If the formation of the BMD, serious efforts at opposition cooperation, and the mobilisation leading up to the BOFEPUSO strike of 2011 energised the opposition, the opposition has been demoralised by events since mid-2011. More recently, at the end of 2013, primary elections generated particularly serious challenges for the BDP. All of these developments can be expected to influence electoral behaviour, including turnout and the relative support for various parties. At the same time, given the constituency based electoral system, characteristics of the candidates and local issues also influence voting behaviour.
Since 2009, by-elections have taken place in four parliamentary constituencies and 15 council wards. Although the dynamics surrounding by-elections and general elections differ, by-elections may give some indication of emergent patterns, if there are any. A couple of caveats are in order, however. First, the BDP's exclusion from the Francistown West by-election limits its value for evaluating trends, since the BDP can be expected to field a candidate in the general elections later in 2014. Second, the other by-elections took place too early to have been affected by controversies surrounding the primary elections for the 2014 elections but can be expected to reflect other changes in the party system.To evaluate changes in vote shares since 2009, I made several calculations. First, I calculated the vote shares and average changes in vote shares since 2009 for all by-elections, both parliamentary and local. Because the dynamics of local and parliamentary elections may differ, I calculated the vote shares and average changes in vote shares since 2009 in the four constituencies affected by parliamentary by-elections and the 15 wards affected by council by-elections. All figures for parliamentary by-elections exclude the data for Francistown West, since the absence of the BDP in that by-election is clearly atypical. Second, I calculated the average vote share per constituency and per ward in 2009. These figures differ slightly from the overall vote shares for each type of election, which (third) I also calculated. To facilitate comparison over time, the figures for the BCP include votes for BAM in 2009. Likewise, the figures for the UDC include votes for the BNF and BPP in 2009. BDP61.11%: By-elections parliamentary: Average vote share (excluding FTW, average change from 2009 of -1.4%)51.49%: By-elections council: Average vote share (average change from 2009 of +3.28 percent)53.09%: By-elections all: average vote share (excluding FTW, average change from 2009 of +2.50 percent) 59.41%: 2009 parliamentary: average vote share where by-election later held48.20%: 2009 council: average vote share where by-election later held50.59%: 2009 all wards and constituencies where by-election later held 53.26%: 2009 parliamentary: overall vote share (all constituencies)54.21%: 2009 parliamentary: average vote share (by constituency, all constituencies)50.07%: 2009 council: overall vote share (all wards)51.50%: 2009 council: average vote share (by ward, all wards)
