not a popularity contest. Although a less skewed and more evenly distributed pattern of voting at the party level would ideally be more desirable there is no such requirement for this under the current PR voting system. The strategy for each party is to maximise the votes it gets which in turn determines the number of seats it gets in parliament. It appears that the Fiji First Party used this strategy to its fullest advantage and reaped its just rewards. Parties and independents that failed to utilise this strategy and failed to see the big picture cannot really blame the system or the winning party. Comparing individual votes is irrelevant.
Fraenkel Fiji’s 2014 Voting Results – Misplaced Analysis, Faulty Conclusions: A Rejoinder
By Professor Jon Fraenkel
Victoria University of Wellington
Rex Boyd says some sensible things, but the simulation of what would have occurred under a first-past-the-post system is completely bizarre. Did he break down the results by constituency? This would be possible but difficult. We have the electoral data at the micro polling station level. One would have to somehow group the polling stations into constituencies. It would be even more difficult if one were to try to replicate the 1997 constitution's 71 constituencies, and draw boundaries around 46 communal and 25 open constituencies (indeed impossible since we dont know the ethnicity of voters). In fact, what Rex Boyd did was simply to count up the highest vote getters nationwide. That's a very poor guide to what would have happened under a first-past-the-post system. In fact, its no guide at all.
In 2014, around 40% of the nation voted directly for Frank Bainimarama. Under a FPP system, they would have been highly unlikely to do this. Nor would they have been encouraged to do so by Bainimarama himself, as they were during the 2014 campaign. The only reason why Bainimarama did this is because he knew a massive personal vote would enable loads of other FijiFirst candidates to get elected. It is highly misleading, even inflammatory, to suggest that SODELPA would have won the election under a first-past-the-post system. Making such claims does not assist the country to have a sensible discussion about alternative electoral systems, or to appreciate the impact of Fiji’s new system in September 2014.
Rex Boyd is right that under an open list PR system comparison of individual votes for candidates can be misleading. Behind-the-scenes what matters is the summing of individual votes to get party tallies. He is wrong to say 'comparing individual votes is irrelevant'. Once party entitlements have been calculated, the individual tallies are exceptionally important. They decide which candidate is elected within each party.
But Fiji's current system encourages the general public to focus on the individual candidate tallies. It has a ballot paper showing only numbers, which represent the individual candidates. In most countries with open list systems, citizens can a) either cast a party vote, or b) a party vote as well as a candidate vote, or at least c) the candidates are arranged on the ballot paper in columns ordered by political party. Fiji is (as far as I'm aware) unique globally in having a type of ballot paper with the party altogether obscured from view.
Although widely trumpeted as strategically brilliant, this choice of ballot paper format probably arose because of an ad hoc adjustment to the implications of earlier decisions. First, the Attorney-General's office decided to ditch Yash Ghai's proposal for a simpler closed list proportional representation system (where voters just endorse a party, as in South Africa), then they decided not to subdivide the country into three or four constituencies (and to have only a single nationwide constituency, without fully grasping the ramifications of this decision. Only after that did they recognise that this would entail a huge and unwieldy ballot paper with hundreds of names and party symbols. So they tried to simplify this, and came up with the very odd ‘bingo’ or ‘Sudoku’ ballot paper format with only numbers on it. This perhaps explains the protracted delays in early 2014 before the Electoral Decree was finalized and released.
The result was a ‘latent’ list system, meaning by this that the fact that party lists are central to translating candidate votes won into seats for parties is not at all obvious to the voter. That commentators focus on this is thus not at all 'mischevious', as Rex Boyd seems to think.
The electoral system, and the single national constituency, have been built into the constitution. So they cannot be changed by the political parties now in parliament even if they wish to do so. A change would require a constitutional amendment. And constitutional amendments need a three-quarters majority in a referendum.
Electoral commentators sometimes like to play god. Rex Boyd says that ‘a more evenly distributed pattern of voting would ideally be more desirable’, but no specialist can or should specify what pattern of voting is or is not ‘more desirable’. Nor was the 2014 outcome, as so many claim, the product of some brilliant strategy. With 59% of the vote nationally, Fiji First would likely have won the election under any of the best known electoral systems – whether first past the post, closed list or open list proportional representation, or a New Zealand style mixed member system or the Australian style alternative vote system. The idiosyncracies, and potential glitches, of the new system are more likely to show up at the next election if there is a split field and no one party that takes a commanding lead.
By Professor Jon Fraenkel
Victoria University of Wellington
Rex Boyd says some sensible things, but the simulation of what would have occurred under a first-past-the-post system is completely bizarre. Did he break down the results by constituency? This would be possible but difficult. We have the electoral data at the micro polling station level. One would have to somehow group the polling stations into constituencies. It would be even more difficult if one were to try to replicate the 1997 constitution's 71 constituencies, and draw boundaries around 46 communal and 25 open constituencies (indeed impossible since we dont know the ethnicity of voters). In fact, what Rex Boyd did was simply to count up the highest vote getters nationwide. That's a very poor guide to what would have happened under a first-past-the-post system. In fact, its no guide at all.
In 2014, around 40% of the nation voted directly for Frank Bainimarama. Under a FPP system, they would have been highly unlikely to do this. Nor would they have been encouraged to do so by Bainimarama himself, as they were during the 2014 campaign. The only reason why Bainimarama did this is because he knew a massive personal vote would enable loads of other FijiFirst candidates to get elected. It is highly misleading, even inflammatory, to suggest that SODELPA would have won the election under a first-past-the-post system. Making such claims does not assist the country to have a sensible discussion about alternative electoral systems, or to appreciate the impact of Fiji’s new system in September 2014.
Rex Boyd is right that under an open list PR system comparison of individual votes for candidates can be misleading. Behind-the-scenes what matters is the summing of individual votes to get party tallies. He is wrong to say 'comparing individual votes is irrelevant'. Once party entitlements have been calculated, the individual tallies are exceptionally important. They decide which candidate is elected within each party.
But Fiji's current system encourages the general public to focus on the individual candidate tallies. It has a ballot paper showing only numbers, which represent the individual candidates. In most countries with open list systems, citizens can a) either cast a party vote, or b) a party vote as well as a candidate vote, or at least c) the candidates are arranged on the ballot paper in columns ordered by political party. Fiji is (as far as I'm aware) unique globally in having a type of ballot paper with the party altogether obscured from view.
Although widely trumpeted as strategically brilliant, this choice of ballot paper format probably arose because of an ad hoc adjustment to the implications of earlier decisions. First, the Attorney-General's office decided to ditch Yash Ghai's proposal for a simpler closed list proportional representation system (where voters just endorse a party, as in South Africa), then they decided not to subdivide the country into three or four constituencies (and to have only a single nationwide constituency, without fully grasping the ramifications of this decision. Only after that did they recognise that this would entail a huge and unwieldy ballot paper with hundreds of names and party symbols. So they tried to simplify this, and came up with the very odd ‘bingo’ or ‘Sudoku’ ballot paper format with only numbers on it. This perhaps explains the protracted delays in early 2014 before the Electoral Decree was finalized and released.
The result was a ‘latent’ list system, meaning by this that the fact that party lists are central to translating candidate votes won into seats for parties is not at all obvious to the voter. That commentators focus on this is thus not at all 'mischevious', as Rex Boyd seems to think.
The electoral system, and the single national constituency, have been built into the constitution. So they cannot be changed by the political parties now in parliament even if they wish to do so. A change would require a constitutional amendment. And constitutional amendments need a three-quarters majority in a referendum.
Electoral commentators sometimes like to play god. Rex Boyd says that ‘a more evenly distributed pattern of voting would ideally be more desirable’, but no specialist can or should specify what pattern of voting is or is not ‘more desirable’. Nor was the 2014 outcome, as so many claim, the product of some brilliant strategy. With 59% of the vote nationally, Fiji First would likely have won the election under any of the best known electoral systems – whether first past the post, closed list or open list proportional representation, or a New Zealand style mixed member system or the Australian style alternative vote system. The idiosyncracies, and potential glitches, of the new system are more likely to show up at the next election if there is a split field and no one party that takes a commanding lead.
