The first large-scale election observation was held in 1980, at the independence elections in Zimbabwe. Since then, election observation has become a regular feature all over the world and many international organisations, official agencies, as well as non-governmental organisations, field observation teams. They all use similar methodologies, largely derived or developed from the original 1980 model. A third of a century later, has the use of electoral observation outlived its usefulness – or is it, itself, being used to mask forms of electoral cheating? This paper looks at, among others, four 21st century African elections in Kenya (2007), Zimbabwe (2008 and 2013) and Zambia (2016), and does so through the lens and reflections of one of the pioneers of observation from the 1980 prototype.


I was a member of the Commonwealth Secretariat in the 1970s and 1980s. Under its Secretary-General, Shridath Ramphal, it played a major diplomatic and advisory role leading to the Lancaster House negotiations in late 1979, which established the preconditions for Zimbabwe’s independence after years of war between Rhodesian white settlers and two guerrilla armies who sought majority rule.1 Those preconditions included an election, free campaigning by those previously outlawed, and a truce. Ramphal had inserted into the final Lancaster House agreement a provision that the elections and the campaign were to be observed independently of the warring parties and the British government that temporarily assumed responsibility for the country.2 As the second part of January 1980 dawned, just over a week before Commonwealth observers were to be despatched, a small party from the Commonwealth Secretariat was sent to Rhodesia with the brief both to establish a headquarters for observation, and to work out whether and how observation could be conducted. There were no detailed instructions. Electoral observation had not been attempted before, certainly not on this scale, Rhodesia being about three quarters the size of California.

Two of us, Peter Snelson and I, were all there were who conducted a rapid reconnaissance of the country in a single week, and our report formed the only field input for the plan then devised by Ramphal’s representative, Moni Malhoutra, for observation and the deployment of observers. Both in this first week, and in the weeks of observation that followed, with barely a thought for the health and safety of the observers in a war-torn land, we – the organisers – had no advanced idea of what we were doing. We were involved in an extemporisation under hazardous conditions. The extemporisation assumed a pattern of work, it got very lucky, and did the job – in the end, I think, very well.3 Since then, until 2010, I attended many elections in Africa and elsewhere, but always as a solo observer. In 2010, I was invited to lead one of the teams of an electoral observation group in the Sudanese elections which were part of the prelude to South Sudanese independence a year later. The Carter Centre, the African Union, and the Arab League sent observer groups as well. I was amazed that all and sundry used methodologies essentially no different to those we extemporised in 1980. They had become bureaucratically more robust (or fussy), but they had acquired no greater depth or forensic capacity. When, afterwards, acting as a solo observer in the 2013 Zimbabwean elections and 2016 Zambian elections, and watching the observer groups as well as the elections very closely, I concluded that electoral observation, while not worthless, has limited worth.


The success of the 1980 Zimbabwe observation was almost immediately shaded by a less successful observation of the elections in Uganda later that year. Emboldened by its Zimbabwean success, the Commonwealth Secretariat sent an observation group to an election with very different conditions.4 After the fall of Idi Amin in 1979, facilitated by the Tanzanian army as well as Ugandan rebel irregulars, unsuccessful interim governments were formed under Yusuf Lule (April–June 1979), himself a former Commonwealth Assistant Secretary-General; Godfrey Binaisa (June 1979–May 1980); Paulo Muwanga, who served for ten days in May before a presidential commission took over, ending with the elections of December 1980 which were won by the man Amin had overthrown in 1971, Milton Obote – who had taken exile in neighbouring Tanzania.

But Tanzania, who had overthrown Amin, was a key power-broker in Uganda. It wanted stability in its neighbour and Obote was the preferred choice of Tanzanian President Julius Nyerere. That in itself was unavoidable, given the terrible diplomacy and administration of Amin, but the 1980 elections were certainly unfair, and to a greater or lesser extent rigged. Certainly, afterwards, one of the Commonwealth Secretariat members present accepted they were rigged.5 The electoral commission did not show impartial background. The country had insufficient public administrative infrastructure to hold a nationwide election.6 The Commonwealth Observer Group, in validating it despite the misgivings of its own chairman7, was described as ‘shameful’ by the Economist, although a critical but more measured academic response suggested the Commonwealth sought to maintain some means of influencing Obote in his future conduct as president.8 Figures like the current Ugandan President, Yoweri Museveni, immediately took to the bush to wage guerrilla war.

The differences with Zimbabwe was that, under Amin, Uganda lacked not only public administration but effective government; Rhodesia always had a government and the strategy of the guerrillas was to deny it governmental space, but large parts of the country remained under discernible government. The United Kingdom assumed responsibility for the tasks of government in the interregnum between Rhodesia and Zimbabwe, but was anxious to be rid of the whole affair, albeit without losing face or causing domestic backlash back home, as many British citizens were related to the white settlers. Even so, the UK had no strategic interests in Zimbabwe; whereas Tanzania had every strategic interest. The Tanzanians had invaded Uganda only after Amin had first invaded Tanzania. The Tanzanians wanted a guarantee that this could not happen again and wanted a reliable and familiar man in charge. Whereas a polyglot Commonwealth military force went to Zimbabwe to help keep the peace during the electoral period – British, New Zealand, Kenyan and Fijian troops – peacekeeping in Uganda was entirely conducted by the Tanzanian army, and it was not only determined to do so but was under a central command with directive capacities in many parts of the country.

Finally, by and large, the opposing forces kept the truce in Zimbabwe, but disparate armed groups maintained hostilities at various levels throughout the Ugandan elections. Observation required Tanzanian security and the Commonwealth Observer Group was nowhere as intrepid and determinedly independent as its predecessor in Zimbabwe. Basically, it had less capacity to observe. Tanzania was a Commonwealth member with an illustrious record of support for liberation and efforts at egalitarian socialism. It was not to be gainsaid, and the result it sought in the election was validated by the Commonwealth. The observation had become part of the regional power politics of East Africa. The term ‘free and fair’ was first used at the Zimbabwean elections.9 It was used again, albeit in a qualified way, to scepticism and criticism in Uganda10 – but it established these loose terms as undefinable and to an extent adjustable litmus tests for elections that lasted until 2013 when, in the desperation to avoid a repeat of the naked violence and ‘electoral cleansing’ of 2007 in Kenya, the UN Secretary-General, Ban Ki-Moon, on 27 February 2013, probably without the intention of devising a lower standard, talked of the need to be ‘credible and peaceful’.11 We shall come to this in the heart of this paper but, first, a note on the methodologies of electoral observation in 1980 and 2016.

The very early Commonwealth decision in Zimbabwe was that observation had to be decentralised. The British Governor, Nicholas Soames, had undertaken to mount a separate observation exercise in which observers – sent usually by governments, together with journalists – would fly from the capital to and from chosen locations on a daily basis. They would be militarily escorted to ensure safety. And, as many observers who wanted to come would be welcomed. In part, the political rationale was to create a range of opinions about the elections so that, in the event of dispute at the results, at least some voices would declare a proper exercise had taken place. In other part, safety was a genuine factor. So the Commonwealth observation, decentralised, without inbuilt security, was clearly an exercise in risk. Observers were rotated around the various decentralised zones on a weekly basis, and a small secretariat remained in place in all zones to prepare the observation and, crucially, to liaise with all parties and all security forces – there being a total of four major armed groups, three of whom were full armies (also dispersed into centres), one of which was a smaller para-military, and peacekeeping units from four different foreign armies. The organisation of the election, in its purely administrative forms, was decentralised to provinces and districts, so observation mirrored electoral, military, and public administrative decentralisation. By and large, modern electoral observation has sought to spread itself throughout the country being observed, but without rotation of observers (there were four full rotations in Zimbabwe), without a presence of more than a month before the elections, and without on-site secretariats. Certainly there has been an effort not to depend on security forces from the incumbent government, but an entire float of health and safety procedures have meant that observation tends to avoid what may be still war-zones or zones of great volatility. In the South Sudan elections of 2010, UN peacekeeping bases were meant to provide accommodation on request from travelling observers, but the Chinese and Kenyan peacekeeping camps did not comply.12 Although the Ukrainian and Canadian camps did, many regions were under curfew, so observers were discouraged from travelling to areas where observation was most needed for fear of being stranded without shelter as curfew fell. Basically, however, albeit in an often tame way, decentralised observation has continued since Zimbabwe.

What has developed most radically since 1980 has been the use of bureaucratic check lists. These can be easily tabulated to give ‘scores’ for different aspects of electoral conduct. There are generally tick boxes for whether party agents are the prescribed distance from the polling desks, whether special assistance was available for the disabled and elderly, whether all documents, ballots and ballot boxes were in place, whether voters’ rolls were accessible, etc. Observers will tick off (or not) all the boxes and move onto the next polling station. The 1980 Zimbabwe observation used no such formal lists, but by and large sought to observe exactly such indicators of good polling practice. The result over the years is two-fold: observation groups publish tabulated results that give an impression of nationwide electoral practice; governments and electoral commissions mount very good election days with good procedures conforming to accepted international norms, precisely for the benefit of observers. The focus on the election day itself, and the final days of campaigning when observers disperse to decentralised stations, means that ‘conditioning’ of campaigning in the weeks prior, e.g. access to rallies, advertising, media; whether or not harassment and violence took place, and by whom; may be common knowledge and the subject of reportage, but not always observation. Election day itself might be ‘free and fair’ – people could vote freely, all contesting parties were represented on the ballot fairly – but the preceding campaign may have been fraught to the point that it was not ‘credible and peaceful’, let alone always ‘free and fair’. The timing of dispersal of observers to decentralised points becomes critical. A contentious election really requires at least one month of decentralised but coordinated observation. But early dispersal requires careful and detailed briefing as to what is required to be observed, as there are few universal campaign indicators that are not also heavily generalised. For instance, subtle psychological intimidation of voters by means of cultural signs or local language may not be appreciated by foreign observers. A youthful party militant rattling a box of matches in a marginal seat was taken by local people as a threat to burn property if the vote went against the government. The expression, in Shona, ‘the spirits see through all barriers’, was interpreted as a statement that polling booths and especially transparent ballot boxes could not hide the identity of voters who chose the opposition. The severe beating up of an opposition supporter was calculated, perhaps tabulated, to spread fear among his family, relatives and friends. Here, an injured body that lives to tell the tale is of much greater use than a dead one. But this kind of multiplier effect might not make an impact on observers who would report on the small incidence of such beatings. It is where intimidation oversteps its own restraints – threats by inference, exemplary but not wholesale violence – that it becomes naked and of concern to observers and the international community. The 2007 Kenyan elections were a case in point.

Before we look at those elections, what in summary are the things that derived from 1980, both positive practices and oversights? Certainly dispersal ahead of polling day – but with various time-frames involved – is a feature, i.e. the observation is decentralised; it is not confined to observing a results-verification process at an electoral headquarters but seeks to observe voting and counting of votes on a widespread basis. Secondly, close observation of the actual polling process is a continuing feature, with especial attention to rules pertaining to fairness of access and, particularly, the absence of visible intimidation.

The politesse of observation is something that developed. In 1980, observation was treated with suspicion and it could be dangerous. Now, it is almost as if an election could not be ‘proper’ without highly visible observation, and the observers are not so much the intrepid pioneers in the dark as well-accustomed members of a bus tour party. They arrive in uniform gilets and badges, many are veterans of several observations and may be doing it for the per diems, and, in Zambia 2016, they seemed uniformly comfortable – which is to say they did not seem prepared for hardship zones, i.e. in a personal and purely anecdotal assessment many were portly and some were amazingly fat and visibly ate conspicuously well.

But they share a certain ignorance and naivety with 1980. This is to do with the lack of advanced in-depth information about national political culture, political organisation, past electoral trends; and efforts to model possible outcomes. In 1980, there had been no previous elections in which the liberation parties had participated, so there was no relevant electoral history; and computer modelling was not known. In Zambia in 2016, it was foreign embassies who were computer modelling results and trends – but no observer group. This is to say, and I shall elaborate later in this paper, that the observation process is not hi-tech at this moment in time, and is exactly as low-tech as it was in 1980.

KENYA 2007
The Kenya presidential and parliamentary elections of 2007 were highly cynical. Both presidential candidates, Mwai Kibaki and Raila Odinga had worked together in the outgoing government. Both were highly credible and experienced candidates, and it was Odinga who wore such mantle of liberation as a son of Odinga Odinga – a figure of true importance in the ending of colonialism – could wear. Both men were capable of ethnic mobilisation – and well prepared to do so – and deploy such mobilisation forcefully, having also the ability to call it off at a moment’s notice. Each knew the other’s strengths and weaknesses and knew, no matter how difficult the campaign, they could work together again afterwards if they had to – even if there was no love lost at all.

Odinga was seen in the final months of the presidential campaign as the clear favourite. The aggregate of opinion polls in September and October – in a rare African country where opinion polls have some substance – had him ahead. Weekly polls in October and November, with the exception of one week in November, had him remaining ahead of Kibaki. The elections in December had three curious aspects:

1. The early counting had Odinga ahead, so any rigging against him would have occurred in the later stages of counting.
2. In the parliamentary contest, Odinga’s party won 99 seats to Kibaki’s 43 – although many parties were involved in a complicated contest.
3. Even so, Odinga’s clear parliamentary majority over Kibaki, a margin of 2 to 1 seats, was belied by final presidential results that gave Kibaki 4,578,034 votes to Odinga’s 4,352,993.

Great violence along ethnic lines erupted,13 and it took an incisive mediation by Kofi Annan to force the two sides to compromise on a coalition government.

In July, however, a US government-commissioned poll had Odinga ahead of Kibaki by 46 to 40 percent.14 The poll’s margin of era was 1.3%. What Annan’s mediation did was to restore peace, force compromise, institute measures that sought to prevent a recurrence of such violence, and propagate a principle that inclusiveness was a greater virtue than electoral victory. What it did not do was to preserve an ethos of democracy, i.e. that choice was the essence of an election. This election was almost certainly rigged. 15 Nevertheless, the principle was carried forward by Thabo Mbeki at the 2008 Zimbabwean elections – although he had certainly pioneered a similar principle in earlier
talks in Democratic Republic of Congo. What did all this, however, mean for electoral observation? There were many observer groups present, including from the EU.

The EU report concluded that ‘the 2007 General Elections in Kenya fell short of key international and regional standards for democratic elections’.16 It used no stronger language, although much of its language continued this disapproving tone. The report was comprehensive, but what it gives no evidence of contemplating – and the argument would be that such contemplation could unduly influence an open-minded judgement of results – available electoral data. By that, I mean the following:

1. Past election results, projected forwards with account being taken for a. boundary changes b. demographic changes
2. opinion polling data, interrogated for location and demographic factors
3. variance of polling data with regard to urban locations and ethnic mix
4. early trends in counting adjusted for urban factors
5. final voting profile against each of the preceding four data-sets.

The observer groups would have, thereby, some sort of template against which to measure sudden shifts in voting pattern. Those sudden shifts would demand enquiry.

Having said that, and again this is rumour and anecdotal, there are already the fabled ‘Israeli consultants’, ex-Mossad number crunchers according to the rumours – but they could be anyone with some training – who now rig the results so that they are entirely in line with plausible variations against the template indicated above. Such rumours were abundant in the Zimbabwean elections a year after those in Kenya. So that ‘rigged’ elections are no longer won in a banal way with 90% of the vote cast for an incumbent, but with 50.1%, and founded on a spread and pattern of votes gained that are both plausible and within margins of error of early opinion polls and forecasts. The algorithms for this could be written by a schoolboy. They may or may not be used in rigging; they may exist only in rumour; but observer groups should at least have the capacity to understand the possibilities and plausibilities of both honest and dishonest elections.

It also poses a key point of interrogation. Africa-wide protocol now has polling figures posted at every polling station. If there are 3000 polling stations, there are 3000 sets of results. It is impossible to have observer teams at each one. However, party agents with smart phone cameras can be at each one, and each set of results can be photographed. Or party agents and election officials can agree the final result, i.e. a ‘parallel vote tabulation’ and an official result can be made identical by a bureaucratic process. Where there is no uniform such process or reliable photographs, the central vote verification system becomes critical – and it is at this point that modern coordinated rigging must occur. Observer access to the verification procedures becomes critical – preferably armed with data that indicates what is plausible and what is not.

This does not mean that intimidation, violence and fake ballots have disappeared. Fake ballots lose their utility in the sense that a paper trail is left. It needs to be discovered of course, but the risk of discovery is greater the more widespread is their use. It also means that a large number of people would be involved and people talk. Intimidation is slowly becoming less often a tactic of first recourse as, even if it takes place before observers arrive, people affected can talk to observers and broadcast on social media moving pictures of violence. The cleanest rig is still one conducted by a small number of highly-skilled people centrally. But the election that combined mass intimidation with a slow-motion count that suggested manipulation was Zimbabwe’s of 2008.

These elections were set against previous attempts by the MDC opposition leader to achieve electoral victory, earlier failures set against accusations of rigging.17 The problem was that no one plausibly explained how the rigging was conducted. However, by 2008 the government of Robert Mugabe and his ZANU-PF government had clearly achieved economic failure – with drastically reduced agricultural output and revenues because of farm seizures and, above all, with hyper-inflation of gigantic proportions. What was amazing was that the government failed to recognise its own unpopularity, so what became the first round of the elections was not as greatly marred by violence and intimidation as became the case in an unexpected second round. Election day itself, in keeping with the now-established practice of ‘a good show for observation’, was peaceful and the electoral process was well conducted.

The build-up to polling day was in some ways soporific. It was as if ZANU-PF was so confident of victory it wasn’t really making an effort. A survey of its advertisements gives an indication. ZANU-PF published 21 different newspaper advertisements. Only one, at the end of the campaign, listed ZANU-PF’s positive qualities. Nine were largely vituperative attacks on opposition leader, Morgan Tsvangirai. Three were attacks on the third presidential candidate, Simba Makoni. Three attacked international sanctions. Five were to do with electoral issues such as land and education. Almost all were shoddily designed, as were street posters and hoardings. Almost in recognition of this, in the last week there was the sudden appearance of well-designed hoardings of a highly glossy nature: a besuited Robert Mugabe, shaking his fist, was defending Zimbabwe against foreign ownership; he was set against a beautiful green and blue backdrop of Victoria Falls. But it was the Zambian side of the Falls.18 Campaign research could not have been described as a strong point.

ZANU-PF’s own opinion polling indicated it would win with 57%, with just 27% for Tsvangirai and 14% for Makoni, but the details of the polling methodology used were never released. Nevertheless, all this added to a sense of ZANU-PF confidence on the eve of the elections. So that, as the first results began to be announced, and a trend began to emerge, basically that Tsvangirai was ahead of Mugabe by 2 to 1, the announcement of results suddenly slowed, then ceased. The direction of this trend had been prefigured after polls closed the day before. South African Broadcasting estimated a 52% victory for Morgan Tsvangirai. My own estimate was 56% for Tsvangirai.19 But I also had Mugabe receiving not even 40%, as Makoni picked up sufficient votes to deny him that. Tsvangirai’s MDC first claimed 60%, but later moderated that to 50.3%. Some years afterwards, Ibbo Mandaza, a Zimbabwean intellectual ‘insider’, insisted it was as much as 70%.20 The basic thrust of all these estimates was a defeat for Robert Mugabe in the presidential vote and for his ZANU-PF party in the parliamentary vote.

The election had concluded on the evening of 29 March 2008. Tallies for each polling station were posted outside the stations on the morning of 30 March. No further central announcements were made by the electoral commission from the afternoon of 30 March. The rumours of an Israeli ‘consultancy team’, arriving on 31 March, began circulating the afternoon of its supposed arrival. The derivative rumours then had the Zimbabwean military commanders demanding the ‘arrangement’ of a 52% vote for Mugabe, and a combined 48% for Tsvangirai and Makoni. They otherwise threatened a coup on behalf of Mugabe. The ‘consultants’ were anxious, however, to deliver final results that were mathematically credible – and this is where the modern science of rigging had its genesis. The results were finally announced on 2 May – so credibility was a painstaking exercise – and Tsvangirai in the end, although he outpolled Mugabe, received less than 50% of the votes cast, and a runoff was required.

The thing is that the results could not be dismissed as merely incredible. The days of dictators winning sham elections by 90% were over.

a. Those polling station results that had been independently verified, e.g. by photographs on smart phones with date and time records, had by and large to stand.
b. Those not accurately verified, e.g. if photographs were blurred because of poor camera function, could be adjusted.
c. It was important in adjusting Mugabe’s tally upwards not simply to adjust Tsvangirai’s downwards; so many votes were appropriated from Makoni’s total, and Makoni had no national network to protest on his behalf.
d. Parliamentary results and presidential results had to indicate a degree of symmetry, i.e. it was unlikely a huge number of those who voted for the MDC had also voted for Mugabe. It was not impossible that some might, e.g. if there were a ‘favourite son’ parliamentary candidate, but it could not become too obvious a trend. Even so, the decision to allow the MDC a very narrow parliamentary majority meant that the vote for Tsvangirai could not be disproportionately lower than his party’s parliamentary vote.
e. Voting patterns had to be in general keeping with the results from earlier elections, and although there is much suspicion those too had been rigged, they had been rigged for purposes of a previous day, and transferring patterns to an election of different circumstances required a mathematical logic of its own.
f. Finally, the ‘adjustment’ in the month-long count, in a normally extremely efficient country, had to be conducted centrally, so the role of observers at the central verification count became of great importance. It was reported that an observer slept each night in the room with the assembled ballot boxes. But there were more 9132 polling stations, and each would have filled more than one ballot box for each of the presidential, parliamentary, senate, and local council elections contested at once. All those contests attracted 4406 candidates – so a single, surely sometimes sleeping, observer in a room with them all could not physically monitor them all; and seals broken during the night could be easily resealed authentically by resumption of the count each morning.
g. Even so, the figures indicate that a ‘scientific’ rig was a monumental task, so a month was probably very good going – although exactly the need for a month indicated to all and sundry that no ordinary count was being conducted.

The importance of observers, in ‘robust presence’ at all stages of the verification count conducted centrally becomes a key lesson here – as does the requirement for an additional layer of observation, i.e. there requires to be an observable bureaucratically verified point of agreement between official polling station tallies and what are now called ‘parallel vote tabulations. There is in fact a software available, pioneered in 2008 in Cambodia.21 This does not yet seem in use in Africa. The PVT are the tallies verified by party agents and accredited observers within each polling station, i.e. they verify what the polling station staff post as the results, having watched the count. The process of counting requires observation, but so does the moment of agreement between the official result and the PVT. This might seem a clumsy and gargantuan process, but it doesn’t even require heavy-duty software. Computer links via encrypted systems with firewalls and multiple but correlated passwords should provide at least a highly indicative preliminary picture. In some countries with less spread of WiFi, e.g. Sudan in 2010, this becomes difficult – less so in the Zimbabwe of today.

In the absence of such provisions, citizen groups have tried to devise electronic tools. An app, though not as scientifically-based as the one in Cambodia, was fielded as an independent tabulation-awareness tool in Gabon in 2016.22 In the Zimbabwe of 2008, not all the smart phone cameras were reliable. Many pictures of polling station tallies were taken on Chinese clone phones. The need for distribution of reliable phones should have been a priority. In the UK, the Liberal Democratic Party sent a shipment of such phones to Zimbabwe, but they never made it past South African customs in time.23 But observation should not simply be a closed-shop exercise, i.e. not just an exercise conducted within an accredited group. Observation needs to reach out to civil society actors at least, and that should include disseminating technology for observation and as a part or adjunct of a PVT process. Otherwise, observation as it is now will soon be overtaken by citizen reportage and tabulation, using far more sophisticated tools than observers now deploy.

In the event, in the runoff of 2008, ZANU-PF resorted to very old-fashioned bludgeoning techniques of great and crude violence. Morgan Tsvangirai withdrew. Mugabe swept to victory with a count that lasted a single day.

These elections marked the end of the power-sharing government mediated into existence by Thabo Mbeki after the furore that accompanied the 2008 elections and the long months of delicate and sometimes brutal negotiations that extended into 2009. Morgan Tsvangirai accepted the invented position of prime minister, despite probably having won the presidency – had the count in 2008 been straight-forward and honest. But he did not prove a good prime minister.24 At the same time, his opposition colleague, Tendai Biti, had been a most successful minister of finance, stabilising the economy and providing the citizens of Zimbabwe with the possibility of planning horizons which had disappeared with hyperinflation. However, what this meant for voters was not necessarily an automatic endorsement of Biti’s party, but a judgement as to what power configuration would best safeguard these hard-won gains. This was a key consideration of a particularly comprehensive pre-election poll of voters’ intentions. The survey by South African academic, Susan Booysen,25 for Freedom House, was scanned for its headline figures but left largely uninterrogated by some political quarters in Zimbabwe.

Those headline figures, in a survey of 1198 respondents and a margin of error of 2.8%, had 47% of the respondents declining to provide a voting intention. This was seized upon by Tsvangirai’s MDC strategists as meaning 47% were undecided – but further reading of the report would have seriously challenged such an interpretation.

In particular, the survey was far from generalised. It covered 100 areas, 63 rural and 37 urban, consistent with population distribution. It showed consistent falls in approval ratings for both the MDC and Morgan Tsvangirai across a range of carefully calibrated survey topics – including stabilisation of the economy. And it showed falls in approval ratings across the range of topics in all the MDC urban strongholds. In those same strongholds, ZANU-PF approval ratings rose. On the key topic of ‘trust’, MDC approval ratings dropped below those for ZANU-PF. And on the topic of capability to bring change, ZANU-PF outpolled the MDC by 29 to 22%.

It wasn’t all doom and gloom for the MDC, but there was well enough in the Booysen report to make strategists aware that there was much to fight for, and also to give a range of topics over which to wage an electoral fight. ZANU-PF took the report far more seriously than the MDC, and realised that, if rigging was to be employed, the report gave a template upon which final manipulated voting figures could be based and made to look credible. By this stage, Ban Ki-Moon’s formulation, ‘peaceful and credible’, had entered the lexicon. Zimbabwe’s neighbours had rounded on the country to avoid the naked violence of 2008’s second round, so rigging had to be accomplished without primary reliance on obvious intimidation and strong-arming. Credibility became the key motif in preparations for the contest and assessments of the contest. In short, elections are now managed to convince observers of their credibility – not their probity – and the stage-managers have more in their repertoires than the observers have analytic tools. Elections become complexly stagemanaged validations, rather than exercises in choice. Choice has to be overwhelmingly against the incumbent, to the extent that it would not be credible if the incumbent won, for choice to have a chance to make it through.

As it turned out, the election was won by President Mugabe and his ZANU-PF Party, certainly with irregularities in the electoral process – the voters’ rolls begged various significant questions – but it would seem that the irregularities were designed to deny the opposition a constitutionally-blocking third of the parliamentary seats. Without any rigging, ZANU-PF could plausibly still have won by 52%. Even so, the spread of the vote was such that had the MDC healed the splits with breakaway factions and not faced a split opposition vote, and had it mounted a youth registration drive to ensure a greater youth turnout, the result would have been much closer. As it was, Mugabe defeated Tsvangirai by 2,110,434 to 1,172,349 votes; and ZANU-PF took 197 parliamentary seats to the MDC’s 70, with 3 seats taken by others.

But the point is that, in line with key aspects of the MDC’s post-election complaints, if all the mysteriously long-lived centenarians on the voters roll (100,000), and all those turned away for not being properly registered (305,000), and all those who required ‘assistance’ to cast their votes (207,000) had voted without exception for Tsvangirai, Mugabe would still have won.26

In an election with an army of some 20,000 domestic and international observers, and party agents at 96% of the polling stations, none could provide forensic evidence of how rigging would have changed the final result. The scale of victory could be deemed not fully credible, in terms of questions pertaining to the voters roll, but the fact of victory in itself could not be demonstrated to be implausible.

The question is whether or not there is a possible gradation from what is plausible to what is credible. Pre-election polling suggested the credibility of a Mugabe victory. The eventual size of that victory was the subject of questions to do with plausibility, but there was no coordinated effort to field a set of indices for plausibility.

With a talent for exquisite euphemisms, UN Secretary-General, Ban Ki-Moon, called for these elections to be ‘peaceful and orderly’27 – and the epithet of ‘credible’ was not at this stage used. The encouragement was for them to be well-conducted, ‘orderly’; but, with each introduction of new epithets, ‘free and fair’ sunk further into the background of desiderata. There was in fact a rise in electoral violence. Since 1991 there has always been violence in Zambian elections. Youthful party thugs have always been mobilised to intimidate opposing candidates and prevent them from campaigning. Zambians have always played down this aspect of their elections, partly because of the lingering doctrine from the Kenneth Kaunda era that Zambians were more peaceful than others, and also in the belief that violence in other countries is much worse.28 In 2016 it seemed that both main parties let their youth wings off the leash more than ever before.

The other major aspect was the almost entirely one-way reportage of the nation’s newspapers – in favour of the incumbent president. The independent newspaper that supported the opposition, The Post, was forcibly shut down – ostensibly on grounds of failing to pay its tax bills – but it seems that it became critical to shut it down just before the elections. It meant, in a land where electronic and social media have not yet had the impact felt in countries like Zimbabwe and South Africa, that there was no balance of extremes in an otherwise lacklustre press.

Both these aspects faded in significance when it came to accusations and suspicions of rigging in the electoral count. That count took three days with a very slow start, a surge, and then another slowdown at a critical point. There seemed discrepancies between what the Electoral Commission of Zambia was announcing and the parallel vote tabulations conducted by party agents for the opposition. There were meant to be means for agreeing these official and unofficial counts at each polling station, but forms for doing so were not available at some stations.

And the contest seemed all the keener because of new requirements that the winner had to take over 50% of all votes cast. It could not be simply a case of one person taking more votes than another. Without 50% being achieved, there would have to be a runoff between the top two candidates. This meant an incumbent could always look forward to a second chance in a runoff. It also meant that any rigging had to deliver not only a majority of the votes, but enough to avoid a runoff – otherwise it would all have to be done again. Entering the contest, both sides had crunched forecasts to identical conclusions: that whoever won would do so by no more than 52%. It would be a very close race.

Finally, the Electoral Commission announced a 50.35 victory for President Lungu. Immediately afterwards, the Commission’s results website went down. When, a day later, it came back up, there was reasonable congruity between its count and that of the PVTs. The official figures said that Lungu had gained 1,860,877 votes and Hichilema 1,760,347, with other candidates far down the field.29

Where the suspicions were focused was in the secrecy of vote verification on the part of the Commission. Whereas polling station and constituency centre counts had been open to accredited observers and party agents, the European Union Election Observation Commission regretted ‘that its observers do not have access to the verification of results at the national results centre, despite formal requests’.30 Suspicions were amplified when the announcement of results suddenly slowed down just as the big urban centres were coming in. It had always been opposition strategy to take by a huge margin its normal strongholds, but to capture sufficient of the urban vote in Lusaka, and in the northern cities of Ndola and Kitwe (Lungu’s strongholds) to prevent the incumbent from realising the full benefit of his own heartland. This strategy briefly seemed it was working, but in the end did not. The case of Kabwata constituency in Lusaka is a case in point. In what should have been Lungu territory, the melting pot effect had Hichilema run him close. If this pattern had been more widely replicated in the cities, the outcome would have been different. But for all the suspicious circumstances outlined above, and for reasons of media bias, unreasonable exploitation of encumbency, and violence, Hichilema announced he would petition the Constitutional Court to annul the outcome. This petition failed on procedural grounds and was not properly heard by the court.

Based on the discussion of the elections above, but drawing also from the author’s observation of elections and referenda generally since 1980, the following observations for the future are offered.

There are major caveats at the very beginning. These are not intended for elections conducted in the immediate or short-term aftermath of war and great violence. The 1980 Uganda elections offer an example of the difficulties that come from war – not just within the war-torn country itself, but from external pressures to broker, even gerrymander an electoral settlement that will take things forward towards some degree of stability. In such conditions, stability becomes a greater value than democracy.

Nor are these intended as fully suitable for elections which have no intention from the outset of allowing an incumbent government to be replaced – even if great public protest is directed against the result, and an external mediator then works towards a compromise or coalition government, with governmental places for those who were officially ‘defeated’, but who probably ‘won’. In such conditions, inclusivity becomes a greater value than democracy.

These suggestions and categories are offered to observer groups at elections where the surrender of power is possible – even if reluctant; and the test of freeness and fairness, or whatever set of appellations used, lies in the quality and forcefulness of this reluctance. I now add to these appellations – but noting that these are not necessarily exclusive categories; an election might be described in any combination of the broad but hopefully helpful and robust categories below.

Moreover, I should stress that observation cannot be a condensed affair, nor should it be contained only within the observer group but also in its links with other observer and civic groups. A ‘condensed’ observation, in my mind, is one which has taken less than a month on the ground. At the very least, an advance team of experts, or those briefed on the constitutional, electoral, and political affairs of the country should be in place as a reconnaissance unit a month before polling day. Most observer groups now contain such experts, but they are called upon to answer questions as they arise in the final phases of observation, and not to present to the observers their own latest observations gleaned from time in the country.

That reconnaissance team needs also to be energetic and mobile, traversing the country. It is no country for old men (or women, or fat, unfit, timorous and frightened men and women) to do this kind of thing properly. At the 2010 Sudanese elections, while assigned to areas of South Sudan still enjoying forms of contention or war-readiness, we took a simple executive decision: if we saw an EU, AU, or Carter Centre car, we weren’t out far enough. We never saw an Arab League car. We went until there were no other observers for miles around – but then asked ‘why?’

To be Free and Fair
With the clear recommendation of time-on-location before polling day, mentioned in the preceding paragraph, what would be the requirements for an observer group to declare elections ‘free and fair’?

1. When there has been a clear moratorium on violence by all organised contenders.
2. When there has been no hinderance on physical campaigning, and state security has been made available to all organised contenders for their major rallies.
3. When all media has been free to report on the election, even if that means a moratorium on debt owed by any section of the media – free expression and publication being an electoral good over and above considerations of debt, especially debt to government offices.
4. When that media is comprehensive, i.e. it should include access to broadcasting; this includes broadcasting in vernacular languages on e.g. vernacular radio.
5. When the incumbent government does not make use of incumbency during the campaign, e.g. by launching or opening huge public works as a campaigning strategy, or by utilising state funding for its own campaign while denying exactly such funding to organised challengers.
6. When polling day conforms to the now established protocols of access and facilitation and secrecy.
7. When procedures are in place at each polling station to agree official and PVT final figures and for these to be transmitted to electoral commissions in a recorded manner.
8. When all electoral commission verification procedures are witnessed by observers.
9. When the observers detect no untoward variations between sustained opinion polling before the elections, voting patterns in the previous election, voting patterns for parliamentary and presidential candidates, and voting patterns in adjacent densely populated urban constituencies – without clear margins of error being explicated, or other decisive characteristics taken into account.
10. When there is no interruption to the counting process, i.e. once started, it is continuous, observable and gives no pause for tampering or adjustment.

To be Peaceful and Credible
1. ‘Peaceful’ relates to the same condition, 1 and 2 above, pertaining to ‘Free and Fair’.
2. Basically, ‘Peaceful and Credible’ relates to conditions, above, 3 to 6, about conditions of campaigning, but it is a judgement call as to their being not ‘good’ but ‘good enough’, i.e. campaigning met difficulties, especially for opposition candidates, but on a by-and-large basis, campaigns were able to be nationally conducted and a clear opposition message was able to be put across.
3. However, the real test of ‘credibility’ lies in 7 to 10, above, to do with counting, i.e. it was not seamless, not always continuous, and there were lacunae in opportunities for observation; but, by-and-large, margins of error allowed for a not fully perfect result to be regarded as ‘credible’. In the rationale behind ‘margins of error’ need to be made clear by the observer group.

To be Plausible but Problematic
This judgement resides primarily in the tests for 7 to10, above, i.e. the margins of error are larger than satisfactory, larger than fully explicable – but are conditionally explicable – and it is for the observer group to make this explication clear and how it was utilised in its final judgement.

To be Not Fully Plausible
When the conditions outlined in ‘conditionally explicable’ deviate from full plausibility, i.e. the key inference here is of intervention of a sustained and planned and calculated manner in key stages of the counting and verification process.

To be Implausible
What happened in 2016 in Gabon provides a key example where, after a critical pause in the counting and announcement of results, the President was returned by results in a key constituency, where turnout was 20% higher than national turnout, and voting in favour of the President was likewise much higher than in any other part of the country. The delayed, sudden decisive variation is grounds for implausibility.

To be Unacceptable
When both campaigning and counting were subject to so many observable interventions of an untoward nature that no credibility or plausibility attaches to the outcome. This should have been the verdict of observation of the 2008 elections over two rounds in Zimbabwe, despite efforts to render the final results of the first round in a sense ‘plausible’.
The term ‘not conforming to international standards’, in regular use, should be amplified by the addition of one of the last three of my categories, above.

Election observation has become mired to a considerable extent on the conduct of polling day itself, and a short period of time afterwards – but presence in the country to observe a full swathe of conduct in campaigning has not always been available. Above all, concentration on all aspects of counting, and testing of the count in manners outlined above, have not been sufficiently robust at all recent African elections. Insofar as African governments now prepare almost immaculate polling days – feats of organisation involving thousands of polling stations – election observation has accomplished something. Observation still needs to give thought to a more extended presence during campaigning. But, importantly, it must address its own key next stage, which involve the forensics of observing and testing the count. What does this mean? In summary it should involve testing the electoral commission’s count:
1. Against PVTs.
2. Against opinion polls, patterns from previous elections; results in adjacent constituencies; and the concerns of NGOs who also fielded observers.
3. It should involve monitoring the transmission process as agreed local totals reach central points. How is this done? If manually, against what parallel electronic process?
4. The count at each stage should be tested against computer projections from when the last batch of results came in, calibrated against a range of different conditions, types of constituency, and electoral histories and patterns.
5. It should involve testing of the robustness of data assemblage in the central verification process and observing all aspects and moments of the verification. Bearing in mind my introductory comments that specialist consultant firms are rumoured now to adjust results in a credible, or at least plausible, way – the observers do require the technological and forensic skills to play the ‘adjusters’, not any more merely naked ‘riggers’, at their own game. Even if the rumours of such ‘adjusters’ are merely the imagination of a computerised age, observer capacity to apply computerised forensic tests will at least delay the advent of any such ‘adjusters’ of the future – or at least make their task much harder. In short, the protection of electoral democracy today and tomorrow requires sharp observation skills that cannot simply be borrowed from yesterday.

There can be upwards of 20 000 observers at African elections. The bulk of these will be local civic action group observers. It seems a big number but, in an election of e.g. 5000 polling stations, it means an average attendance at each station of four observers at any one time; if they come from disparate groups, their observations may also be disparate, i.e. there is no means to coordinate or harmonise verdicts. In any case, their attendance on election day will be, as I remarked earlier, to compile checklists of good polling station administration and conduct.

In fact, the number involved in observation is larger in many cases, as interested countries, e.g. those with large aid programmes to the country concerned, will have many volunteers from their own civic action, think tank, and advocacy groups from their own country and, while not accredited as observers, will accompany selected local observer groups. These will report to their own embassy, where the political officers will certainly be ‘number crunching’ the observed results. So, behind the scenes, both the election and observation will be the subject of a parallel observation and assessment. None of its findings are intended for publication, but certainly for reportage to government offices back home. My point is that the official observer groups should just have number crunching capacity themselves, especially those from large international organisations, such as the AU.

Is the continual observation of elections over many years in many countries dispiriting? Yes, to the extent that I don’t enjoy it anymore and would rather not do it anymore. But, as a private observer, I have my own methodology and part of that is to observe election day in the slums of the great cities. I lived in Kliptown, the poorest part of Soweto, for the last two South African elections; I was in George, the poorest part of Lusaka for the 2016 Zambian elections. It is for the people who have no capacity to get on a plane afterwards, as both I and all the other international observers have, who must live with the results of election after election only to have – at five-yearly intervals – an eccentric longhaired Chinese visitor find them in exactly the same under-developed situation, that elections should be held in an honest, yes, free and fair manner. Why should they have to live with the regular delusion that their hopes can change for the better by their own choice? But a paper like this one, which outlines a gradation downwards of categories from ‘free and fair’, indicates a harsh realisation that the poor will not be able to vote their way out of wretchedness.

1 For a dissident guerrilla view, see Wilfred Mhanda, Dzino: Memories of a Freedom Fighter, Harare: Weaver, 2011.

2 Stephen Chan, The Commonwealth in World Politics: A Study of International Action 1965 to 1985, London: Lester Crook, 1988, pp 42-5.

3 Stephen Chan, The Commonwealth Observer Group in Zimbabwe, Gweru: Mambo, 1985.

4 For an early contemplation of the differences, see Simon Hardwick, ‘Administrative aspects of the Uganda election— December 1980: A comparison with Zimbabwe’, Public Administration and Development, 2:2, 1982.



7 :how-upc-rigged-the-1980-elections

8 Justin Willis, Gabrielle Lynch and Nic Cheeseman, “A valid electoral exercise”: Uganda’s 1980 election and the rise of international election observation,  0rise%20of%20international%20election%20observation.doc

9 It was the declared intention of the British government to hold ‘free and fair’ elections. The conclusion of the Commonwealth Observer Group was that it was ‘a valid and democratic expression’: Southern Rhodesia Elections February 1980: The Report of the Commonwealth Observer Group on elections leading to independent Zimbabwe, London: Commonwealth Secretariat, 1980, p 74.



12 Based on personal experience in South Sudan at this moment in time.

13 Nic Cheeseman, Democracy in Africa: Successes, Failures, and the Struggle for Political Reform, Cambridge: Cambridge University Press, 2015, pp 166-7.

14 ‘US-funded exit poll says Raila won elections’, The Nation, 11 July 2008.


See click through to the report. These words introduce the report in the Executive Summary.

17 Stephen Chan, Robert Mugabe: a life of power and violence, Ann Arbor: University of Michigan Press, 2003, Chapters 8 & 11; Stephen Chan, Southern Africa: Old Treacheries and New Deceits, New Haven: Yale University Press, 2011, Chapter 8.

18 Chan (2011), Chapter 10.


20 Mandaza in a private conversation with the author in Harare. Mandaza was the expert commentator on national radio as the results began to be broadcast, then were suspended.

21 T_Report.pdf


23 LibDem spokesman, Ed Davey, Shadow Secretary of State for the Foreign and Commonwealth Office, to the author, discussion in the House of Commons after the Zimbabwean elections, 2008.

24 Stephen Chan, ‘The Tragedy of Tsvangirai’, Prospect, August 2008.


26 See Stephen Chan & Julia Gallagher, Why Mugabe Won: The 2013 elections in Zimbabwe and their aftermath, Cambridge: Cambridge University Press, 2017, for our calculations.


28 Stephen Chan, Kaunda and Southern Africa: Image and Reality in Foreign Policy, London: I.B. Tauris, 1992, espec. pp188-192.


30 European Union Election Observation Mission – Zambia 2016: Preliminary Statement, Generally well administered and peaceful election day marred by systematic bias in state media and restrictions on the campaign, Lusaka, 13 August 2016, penultimate para., p9.