The 9th International Anti-Corruption Conference
Corruption - The Attitudinal Component: Tracking Public Perceptions Of
Official Corruption In South Africa, 1995-1998
Corruption - The Attitudinal Component: Tracking Public Perceptions Of Official Corruption In South Africa, 1995-1998
Robert Mattes Cherrel Africa
Regardless of the actual extent of real instances of corrupt behaviour among public officials, it is vital to assess the degree to which citizens believe that their leaders are involved in such behaviours.
There may be no more important challenge in building a new democracy than creating a widespread sense of legitimacy amongst its citizens: that is instilling a sense of trust and confidence in, and commitment to democratic institutions and processes. In a new democracy, much of this depends on the extent to which people believe that government is effective, and that elected leaders respond to voters' wishes, are honest, and work for the public good.
This paper will focus on the last set of these perceptions: citizens' assessments of official corruption in South Africa. To reiterate, the data reported here do not comment upon the actual extent of corruption in government in South Africa, but only citizens' perceptions of official corruption.
Tracking South Africans' Perceptions of Official Corruption
In this paper, we report results from seven separate surveys of nationally representative samples of South Africans designed by Idasa's Public Opinion Service.
1. Absolute Evaluations of Government Corruption
Our principal method of measuring public attitudes toward public corruption has been to ask the following question in a format widely used in international opinion research.
"What about corruption, that is, where people in government and the civil service illegally used public monies for their own benefit or take bribes. How many officials do you think are involved in corruption?" The response categories are:
Summing the "almost all" and "most" responses together to express perceptions of high levels of public corruption, Idasa's first survey of South Africa's formative democratic culture in September-November 1995 found that, already, 46% of South Africans felt that "almost all" or "most" public officials were involved in corruption.
The Idasa June-July 1997 survey broadened the range of indicators by posing the same question about officials across different branches and levels of government. It found that perceptions of corruption were fairly widespread across these different institutions, with the significant exception of the office of President Nelson Mandela. When we repeated these question in the Idasa survey of November-December 1998, we found that perceptions of corruption had yet again increased across each indicator.
(1995) "What about corruption: that is where people in government and the civil service illegally use public monies for their own benefit or take bribes. Would you say that almost no public officials are engaged in it, a few public officials are engaged in it, most officials are engaged in it, or that almost all public officials are engaged in it?" (1997 / 1998) "What about corruption: that is where people in government and the civil service illegally use public monies for their own benefit or take bribes. With regard to the __________, how many officials do you think are involved in corruption ? [substitute ....]
2. Relative Regime Evaluations
A second way that we have approached this is by asking people for a comparative assessments of corruption under the new democratic dispensation versus the old apartheid-National Party regime. The emerging cross-national literature on democratisation suggests strongly that absolute evaluations of new democratic regimes are less important than how people see those institutions in relation to those of the previous authoritarian regime (Rose, Mishler, and Haerpfer, 1999).
Thus, it could be that regardless of how much corruption South African see in their new government, the key question is whether they believe it is less corrupt than the old.
When we asked this question in our 1995 survey, we found that four-in- ten South Africans (41%) felt that their new democratic government was more corrupt than its predecessor. In addition, another one-quarter (25%) felt that it was no different than in this respect. By 1999, the position had improved very slightly, with 39% saying the new democratic government is more corrupt, but 28% still saying they saw no real difference. Thus, in both surveys, a similar 67% could be said to see either no change, or an increase, in corruption from apartheid to democracy.
Do you think that there is more or less corruption in government than there used to be?" (1998) "Compared to government under apartheid, is government today more or less corrupt, or is about the same as the old government?"
3. The Public Salience of CorruptionA third way that we have tracked public perception about corruption has been through a very different, open-ended type of question that asks people "What are the most important problems facing this country that government ought to address?" People give us their spontaneous answers: no lists are provided. We prompt for up to three answers.
What is significant is that in seven separate national surveys conducted between 1994 and the 1999 election, corruption or related issues have been mentioned by more than one-in ten South Africans only once, in April 1999 (though proportions ranging from 15% to 22% of Eastern Cape respondents spontaneously mentioned the issue between September 1998 and April 1999).
At the same time, the previous three surveys (all from the Opinion '99 election project) showed that the public salience of corruption as an important national issue was on a definite upward curve. Whether that concern has been sustained, or whether it was only the reflection of election campaign rhetoric, remains to be seen.
"What are the most important problems facing this country that government ought to address?"
4. Public Approval of Government Anti-Corruption Actions
During the Opinion '99 project (an election research consortium of Idasa, the South African Broadcasting Corporation, the Electoral Institute of South Africa, and Markinor), we also placed two corruption-related items among a larger set of questions that asked people how well government was performing on a wide range of activities.
During the election campaign, the government received mixed but improving reviews on its efforts to fight corruption and maintain accountability and transparency.
By our last survey in April 1999, 44% felt that the government was doing its job "well" pr "very well" in controlling official corruption (up substantially from the meagre 26% who had said so in September 1998). And by the end of the campaign, 55% approved of the job its had done in maintaining democratic transparency and accountability (compared to 31% in September). The degree to which these responses were a reflection on actual government performance and, or a very successful ANC election campaign, is uncertain. Public opinion on a wide variety of government performance areas reached the highest levels of the entire post-1994 period right before the actual election (Mattes, Taylor and Africa, 1999).
"How well would you say the government is handling each of the following? Fighting corruption in government?"
"How well would you say the government is handling each of the following? Maintaining transparency and accountability."
Finally, our 1995 survey contained a wide range of other question on official honesty, and more specific questions around parliamentary salaries and ethics reforms (Mattes and Calland, 1999). None of these items have since been repeated.
Accounting for Perceptions of Corruption
The obvious question, then, is "Why do South Africans hold such fairly negative views of official corruption?"
If we had sufficient questionnaire space, or resources to fund a survey solely focussed on corruption, we would have directly answered this question by asking people why they felt, or did not feel, that government officials were corrupt? We would assess direct personal experience with, or victimisation by, corrupt practices or officials in a wide range of different settings. We would also explore the full range of ways the people can learn about government corruption through various types of information networks.
Because Idasa's surveys have been focussed tracking the emergence of a democratic culture in South African society, we have only tracked public perceptions of official corruption as one (among many) key variables that may account for how people may see their new democratic regime. But what our surveys do permit us to do is to examine this central question in an indirect way through an analysis of the demographic correlates of perceptions of corruption. We cannot say whether these perceptions are a true reflection of the actual extent of official corruption (which would require actual behavioural data on corrupt behaviour), or a direct result of personal experience with corrupt officials. But a demographic analysis does allows us to begin to address a number of potential explanations.
One possible explanation is that entire the South African population is not as cynical about official corruption as the aggregate results reported thus far suggest. Rather, one might argue that the apparent cynicism of the citizenry is actually the result of smaller subgroups of the population that are especially suspicious of, or hostile to black government. Thus the aggregate data might create a misleading picture by combining the opinions of relatively satisfied black South Africans with those of highly dissatisfied white, coloured or Indian citizens.
Another possible, and often-heard argument is that these perceptions are largely media-created. Public corruption, some say, only appears to be as bad as it does because of greater press freedoms under the new democratic dispensation that shine a greater spotlight on the few instances of corruption that do exist. Alternatively, some say that these perceptions are only a reflection of a news media that overlooked the sins of the apartheid government, yet sensationalises every possible transgression of the new government.
Yet another possibility is that public cynicism is the result of public ignorance: if people simply knew more about government and public affairs, they would be less cynical about public officials.
In order to assess these different possibilities, we examine the bivariate correlations of a range of demographic indicators and perceptions of corruption in 1995, 1997 and 1998. We use the single item on perceived corruption among public officials / civil servants because it is the only one to be asked in all three surveys (see Appendix A).
A lot of the common wisdom on this matter finds little or no empirical support.
If perceptions of corruption were fanned by the news media, we would expect to see increased perceptions of corruption amongst those who most often get news from various media sources (we asked about getting news about politics from television, radio, and newspapers). In fact, no such general effect is visible. Only in the 1995 data do we seen any effect and it is in precisely the opposite direction than expected: those who use the news media most often were the least pessimistic about public corruption. However, no general effect is discernible in 1997 or 1998.
Neither is there any evidence that more (or less) knowledgeable, articulate sections of the public are less (or more) likely to perceive high levels of public corruption. We see no significant effects of education, income or rural-urban status.
There is, however, evidence across all three surveys that perceptions of official corruption vary significantly by race, as well as that levels of cynicism about public corruption are especially high amongst minority race groups.
White South Africans have consistently been most pessimistic, and have become sharply more cynical since 1995. As of 1998, two-thirds of white respondents (66%) said that "almost all" or "most" were involved in corruption. They are also now matched in cynicism by Indian respondents (65% who felt this way).
However, while these two groups are especially cynical, it is also the case that 52% of black respondents also now say that "almost all" or "most" public officials are engaged in corruption.
The same general patterns are true with regard to relative assessments of corruption now versus under apartheid. While, whites and Indians are far more likely to see increased corruption under the democratic government, more than one-third of black respondents (35%) also believe that corruption has increased under the new order.
The demographic analysis also reveals a strong impact of political geography. There are strong provincial impacts on beliefs about corruption. These is true for the entire sample as well as among and blacks on one hand, and among white, coloured and Indian respondents on the other. In other words, even once we examine only the opinion of black respondents, we still find important differences in the perceptions of corruption between, say, black respondents in Eastern Cape versus black respondents in Mpumalanga.
In general, the highest levels of perceived corruption are located among respondents in KwaZulu-Natal, and Eastern Cape. This is true both with regard to beliefs about public servants in general, as well as about provincial government. We also see some important shifts within provinces with regard to public perceptions of corruption (Northern Province where they have improved notably, and KwaZulu-Natal and Eastern Cape where they have deteriorated significantly).
Further analysis is required to understand these regional variation. Some plausible factors are: the presence or absence of the legacy of homeland administrations from province to province; the differing legacies of different homelands with regard to corruption; and the way in which governments in different provinces have communicated their strategies in fighting corruption to their respective electorates.
Connecting Perceptions of Corruption to Democratic Legitimacy
Does all of this matter? To what extent to popular perceptions of official corruption affect the way South Africans view their democratic institutions or the way they are likely to behave toward government?
We examine the impact of beliefs about official corruption on three import components of democratic legitimacy. First of all, we look at whether it influences the degree to which people are "satisfied with the way democracy works in South Africa." Secondly, we examine the extent to which perceptions of corruption influence people's "trust" in government. And third, to what extent do perceptions of corruption affect people's support for democracy (measured as the view that "democracy is always best" even if democracy does not appear to be working)? The widespread, almost consensual acceptance by citizens that democracy is "the only game in town" is seen by students of democratisation as essential to democratic consolidation (Linz & Stepan, 1996).
We do this by examining the bivariate correlations of perceptions of corruption with each of these three key measures. We also use a method known as multiple regression to examine the remaining multivariate correlation once we consider the simultaneous impact of other important factors. Based on previous work (see Mattes and Christie, 1997; Mattes and Thiel, 1998; and Bratton and Mattes, 1999), the most likely factors to also affect these components of democratic legitimacy, and which are present in each of the three surveys are: evaluations of the (1) performance and (2) representativeness of national government; (3) people's expectations of national economic performance; and (4) the extent to which people feel politically competent or efficacious (see Appendix B).
What we find is that perceptions of corruption have consistent, independent impacts on people's satisfaction with democracy. In other words, the more citizens feel that public officials are involved in corruption, the less likely they are to be satisfied with the way democracy works. This is true both for the entire sample, as well as when we segment the sample and examine these relationships among black respondents on one hand, and among white, coloured and Indian respondents on the other.
Perceptions of corruption, however, have no independent impact on support for democracy. Yet satisfaction with the way democracy works does have a strong impact. Thus, corruption to the extent that perceptions of increasing levels of corruption reduce popular satisfaction with democracy, they will also - indirectly - limit the extent to which South Africans reach consensus on democracy "as the only game in town."
Perhaps surprisingly, perceptions of government corruption - at least in these three surveys - have only mixed impacts on trust in government. We find strong relationships in our first survey in 1995, but none in 1997, and only a weak effect in 1998 among black respondents.
Finally, we have not, in this paper, attempted to address linkages between perceptions of corruption and citizen behaviours because our measures of these things are not consistent from survey to survey. However, Idasa's Public Opinion Service has conducted an extensive analysis of our 1997 results for the reasons behind tax morality. This study found that, for white respondents, perceptions of corruption have a significant impact on willingness to pay income tax (Lieberman, 1998).
In this paper, we have reviewed a number of approaches used by Idasa's Public Opinion Service to measure differing elements of citizens beliefs with regard to official corruption. With the passage of time, we have been able to develop nascent time series with regard to several of these measures. The aggregate results in these time series exhibit a satisfying degree of overall stability, indicating an important component of the reliability for our measures. The correlational analysis (both the bivariate demographic correlates, as well as the bivariate and multivariate attitudinal correlates) also yield important relationships among items in each individual survey, as well as relatively stable relationships over time (suggesting inter-item validity as well as reliability). Most overtime changes in aggregate results to individual items or in relationships amongst items, also appear to make sense in light of what empirical democratic theory, or plain political "horse sense" would predict.
However, in our exploration of this data we have confronted some important limitations. Most importantly, if we are to more fully understand the roots of public perceptions of official corruption we need to devote larger portions of our questionnaires to these issues. We need to assess separately the extent to which individuals have been personally involved in corrupt clientele relationships or victimised by corrupt officials (or know people who are, or have been). This requires direct questions about dealings with government officials, but it also requires more indirect questions about general survival strategies (of which corruption may be one type). We also need to ask other sets of questions to help us analytically separate direct or indirect experience with corruption from a general distrust or suspicion of the incumbents of government.
Idasa's Public Opinion Service is currently co-ordinating the first wave of a regional, cross-national project called the Southern African Democracy Barometer. In this project, the questionnaire, drafted jointly by researchers from seven different participating countries, will continue to ask the types of perceptual questions reported in the first part of this paper, but also begin to explore the various ways in which citizens may be subjected to or involved with different types of relationships or networks involving official corruption. For more information on this new project, please contact the authors.
Demographic Correlates of Perceptions of Government Corruption (All South Africans)
Demographic Correlates of Perceptions of Government Corruption (Black Respondents Only)
Demographic Correlates of Perceptions of Government Corruption (White, Coloured and Indian Respondents)
NA = Item Not Asked In That Year
Satisfaction With Democracy (All Respondents)
Trust In Govt (All Respondents)
Commitment to Democracy (All Respondents)
NA = Item Not Asked In That Year
Satisfaction With Democracy (Blacks)
Trust In Govt (Blacks)
Commitment to Democracy (Blacks)
NA = Item Not Asked In That Year
Satisfaction With Democracy (White, Coloured, Indian)
Trust In Govt (White, Coloured, Indian)
Commitment to Democracy (White, Coloured, Indian)
NA = Item Not Asked In That Year
Lieberman, Evan (1998) "Paying for `Us' Versus `Them'? Citizens' Attitudes Toward Income Tax Compliance," POS Reports No. 6 (October)
Linz, Juan and Alfred Stepan (1996) "Towards Consolidated Democracies," Journal of Democracy 7 (January).
Mattes, Robert and Richard Calland. (1996) "Parliamentary Ethics and Government Corruption: Playing With Public Trust," POS Reports No. 3 (February).
Mattes, Robert and Jennifer Christie (1997) "Personal Versus Collective Quality of Life and South Africans' Evaluations of Democratic Government," Social Indicators Research 41.
Mattes, Robert and Hermann Thiel (1998) "Public Opinion and Consolidation In South Africa," Journal of Democracy 9 (January).
Mattes, Robert, Helen Taylor and Cherrel Africa, (1999) "Public Opinion and Voter Preferences," In Election '99 South Africa: From Mandela to Mbeki, edited by Andrew Reynolds.
Rose, Richard, William Mishler and Chrisian Haerpfer (1998) Democracy and Its Alternatives: Understanding Post-Communist Societies (Baltimore: Johns Hopkins University Press).