Journal of Democracy 15.4 (2004) 147-158
More than a decade has passed since the third wave of global democratization broke on Africa's shores. This African wave came to world attention in February 1990, when the South African apartheid government decided to lift the ban on the African National Congress and release Nelson Mandela from three decades of imprisonment. Only four years later, vast crowds cheered Mandela's presidential inauguration as the new nation's multicolored flag replaced the hated banner of the apartheid state. And in 1999, young Nigerians danced in the streets to celebrate the downfall of a brutal military dictatorship and to welcome a new civilian regime.
Mass public celebrations have often greeted the ouster of longstanding authoritarians and the birth of democracy at the polls. But how long does such enthusiasm last? Are Africans' preferences for democratic government enduring or ephemeral? This essay reports that, consistent with trends in other world regions, mass democratic commitments in Africa are far from fixed. Instead, popular support for democracy tends to drift downward over time. Reassuringly, however, it seems that Africans' commitment to democracy can be refreshed by alternations in power by way of elections.
In short, there are signs of an incipient pattern in the evolution of public opinion in Africa's new democracies. Mass political optimism invariably marks the immediate aftermath of founding elections, and the public mood is especially buoyant in countries where the preceding regime has been deeply repressive—as in South Africa and Nigeria. Indeed, landmark democratic transitions often generate rashly unrealistic expectations about the benefits that democracy will bring. But on the "morning after," initial popular exuberance dissipates. If political life reverts to familiar patterns, people begin to question the desirability and quality of the new order, and they tend to sober up sharply if elected elites betray their popular mandate by indulging in corrupt or manipulative behavior. In cases where a rotten team of incumbents survives subsequent elections, it does not take long until the general public becomes disillusioned with democracy.
Can alternations in power restore popular faith in democracy? More than any other political event, a peaceful vote and subsequent transfer of power from one group to another should serve in the public mind to validate "rule by the people." But in reality, does electoral alternation help to relegitimize democracy?
Surveys of public opinion in Africa offer an opportunity to test whether the election of new ruling parties can in fact reinvigorate stalling attitudes toward democracy. This essay draws upon data from the Afrobarometer—a comparative series of public attitude surveys on democracy, market reform, and civil society. The first round of surveys took place between July 1999 and October 2001, and covered more than 21,000 adult citizens in 12 countries. The second round reached more than 23,000 persons in 15 countries between June 2002 and October 2003. In each round, trained fieldworkers conducted face-to-face interviews in a local language of the respondent's choice. For the sake of consistency, however, the key term of "democracy" was expressed in a national language, whether English, French, Portuguese, or Swahili. By this standard, an average of more than three-quarters of the respondents in the surveys' randomly selected national samples was able to attach a meaning to the term.
In order to arrive at reliable conclusions about trends in public opinion, analysts generally prefer to have at least three observations, separated by intervals of several years. Otherwise, momentary shifts in volatile attitudes—or mere measurement errors—may be misinterpreted as lasting changes in the public mood. Because so far only two observations are available for most countries in the Afrobarometer, we take several precautions to diminish the chances of arriving at faulty conclusions: We note if interview questions change between surveys, we break down Afrobarometer "averages" by country, and we draw attention only to differences of 10 percentage points or more. For any given Afrobarometer survey the margin of sampling error is plus or minus 3 points, which doubles to 6 points when two surveys are compared. Therefore, we use an even larger margin—at least 10 percentage points—before speculating about changes in public...