Many political observers and activists had a keen interest in last weekend’s bye-election in Tonota North. It was inevitable that people would look at the outcome of the elections for clues about the damage caused by the BDP’s split, the level of electoral support for the BMD, and the possibilities for opposition cooperation. It was never clear, however, whether these elite-level considerations would resonate with the voters in Tonota North more than local considerations, such as the local reputation of the candidates, the extent to which each candidate had the support of local party activists, and campaign activities. Since the results of the election reflect all of these things, how can we figure out what the results tell us about the damage caused by the BDP’s split or about the voters’ assessments of the candidates? How can we evaluate the effects of opposition unity?
To make sense of the election results, I analyzed four aspects of the electoral results. First, I looked at changes in the difference in the vote share for the winning party (BDP) and the vote share for the runner-up party (BCP); this is known as the margin of victory. Second, since the margin of victory does not take into consideration the effect of changes in the number of contestants, I also looked at changes in the vote shares for the top two individual parties. Third, since people sometimes spoil their ballot as a form of protest when they do not like any of the parties, I analyzed the spoiled ballots as a share of all ballots casts. Fourth, I evaluated the turnout, both for the election as a whole and by supporters of individual parties.
The analysis involves comparisons of vote shares for the 2010 bye-election with the 2009 and 2004 general elections. Discussions of election results often ignore spoiled ballots and refer to vote shares based on valid votes only. A comparison of votes based on total votes cast, however, makes it possible to assess the effect of any changes in the spoiled ballots. It is also important to compare the valid votes since they are the basis for the election outcome. The comparison of the votes cast for the top two candidates gives an indication of the competitiveness of each election. Therefore, vote shares have been calculated based on (1) the total votes cast, (2) the valid votes cast for all parties, and (3) the votes cast for the top two candidates only. Note that, in 2009, because there were only two candidates, the vote shares based on the valid votes cast and the votes cast for the top two candidates are identical.
The calculations for the 2010 bye-election are based on the vote distribution as reported on the IEC website: BDP = 4575; BCP = 2617; MELS = 62; spoiled = 114; total vote = 7368; valid vote = 7254. The figures for 2009 and 2004 are drawn from the election reports produced by the IEC.
Swing in the margin of victory
The margin of victory refers to the difference in the percentage of the vote received by the winner and the next closest candidate. I have calculated the margin of victory based on (1) the shares of the total vote, (2) shares of the valid votes, and (3) the votes received by top two candidates only. As noted above, there is no difference in the vote shares based on the valid votes and the top two candidates in 2009 because there were only two candidates.
For example, the margin based on the total vote (including spoiled ballots) in 2010 is the vote share for the winner (the BDP with 62.1%) minus the vote share for the runner-up (the BCP with 35.5%) or 26.6%.
The “swing” refers to the change in margin of victory from one election to the next if calculated in the same manner. A positive swing indicates that the margin of victory increased and the constituency has become less competitive. A negative margin swing indicates that the margin of victory decreased and the constituency has become more competitive.
For example, the change in the margin of victory in the total vote between 2010 and 2009 is the margin of victory in 2010 (26.6%) minus the margin of victory in 2009 (30.3%) or – 3.8%, which means that the margin of victory narrowed by 3.8%. (Sometimes, as in this case, the figures seem slightly off as a result of rounding errors. I calculated all of these figures using excel).
Whether one compares the margin of victory based on the total vote, shares of the valid vote, or only the votes for the top two candidates, it is clear that a rather substantial margin swing occurred between 2004 and 2009. There was a much smaller – but noteworthy – swing between the 2009 general election and the 2010 bye-election.
Change in the vote share of individual parties
Changes in the margin of victory can be somewhat misleading when the number of candidates changes from one election to the next, as in Tonota North. When the number of candidates changes, it is possible for the margin of victory to shrink or expand in part as a result of changes in the division of votes between the second place candidate and any other candidates. For instance, a party might win two elections with 65% of the vote, but face one candidate who received 35% of the vote in one election and two candidates who received 20% and 15% of the vote in the other election. The margin of victory will be narrower in the first election with no change in support for the winning party.
The tables below present changes in the percent share of each party from one election to the next based on (1) total votes and (2) valid votes. Negative figures indicate that a party’s vote share is decreasing and positive figures indicate that a party’s vote share is increasing.
For example, the swing in the BDP share of the total vote from 2010 to 2009 is the BDP vote share in 2010 (62.1%) minus the BDP vote share in 2009 (64.2%) or -2.1%.
The BDP lost more support in terms of the total vote between 2009 and 2010 than between 2009 and 2004. It lost slightly more support in terms of the valid vote between 2009 and 2004 than between 2009 and 2010. This difference suggests that something related to the spoiled ballots changed between 2004 and 2009. Indeed, the percentage of spoiled ballots fell by more than half between 2004 (3.8%) and 2009 (1.5%), and there was only a slight change between 2009 and 2010 (a drop from 1.9% to 1.5%).
The BCP experienced most of its growth in support between 2004 and 2009. The increase in the BCP’s vote share from 2004 to 2009 is much greater than the loss of support experienced by the BDP over the same period. It is somewhat greater than the vote share received by the BPP in 2004. This suggests that the BCP reaped the benefits of opposition consolidation in 2009. The analysis also suggests that the BCP benefited from the decline in spoiled ballots between 2004 and 2009 and the slight erosion in the BDP’s support. The increase in support for the BCP from 2009 to 2010 is somewhat lower than the loss in support experienced by the BDP over the same period. The difference between the decline in the BDP’s vote share and the growth in the BCP’s vote share is 0.5% of the vote.
Spoiled ballots
Changes in spoiled ballots as a percentage of total ballots cast may be an indication of protest voting, in which voters intentionally spoil their ballot to indicate their dissatisfaction with the available options. Differences in the percentage of spoiled ballots can also reflect differences ÔÇô legitimate or otherwise – in the procedures for deciding that a ballot is spoiled. The percentage of spoiled ballots cast in the bye-election (1.5%) was somewhat lower than in the general election (1.9%); the difference is 0.3%. These figures compare favorably with the average percentage of spoiled ballots (for all constituencies) in the 2009 election (1.9%) and the 2004 election (2.1%). (The difference in the averages for 2004 and 2009 is not statistically significant). The percentage of spoiled ballots in 2004 (3.8%) was unusually high.
Turnout
Voter turnout as a percentage of registered voters is usually lower for bye-elections than for general elections. The table below compares turnout for four recent parliamentary bye-elections with turnout for the same constituencies in the 2009 and 2004 general elections.
The table reveals quite a lot of variation in turnout rates. A turnout of 58.1% with a decline in the vote of only 13.2% from the previous general election is not unheard of (compare Kgalagadi North), but is quite good for a bye-election. Before the bye-election, I had speculated that rates of abstention might be higher among BDP supporters given the disputes following bulela ditswe and the high stakes for the BMD as it mobilized to demonstrate its electoral support. The BDP also mobilized in an effort to defend its vote share, however, and demonstrate that the BMD had little effect. There was an obvious outpouring of resources by all parties. As in the general elections, however, the BDP enjoyed an advantage compared with the opposition parties in terms of financing and the ability to tap state resources. According to all reports, the BDP invested massively in this campaign.
To get a sense of which party was most successful in turning out its supporters, I calculated the decline in the absolute number of votes received by each party as a percentage of that party’s total votes in 2009. For comparison, I also include the votes for MELS and the decline in the number of valid votes. (In this chart, positive numbers indicate declines and negative numbers indicate gains).
The BDP experienced a loss of 21.3% of its total number of votes between 2009 and 2010; the BCP lost 14.7% of its total number of voters between 2009 and 2010. These figures can be compared with the 18.3% overall decrease in the total number of valid votes. Thus, while both parties saw their total vote decline, the BCP was more successful in retaining its voters than was the BDP. Although this is more speculative, the figures suggest that the people who voted for MELS were drawn disproportionately from past BDP supporters.
The relatively stronger turnout by BCP supporters is all the more impressive in light of the advantages the BDP enjoys in terms of resources and the party’s massive investment in the campaign. It seems that the mobilization effort by the combined opposition did pay off in terms of getting supporters to the polls.
Synopsis and concluding observations
I had expected about a swing in the vote of about 5% from the change in the BDP’s candidate and the fall-out from the BDP primary elections. I would have interpreted a swing of more than 7% as convincing evidence that the mobilization effort by the other opposition parties had weakened support for the BDP. The incumbent was known to be ill prior to the 2009 elections, however, and it is possible that voting behavior had already changed as a result between the 2004 and 2009 elections.
There was a swing in the margin of victory of 3.9% between the 2009 general election and the 2010 bye-election for Tonota North. The swing arises from a 2.4% decline in the BDP’s vote share and a 1.5% increase in the BCP’s vote share. These figures are based on the valid votes. There was a 3.8% swing in the margin of victory based on the total vote, with a 2.1% decline in the BDP’s vote share and a 1.6% increase in the BCP’s vote share.
These figures and the analysis of the changes in turnout suggest that the BCP did not benefit fully from the decline in BDP support. Had the decline in BDP support been completely balanced by an increase in BCP support (i.e., if the BDP lost 2.4% and the BCP gained 2.4%), then the swing in the margin of victory based on the valid votes would have been 4.8%. This is very close to what I had predicted as the likely effect of the change in BDP candidates and disgruntlement related to bulela ditswe.
There was a much bigger swing between 2004 and 2009 that resulted in a considerable increase in the competitiveness of the constituency. One might think that the earlier swing reflected a change in voting behavior in response to the incumbent’s illness. The analysis suggests, however, that the 2004 to 2009 swing mostly reflects a consolidation of the opposition vote behind the BCP. The decline in the BDP’s vote share during this period is so modest that it probably does not reflect a change in voting behavior in response to the illness of the incumbent.
The analysis of the overall turnout of voters in 2009 and 2010 and changes in the total number of votes received by each party suggests that the BDP lost voters mostly to protest votes for MELS or to abstention; it seems that only a few former BDP voters chose to vote for the BCP. On the other hand, these data suggest that the BCP, with the assistance of the combined opposition, was more successful than the BDP in getting its voters to the polls.
The swing away from the BDP is slightly less than the 5% that I had expected based on the change in candidate and the disputes following the BDP’s bulela ditswe contest. The limited swing reflects (1) the BDP’s success in turning out the vote relative to many other bye-elections and (2) the relatively limited extent to which the BCP gained from the BDP’s losses. It also suggests that the BDP’s split and the mobilization of the BMD had little effect on support for the BDP in this rural constituency. Relative to pre-election expectations, these elections results may seem more comforting for the BDP than for the other parties. Nonetheless, the BDP’s vote share has declined and the margin of victory has narrowed somewhat. Despite the BDP’s investment in the constituency, the BCP was even more successful in getting its supporters to the polls. The support of the other opposition parties may have contributed to the BCP’s success in this regard. The results of this bye-election suggest that all of the parties have things to address as they look towards the future.
*Amy Poteete is a Professor at the Concordia University
