It takes two to toyi-toyi : one party dominance and opposition party failure in South Africa’s 2019 national election

Full text

RESEARCH ARTICLE
It takes two to toyi-toyi:
1
one party dominance and
opposition party failure in South Africa’s 2019 national
election
Collette Schulz-Herzenberg
a
and Robert Britt Mattes
b
a
Department of Political Science, Stellenbosch University, Stellenbosch, South Africa;
b
Government
and Public Policy, University of Strathclyde, Glasgow, UK, Nelson Mandela School for Public
Governance, University of Cape Town, South Africa
ABSTRACT
Why do dominant parties continue to win elections despite significant governance
failures? Scholars of one party democracies tend to locate explanations at the
macro-level: manipulation of rules, control over state media, or selective
distribution of benefits to supporters. Other scholars emphasize ethnic or religious
identities which trump consideration of policy and performance. We employ a
multinomial regression model of voter decisions in South Africa’s 2019 general
election to explore how the ruling African National Congress managed to secure
58% of the vote amidst a massive corruption scandal and waning public services.
We find that dissatisfied government supporters do not ignore poor performance,
but must perceive a legitimate alternative amongst the opposition before they
switch their vote. Otherwise, they exit the electorate. This allows the governing
party to win significant proportions of a diminishing electorate. Thus, decisions
about whether to vote are not just a result of resources, mobilization or efficacy,
but are also rooted in perceptions of governing and opposition parties. Voter
turnout and vote choice are intimately linked, rather than separate
causal processes. Moreover, continued one-party dominance may be as much a
function of opposition party failure as it is of government control over rules, rents
or resources.
ARTICLE HISTORY Received 9 November 2022; Accepted 6 June 2023
KEYWORDS Party dominance; voter choice; voter turnout; South Africa; opposition parties
Introduction
Why do electorally dominant political parties seem to retain commanding shares of elec-
tion-day support, even when they begin to fail in terms of governing their country?
Common explanations tend to focus on the dominant party’s control of the media
which enables it to broadcast complementary messages and curb access to opposition
party communication and, or control of the state which enables it to shape the rules
of party funding, provide government jobs to key supporters, deliver particularistic
© 2023 Informa UK Limited, trading as Taylor & Francis Group
CONTACT Collette Schulz-Herzenberg csh@sun.ac.za
Supplemental data for this article can be accessed online at https://doi.org/10.1080/13510347.2023.
2228710.
This articlehas been corrected with minor changes.These changes do not impact the academic contentof the article.
DEMOCRATIZATION
https://doi.org/10.1080/13510347.2023.2228710

policy benefits to co-partisans, or manipulate the electoral management body and elec-
toral rules.
2
Relatively few scholars, however, focus on the quality of the choices offered
by opposition parties and the fundamental connection between vote switching (which
can weaken one-party dominance) and voter exiting (which may maintain it).
3
While
the scholarly literature sees voter turnout and voter choice as distinct processes based
on different political, sociological, and psychological dynamics, we argue that in histori-
cally one-party dominant democracies, decisions about who to vote for are intimately
related to decisions about whether to vote. Thus any analysis of one-party dominance
needs to examine voter perceptions of both the governing party as well as opposition
parties. While positive voter evaluations of ruling party performance maintain
support, negative evaluations will lead to vote switching only if dissatisfied voters see
an effective alternative. But if opposition parties are perceived as severely flawed, dissa-
tisfied voters will exit the electorate, and thus enable the ruling party to retain large pro-
portions of election-day vote shares amongst a shrinking electorate.
We test this argument through an analysis of South Africa’s 2019 national legislative
election. We depart from typical research designs in two important ways. First, we
focus on the role of citizens’perceptions of key attributes of political parties, particu-
larly opposition parties. Second, we examine how these images, along with a range of
citizen characteristics and performance evaluations, simultaneously affect decisions
about whether to vote, and for whom to vote. The results demonstrate that perceptions
of opposition parties play a crucial role, even after taking the impact of other typical
variables into account, and are vital to understand how one-party dominance can be
sustained by sharp declines in voter turnout.
South Africa
South Africa provides us with an ideal laboratory in which to test this argument. On
the eve of the 2019 election, its sixth since the transition to democracy, large majorities
of South Africans had become increasingly frustrated with the performance of the
ruling African National Congress (ANC), which had run the country since 1994, as
well as the overall democratic process. While grinding levels of poverty and unemploy-
ment clearly took their toll, the chief culprit was increasing government dysfunction
linked to the 2009–2018 presidency of former President Jacob Zuma. A high profile
judicial commission had recently publicized evidence of widespread corruption invol-
ving the use of state jobs and contracts to reward Zuma’s personal allies and enabled
the widespread theft or diversion of public funds away from the delivery of social
welfare and the maintenance of state institutions (what became known in South
Africa as “state capture”).
4
By the 2019 election, voter identification with the ANC
had fallen by 23 percentage points (from 52% to 29%) since 2006 [Figure 1].
While the ANC responded by trying to distance itself from that period by forcing
Zuma’s resignation, and replacing him with party stalwart, and father of the South
African constitution, Cyril Ramaphosa, one might reasonably have predicted a dire
outcome for the electoral fortunes of the governing party in 2019. Yet the ANC
came away with a resounding victory, winning 58% of the popular vote. To be sure,
this represented a decline of four percentage points since 2014 (62% to 58%, losing
19 of its 249 parliamentary seats). Yet the loss was much smaller than anticipated,
and identical to the party’s losses in 2009 (from 70% to 66%) and 2014 (from 66%
to 62%). Given the events of the previous five years, the ability of the ANC to retain
2C. SCHULZ-HERZENBERG AND R. B. MATTES

the support of almost six-in-ten voters, especially given South Africa’s electoral system
of closed-list proportional representation, is a remarkable achievement.
5
Conventional interpretations of South African electoral dynamics have seen endur-
ing ANC predominance as evidence that strong racial or partisan identities trump
voter considerations of government performance.
6
Yet this misses the fact that the
ANC’s continued electoral dominance in the face of sharp declines in voter identifi-
cation has gone hand-in-hand with a sharp decline in aggregate voter turnout,
which dropped by almost forty percentage points from an estimated 86% of the
voting age population (VAP) in 1994 to just 49% in 2019.
7
While opposition parties collectively increased their overall share of the election-
day vote (from 38% to 42% between 2014 and 2019), this gain was smaller than it
could have been had voter turnout remained stable. Indeed, the election-day vote
share of the Democratic Alliance, the country’s largest opposition party, actually
declined between 2014 and 2019 (from 22% to 21%), despite the ANC’s declining
vote share. But rather than simply examining election-day voting percentages, a
more illuminating approach is to focus on voter choice as a proportion of the
voting age population (VAP). Seen through this prism, ANC voting support
amongst the eligible electorate dropped from 39% in Zuma’sfirst election in 2009,
to just 28% in 2019 [Figure 2].
8
Yet opposition party support as a proportion of
VAP has remained flat over the last three elections (21%). Thus, South Africa’s oppo-
sition parties have not been able to take advantage of heightened dissatisfaction with
the governing party.
At the individual level, data from the 2014 and 2019 South African National Elec-
tion Study post-election surveys suggests that 10% of 2009 voters switched their vote in
2014, and 9% switched between 2014 and 2019. However, levels of vote switching pale
in comparison to the number of voters who exited the electorate between the 2014 and
2019 elections (27%). In 2019, the ANC (27%) and DA (25%) both lost one quarter of
their 2014 voters to the non-voter column, while the EFF (39%) lost a whopping four-
in-ten [See Figure 1 in Supplemental Material]. Thus, voter exiting has far outstripped
vote switching.
9
More importantly, the proportion of voters leaving the electorate has
not been symmetrical over time. Non-aligned voters and opposition identifiers have
been far more likely to exit the electorate than ANC supporters [See Figure 2 in Sup-
plemental Material]. This process of “asymmetric demobilisation”helps explain why
Figure 1. Partisan identification in South Africa, 2005–2018 (Source: Afrobarometer).
DEMOCRATIZATION 3

South Africa’s opposition parties have not significantly increased their support even as
the ANC has become less popular.
Two key factors underlie these trends. On one hand, voter evaluations of ANC per-
formance in government, as well as the image of the party have turned decidedly nega-
tive. On the other hand, voter images of opposition political parties also remain
relatively negative. Since 1994, successive ANC governments have been unable to
create jobs fast enough to reduce poverty or narrow inequality, the issues consistently
identified by large proportions of respondents in Afrobarometer and SANES surveys as
the “most important problem”, in any appreciable way. And, as noted above, the ANC
had presided over steady increases in state corruption, culminating in a judicial com-
mission of enquiry into state capture.
10
Accordingly, voters have consistently given the
ANC government very negative evaluations of its performance in handling the
economy and corruption.
However, the ANC government has, and continues to deliver a great deal of
material benefits to people in the form of affordable houses, steadily expanding
access to basic development infrastructure including sewage, water, electricity,
roads, schools and health clinics, and has created one of the largest social welfare
systems in the world.
11
Thus, in contrast to macro-economics, high levels of public sat-
isfaction with government micro-economic performance provided a great deal of
buoyancy to ANC electoral fortunes, at least between 1994 and 2009 [Figure 3]. But
over the past five years, public evaluations of government performance on these
issues declined considerably as increasing corruption progressively eroded develop-
ment and welfare budgets.
12
Accordingly, ratings of the job performance and trust-
worthiness of the president, parliament and the party itself also declined
considerably. By the time of the 2019 election, only 26 percent of respondents told
SANES interviewers that the ANC had done a “good”or “very good”job running
the country over the previous five years. Thus, the party was more vulnerable than
ever to a capable and strategically savvy opposition.
To be sure, several important factors lie beyond the control of opposition parties.
For instance, the legacy of the struggle against apartheid means that the ANC con-
tinues to command support amongst older voters who credit it with their liberation,
Figure 2. ANC support, opposition support, and non-voters as proportion of VAP, 1994–2019 (Source: Schulz-
Herzenberg, “Trends in Voter Participation”, p. 56).
4C. SCHULZ-HERZENBERG AND R. B. MATTES

possibly trumping any considerations of current performance. Similarly, the electoral
system and party financing system have both historically favoured the ANC to the det-
riment of other parties.
13
But there are several other areas where opposition parties continue to hurt them-
selves. Observers of South African politics point out, for instance, that none of the
opposition parties provide strong evidence of a well-thought-out strategy to court
voters on a continuous basis by using their parliamentary platform or other events
as opportunities to generate free media publicity between elections. Most ignore the
inter-party campaign period as important opportunities for communicating coherent
and credible messages, mounting their campaign in months leading up to the election,
at which point it is far too late to shape or reshape their public image in any significant
way.
14
South Africa’s opposition parties have focused their campaigns on issues of
interest to, at best, a small core of party supporters.
15
With Zuma’s resignation, and rapidly declining levels of voter satisfaction with
government performance, the ANC was at its most vulnerable in 2019, particularly
to a negative campaign focussed on the tangible consequences of the corruption of
the Zuma years: mounting deficits, shrinking development budgets, and neglected
and damaged development infrastructure, such as electricity plants, light rail ser-
vices, and the national airline. Yet the DA responded by running a relatively
anodyne, positive campaign focussed on inclusiveness and national unity, in which
the impact of corruption played a marginal role.
16
The EFF, similarly, ignored the
concerns of the majority of voters, and chose to focus instead primarily on land
redistribution and the nationalization of key industries –issues that few, if any,
South African voters prioritized.
17
Indeed, most voters struggled to say whether
any of the three largest parties (ANC, DA, and the Economic Freedom Fighters)
were best placed to address the most widely cited national issues, unemployment,
crime and corruption.
18
Thus, just 23% of those surveyed after the 2019 election
felt that any opposition party could do a better job than the ANC dealing with the
problem they had listed as most important. Only 32% gave any opposition leader
a higher favourability rating than the ANC’s Ramaphosa. Perhaps most importantly,
just 41% perceived any opposition party to be more inclusive than the ANC. As a
Figure 3. African National Congress –performance evaluations and party images, 2004–2019. (Source: Afroba-
rometer and SANES).
DEMOCRATIZATION 5

result,just31%thoughtthatany opposition party was more trustworthy than the
ANC.
19
Our argument
With few exceptions electoral scholars have tended to model individual decisions
about whether to vote, versus who to vote for, as based on very different processes
and criteria.
20
Standard models see decisions about who to vote for as based on parti-
san identification, evaluations of government performance, ratings of parties and can-
didates, or views of voter and party positions on key campaign issues, as the most
proximally important sources of voter choice.
21
The decision about whether to vote,
in contrast, is generally seen as a result of the differing opportunity costs of voting
associated with age, gender, social class, place of residence, group membership, cogni-
tive sophistication (interest, efficacy and competence), or assessments of the larger
democratic system.
22
However, a different approach, best exemplified in the classic studies of Anthony
Downs
23
and William Riker and Peter Ordeshook
24
sees voter turnout as a function
of comparative evaluations of candidates and parties. That is, voters who see no differ-
ence in the expected utility of voting for the governing party versus any opposition
party have little incentive to bear the costs of registering, following the election cam-
paign, or getting to the polling booth. Downs called this the expected party differential
(the comparison of expected utility income one might receive from the governing
versus opposition parties).
25
When this differential is zero, it is rational to exit the elec-
torate (what Downs called “rational abstention”).
26
In other words, voters who become
dissatisfied with their preferred party, still require a credible alternative in order to
switch their vote.
But the kinds of cleavage structures characteristic of one-party dominant systems
means that providing those voters with credible alternatives may require more than
simply adopting different policy positions. Seymour Martin Lipset and Stein Rokkan
have argued that the cleavage structure produced by the industrial revolution makes
left-right position a valuable cue in Western societies, providing a useful heuristic
through which voters can organize the political world, and on which they can place
themselves and political parties.
27
But in societies that lack this history, voters may
have limited familiarity with the left-right dimension.
28
And where societies have
experienced other types of social revolutions, the electorate might be characterized
by other, more meaningful cleavages, such as Catholic versus Protestant, religious
versus secular, or modern versus traditional.
29
Thus, we think South African voters may use different criteria, focussing on attri-
butes as much as policies, to compare parties and candidates. By party attributes,we
refer to the intrinsic values or images associated with a party that provide voters
with a mental picture of that party
30
, such as “caring,”“competent,”“moderate,”“pro-
gressive,”“unifying”or “inclusive”.
31
Candidate and partisan attributes related to the
key cleavage structures of a society take on particular significance.
32
And where that
cleavage is linked to social identity, party attributes can include candidates’personal
and demographic characteristics such as race, ethnicity, language and accent.
33
These attributes provide voters with low cost informational cues with which to infer
party issue stances, particularly on distributional issues.
34
The implications for
voting behaviour are clear –the “who”a party stands for (in terms of racial identity)
6C. SCHULZ-HERZENBERG AND R. B. MATTES

may be as, or more important to voters than “what”(policy) they stand for, or at least
provide a cue to the policies they stand for. As such, perceptions of attributes are likely
to be less ephemeral and more stable than voter evaluations of government perform-
ance, candidates or policy positions.
35
In South Africa, the deep divisions created by colonialism and apartheid created a
stark and enduring division between the interests and values of the black, African
majority and white, European minority (who constituted the subordinate and super-
ordinate groups in South Africa’s“ranked”society).
36
Because the proportion of
voters previously oppressed under apartheid (plus their children) is so large, any pol-
itical party that had participated in apartheid government structures (or symbolically
connected to those parties) faces a huge challenge in gaining legitimacy amongst the
wider electorate. For their part, the ANC has worked hard to maintain its position
as the champion of the previously oppressed, and to position all other parties on the
other side of that divide.
37
Thus, voter perceptions of whether political parties have
managed to transcend the main apartheid cleavage and appeal to all voters, rather
than one specific group, should constitute an extremely salient attribute of South
Africa’s parties. Therefore, we see inclusiveness as a key attribute for political parties
in South Africa. Indeed, previous studies have shown that voter perceptions of
whether political parties are inclusive (representative of all South Africans) or exclusive
(representative of one group only) exercise an important influence on the vote.
38
Nega-
tive images of parties as racially or ethnically exclusive enclaves overwhelmingly repel
voters.
39
Another key party attribute in a historically one-party dominant system, such as
South Africa, is the perceived ability to govern, or what we call competence. Given
that the ANC has occupied the national government, and government in seven of
the country’s nine provinces since 1994, voters may legitimately wonder whether a
given opposition party would be able to run government if it was ever elected to
power, not only in terms of the skills possessed by existing party leadership, but
whether the party would be able to assemble a sufficiently large number of allied
policy experts to populate the leadership of government departments and agencies.
40
Finally, while party leaders, as candidates, play an important role in any electoral
system, the fact that South Africa’s system of closed-list proportional representation
(with one half the legislature elected on a national ticket) concentrates an exaggerated
amount of attention on the national party leader who, if the party were to win a
majority of seats, would become president.
41
Thus, we also consider the relative favour-
ability of party leaders as an additional important party attribute, particularly for oppo-
sition parties.
Research design and expectations
In order to explore the common driving forces behind both the continued electoral
dominance of the ANC and the rapid decline in voter turnout, we make use of the
2019 South African National Election Survey, conducted as part of the cross-national
Comparative National Elections Project (CNEP).
42
Based on personal interviews of a
nationally representative, multi-stage, stratified, area cluster sample of 1,625 respon-
dents, the questionnaire measures a broad range of voter characteristics, government
evaluations, and candidate and party attributes, and also whether, and how respon-
dents voted in the previous, 2014 national election.
DEMOCRATIZATION 7

We focus on the decisions of voters in the 2019 election who had previously voted
for the ANC in 2014. To construct the dependent variable, we take those 2019 respon-
dents who say they voted for the ANC in 2014 (n = 690) and divide them into three
discrete groups: (1) those who continued to vote for the ANC again in 2019 (“loyalists”
n = 468); (2) those who voted for any opposition party in 2019 (“switchers”n = 49);
and (3) those who exited the electorate in 2019 (“exiters”n = 173). We use multinomial
logistic regression to assess how evaluations of government performance and party
images simultaneously affect both voter choice (particularly vote switching) and
voter turnout (particularly voter exiting) as a mutually contingent, trichotomous
choice while controlling for a range of independent variables that tap other important
theories of both voter turnout and vote choice.
Our analysis proceeds in four stages. In the first stage, we regress our trichotomous
dependent variable on the control variables that measure socio-demographic status,
psychological engagement, organizational membership, and system (dis)affection.
This allows us to account for four different, though related theories of voter
turnout. First, the resource model of voter turnout sees age, education, urban-rural
location, gender and socioeconomic status as providing opportunities to become
informed, register and vote.
43
A second approach focuses on many of the same
factors, such as education, political interest, and political efficacy, but does so in
terms of how they shape cognitive motivations to vote.
44
A third group of scholars
sees the decision to vote as largely a collective act shaped by the social and organiz-
ational context in which individuals are embedded.
45
Decreasing levels of partisan
identification, in particular, have been seen to lie behind decreasing levels of turnout
across western democracies.
46
Afinal group of scholars explains secular declines in
voter turnout as a result of growing voter disaffection with democracy in general,
either in terms of their perceived (in)ability to influence the political system, or (dis)-
satisfaction with democracy.
47
To test these expectations, we include standard measures of demographics factors,
such as age, gender and urban/rural status,aswellasmeasuresofpsychological
engagement such as education, strength of partisanship, interest in politics, internal
political efficacy, and political competence. And, we also include mobilization vari-
ablessuchastradeunionandorganizational membership, and indicators of respon-
dent disaffection with the larger political system, including (dis)satisfaction with
democracy, and external political efficacy (see Supplemental Material for variable
coding).
However, our main argument is that, in historically one party dominant systems,
decisions about whether to vote will also be shaped by the same factors that explain
voter choice. While a long tradition of scholars have seen partisan identification, per-
formance and candidate evaluations as key determinants of vote choice,
48
we argue
that voters who are dissatisfied with the policies, candidates or performance of their
own party will only switch their votes if they perceive a viable alternative. Otherwise,
they will exit the electorate.
Thus, in the second stage of our model, we add voter evaluations of the ANC’s
recent government performance, presidential job performance, and the current state
of the economy. We think respondents’evaluations of ANC performance, on their
own, will have no effect on whether or not they vote. But those respondents who
provide positive evaluations of the performance of the ANC in government over the
previous five years, who are satisfied with current performance of the economy, and
8C. SCHULZ-HERZENBERG AND R. B. MATTES

with the performance of the incumbent president and ANC standard-bearer, Cyril
Ramaphosa, will be very unlikely to shift their vote.
In the third stage, we add respondent perceptions of three party attributes of
opposition parties: the inclusiveness of the main opposition parties; the competence
of all opposition parties relative to the ANC; and the rating of all opposition party
leaders, relative to the ANC presidential candidate.
49
We expect that those respon-
dents who voted for the ANC in 2014, but who now see either of the main opposi-
tion parties as inclusive, see any opposition party as more competent than the ANC,
or rate any opposition leader more favourably than Ramaphosa, will be simul-
taneously less likely to exit the electorate, and more likely to shift their vote to an
opposition party. Finally, in the fourth stage, because our argument is primarily
about the combined effect of dissatisfaction with government performance and per-
ceptions of partisan choice, we re-estimate the third model amongst only those 2014
ANC voters who were currently dissatisfied with ANC performance. We expect
(negative) images of opposition party attributes to have their strongest impacts
amongst this sub-group of voters, contributing mainly to voter exiting from the
electorate.
Results
The multinomial regression models estimate the predictors of two of the three cat-
egories in our dependent variable: those who switch their vote, and those who exit
the electorate, comparing each group to those 2014 ANC voters who remained loyal
and voted for the ANC again in 2019 [Table 1]. Thus, the multinomial models test pre-
dictors of “switchers”(with “loyalists”as the reference category), and predictors of
“exiters”(again, with “loyalists”as the reference).
In terms of vote switching, Stage 1 shows that amongst 2014 ANC voters (column 1
in Table 1), who live in urban areas, and have higher levels of education were more
likely to defect from the party in 2019. As we expected, respondents with higher
levels of internal efficacy, high interest in the campaign, and higher levels of satisfac-
tion with democracy were less likely to switch (and more likely to remain loyal), reflect-
ing the fact that in a one-party dominant system, evaluations of the larger democratic
system are strongly entwined with evaluations of the dominant party. Perhaps surpris-
ingly, amongst 2014 ANC voters, strong party identifiers were neither less nor more
likely to defect in 2019.
In terms of vote exiting in Stage 1 (column 2 in Table 1), younger and male 2014
ANC voters were more likely to exit the electorate in 2019 (compared to those who
remain loyal ANC voters). Respondents with (stronger) partisan ties and those who
were interested in the election campaign were more likely to remain loyal to the
ANC. Respondents with positive views of the democratic system (who think politicians
are interested in their opinions and were satisfied with democracy) were also less likely
to exit and more likely to remain loyal. However, education, and perceptions of
internal efficacy (ability to influence government or understand politics) were not stat-
istically linked with exiting (compared to loyalists).
Overall, the resource, psychological and system (dis)affection variables accounted
for 25% of the variance in our trichotomous dependent variable (Nagelkeke R
2
).
Amongst the 2014 ANC voters, the model correctly predicted 92% of those who
DEMOCRATIZATION 9

Table 1. Explaining vote loyalty, switching and exiting (2014 ANC Voters, 2019).
Models
Step 1:
Demographics / System (dis)affection
Step 2:
Government Performance evaluations
Step 3A:
Party images
Step 3B:
Party images
Column 1 Column 2 Column 3 Column 4 Column 5 Column 6 Column 7 Column 8
Voter Switchers (2)
vs Loyalists (1)
Voter Exiters (3) vs
Loyalists (1)
Voter Switchers (2)
vs Loyalists (1)
Voter Exiters (3) vs
Loyalists (1)
Voter Switchers (2)
vs Loyalists (1)
Voter Exiters (3) vs
Loyalists (1)
Loyalists (2) vs
Switchers (1)
Voter Exiters (3) vs
Switchers (1)
Intercept −2.511** (.961) 1.260* (.541) −1.333 (1.030) 1.204* (.546) −9.551*** (2.443) .306 (.664) 9.551*** (2.443) 9.857*** (2.462)
Social
Demographics
Age .001 (.013) −.024*** (.008) −.006 (.013) −.025*** (.007) .010 (.017) −.021** (.008) −.010 (.017) −.031 (.017)
Education .226* (.097) .018 (.063) .241* (.098) .016 (.061) .164 (.124) −.009 (.064) −.164 (.124) −.173 (.128)
Gender (Male) .505 (.339) .862*** (.206) .580 (.341) .854*** (.200) .942* (.420) .886*** (.209) −.942* (.420) −.056 (.429)
Urban-rural (Urban) 1.348* (.547) .083 (.227) 1.210* (.532) .073 (.222) 1.257* (.631) .059 (.230)−1.257* (.631) −1.198 (.644)
Trade union
membership (Yes)
.280 (.472) −.510 (.371) .244 (.469) −.522 (.359) .773 (.563) −.473 (.369) −.773 (.563) −1.245* (.615)
Organizational
membership (Yes)
.311 (.347) −.346 (.215) .418 (.348) −.358 (.210) .452 (.426) −.433* (.219) −.452 (.426) −.885* (.441)
Strength of partisan
identification
−.200 (.141) −.170* (.086) −.172 (.139) −.174* (.085) −.189 (.175) −.195* (.089) .189 (.175) −.006 (.179)
Cognitive
Involvement
People like me able to
influence
government
−.427** (.149) −.090 (.084) −.425 (.149) −.078 (.082) −.363* (.170) −.106 (.086) .363* (.170) .257 (.173)
Politics is not
complicated
.219 (.150) .126 (.097) .182 (.149) .121 (.096) .008 (.170) .116 (.100) −.008 (.170) .108 (.172)
Interested in the
campaign
−.291* (.153) −.396*** (.091) −.311*** (.151) −.399*** (.088) −.139 (.188) −.392*** (.092) .139 (.188) −.253 (.193)
System
(Dis)Affection
Politicians care about
what people like me
think
−.293 (.173) −.202* (.097) −.224 (.174) −.205* (.095) −.098 (.212) −.179 (.100) .098 (.212) −.081 (.217)
Satisfied with
democracy
−.281* (.111) −.144*(.065) −.173 (.114) −.151* (.066) −.107 (.142) −.156* (.069) .107 (.142) −.049 (.147)
10 C. SCHULZ-HERZENBERG AND R. B. MATTES

Government
Performance
Evaluations
Positive rating of ANC
performance in
government, last 5
years
−.794*** (.178) −.034 (.087) −.297 (.258) .222 (.120) .297 (.258) .519* (.263)
Positive rating of job
performance of
Ramaphosa
−.108 (.153) −.052 (.099) .232 (.199) .065 (.106) −.232 (.199) −.167 (.206)
Positive rating of
national economic
conditions
.096 (.177) .040 (.099) .297 (.213) .038 (.103) −.297 (.213) −.259 (.219)
Party Images
DA Inclusive image .535* (.241) .052 (.127) −.535* (.241) −.484* (.248)
EFF inclusive image −.072 (.244) −.145 (.128) .072 (.244) −.074 (.250)
Opposition party
more competent
than ANC
4.168* (1.772) .905*** (.277) −4.168* (1.772) −3.263 (1.779)
Opposition leader
more favourable
than Ramaphosa
2.933*** (.504) .268(.263) 2.933*** (.504) −2.665*** (.520)
Cases/respondents 690 690 690 690
Nagelkerke R
2
.253 .291 .417 .417
Cox & Snell R
2
.202 .232 .333 .333
Model Significance *** p< .001 *** p< .001 *** p< .001 *** p< .001
Loyalists Correctly
Predicted
92% 93% 91% 91%
Switchers Correctly
Predicted
7% 14% 50% 50%
Abstainers Correctly
Predicted
31% 29% 32% 32%
*p= <.05, ** p= <.01, *** p= <.001. Cells report multinomial logistic regression coefficients (with standard errors in parentheses). Question items and response coding plus odds ratios
and confidence intervals are available in the supplemental material.
DEMOCRATIZATION 11

voted ANC again in 2019, 31% of those who exited the electorate, but only 7% of those
who voted for an opposition party.
In Stage 2, we add citizen evaluations of performance. As expected, while perceptions
of government performance have no impact on whether voters leave the electorate (versus
remain loyal), they have a strong impact on whether they switch their vote. That is, 2014
ANC voters who in 2019 thought the government had done a good job running the
country since 2014, are substantially less likely to defect (given an Odds Ratio (OR) of
0.452, 58% less likely for each incremental gain in approval).
50
Put in the converse, in con-
trast to those scholars who argue that racial solidarity or partisanship prevent South
African voters from basing their vote on actual performance
51
those ANC voters who
subsequently perceived declines in the quality of government are far more likely to
switch to an opposition party, rather than remain loyal. We also see that the impact of
system (dis)affection (satisfaction with democracy) loses significance as a predictor of
vote switching, but not voter exiting. Those who are satisfied with democracy are less
likely to exit than those who are dissatisfied (OR = 0.860, or 14% less likely for each incre-
mental gain in satisfaction). Overall, the addition of performance considerations increases
the explanatory power of our model somewhat (Nagelkeke R
2.
0.253 to 0.291), and makes
a slight improvement in the correct prediction of switchers (from 7% to 14%).
However, the addition of voter images of opposition party attributes (relative to the
governing party) in Stage 3 leads to a substantial increase in predictive power. Column
5 in Stage 3A displays the impact of party images on vote choice. 2014 ANC supporters
are far more likely to switch their vote to an opposition party (rather than remain loyal)
if they see any opposition party as more competent than the ANC (almost 65 times
more likely, O.R. = 64.580), or rate any opposition leader more favourably than Presi-
dent Ramaphosa (almost 19 times more likely, O.R. = 18.775). But racial imagery also
matters. Those ANC voters who perceive the main opposition party, the DA, as inclus-
ive (“representative of all South Africans”) are seven times more likely to switch their
vote (O.R. = 1.708). The problem for opposition parties, of course, is that relatively few
2014 ANC voters held these views.
In terms of voter turnout, the perception amongst 2014 ANC voters that any oppo-
sition party is better able to govern the country than the ANC matters. But in apparent
contrast to our expectations, the sign in column 6 isactually positive, seeming to suggest
that those who see an effective alternative are actually more likely to exit the electorate.
However, this pairwise comparison estimates the effect relative to loyalists, which is not
the comparison that matters here. What we are really interested in is the impact of an
effective alternative on the choice between vote switching and vote exiting. This choice,
then, is modelled in the Stage 3B, column 8,
52
and here we see that those 2014 ANC
voters who rate any opposition party leader more highly than Ramaphosa are far less
likely to exit the electorate and far more likely to switch their vote. Based on the
Odds Ratio of 0.07, a positive rating of any opposition leader (relative to Ramaphosa)
decreases the odds of exiting (rather than switching) by 93%. Likewise, those who per-
ceive the DA as inclusive are 38% less likely to exit the electorate (rather than switch)
(OR = (0.616)). Conversely, those respondents who see the DA as representative of one
group only, or who don’t know who the DA represents are correspondingly more likely
to exit. In other words, 2014 ANC voters who saw no effective partisan choice in 2019
were far more likely to leave their electorate, than switch their vote.
Importantly, the inclusiveness/exclusiveness of the populist EFF has no impact
amongst these respondents. This suggests that for a self-avowedly “pro-black”party
12 C. SCHULZ-HERZENBERG AND R. B. MATTES

such as the EFF, this attribute does not matter to ANC voters, while it is a major issue
for a historically “white”party like the DA. Also, as importantly, evaluations of the per-
formance of the ANC government over the past five years lose statistical significance.
Thus, amongst those voters who had gone to the polls five years earlier to vote for the
governing party, it is views of the attributes of the opposition parties and candidates
(rather than ratings of government performance) that determines whether or not
they vote again, and whether or not they switch their vote.
The addition of the four opposition party variables to tap perceptions of electoral
choice in Stage 3 increases the overall ability to discriminate between vote exiters,
switchers and loyalists by more than half (from 0.291 to 0.417). Moreover, the rate
of correct predictions of switchers increases substantially, from 14% to 50%. Thus,
in South Africa’s evolving one party dominant system, vote switching appears to be
much more a function of the choices offered by opposition parties than the perform-
ance of the governing party [for a more intuitive interpretation of the odds ratios, we
provide plots of the estimated probabilities of vote switching for each of the party
image variables in Figure 3 of the Supplemental Material].
Up to this point the models have displayed the separate and independent effects of the
various independent variables amongst respondents who voted for the ANC in the pre-
vious election (n = 690). At its core, however, our argument pertains most directly to the
behaviour of those 2014 ANC voters who had become dissatisfied with the ANC in gov-
ernment by the time of the 2019 election. Thus, in the fourth stage of our analysis, we
focus only on those 2014 ANC voters who were also dissatisfied with the performance
of the ANC over the previous five years (n = 302) [Table 2]. While caution is required
when interpreting results based on small sub-samples, the power of voter images of oppo-
sition party attributes is evident even amongst this far more restricted sample. In terms of
vote choice (column 1, Stages 4A and column 3 Stage 4B), dissatisfied 2014 ANC voters
who rated any party leader more favourably than Ramaphosa were almost 14 times more
likely to switch, rather than remain loyal (OR =13.881), and far less likely to exit the elec-
torate rather than switch (OR = .047). Dissatisfied ANC supporters who saw any opposi-
tion party as better able to govern the country (competence) also displayed the same
patterns
53
(for a visual illustration of the size of these effects, see the plotted predicted
probabilities in Figure 4 in the Supplemental Material). While the samples differ, the
explanatory power of this model is even stronger amongst dissatisfied 2014 ANC
voters, than 2014 ANC voters in general (Nagelkerke R
2
= .496, compared to .417). More-
over, opposition party images correctly predict 68% of those dissatisfied 2014 ANC voters
who switched their vote in 2019, and 41% of those who abstained.
Implications
Our analysis of voter decisions in South Africa’s 2019 election provides compelling evi-
dence for the powerful effects of negative voter images of opposition parties. For many
years, analysts of South African politics explained the electoral dominance of the ANC
as the result of strong racial or partisan identities which prevented voters from consid-
ering switching their vote.
54
However, the evidence shows that erstwhile ANC suppor-
ters who are dissatisfied with government performance are willing to at least consider
switching. But to consummate that switch, dissatisfied supporters must see one or
more opposition parties, or their leading candidates, as an effective or legitimate
alternative. Otherwise, withdrawal from the electorate becomes a rational alternative,
DEMOCRATIZATION 13

especially for citizens who find it difficult to follow political affairs, have no interest in
election campaigns, or are disenchanted with the larger democratic system.
55
This dynamic helps explain how a dominant political party such as the African
National Congress has maintained a preponderance of election-day voters even in
the face of mounting voter dissatisfaction and disillusionment. The ability of disillu-
sioned ANC voters to hold the governing party to account has been obstructed by
their negative views of opposition parties whom they understand to be ill-equipped
to lead, govern or represent them, which then lead to abstention rather than vote
switching. Thus, while most scholars have pointed to recalcitrant voters unwilling to
hold the ANC to account as the key problem of South Africa’s struggling democracy,
the results of this analysis points the finger at the opposition parties.
In the last few years, increasing numbers of South African analysts have come to
assume that the continued failures of the ANC government portend significant
Table 2. Explaining vote loyalty, switching and exiting (Dissatisfied 2014 ANC Voters, 2019).
Models Step 4A Step 4B
Column 1 Column 2 Column 3 Column 4
Voter Switchers (2)
vs
Loyalists (1)
Voter Exiters (3)
vs
Loyalists (1)
Voter Loyalists (2) vs
Switchers (1)
Voter Exiters (3) vs
Switchers (1)
Intercept −22.182*** (1.471) −.009 (1.173) 21.182*** (1.584) 21.173*** (1.597)
Social Demographics
Age .003 (.018) −.018 (.013) −.003 (.018) −.020 (.020)
Education .237 (.142) .247*(.097) −.237 (.142) .009 (.149)
Gender (Male) .597(.482) .739* (.329) −.597(.482) .142 (.515)
Urban-rural .254 (696) −.818* (.374) −.254 (696) −1.072 (.733)
Trade union membership −.007 (.794) .493 (.544) .007 (.794) .500 (.849)
Organizational membership .931* (.493) −.668 (.373) −.931* (.493) −1.599** (.554)
Strength of partisan
identification
−.024 (.205) −.149 (.145) .024 (.205) −.125 (.218)
Cognitive Involvement
People like me able to
influence government
−.382 (.206) −.155 (.136) .382 (.206) .227 (.212)
Politics is not complicated .033 (.198) .201 (147) −.033 (.198) .168 (.202)
Interested in the campaign −.043 (.219) −.445** (.153) .043 (.219) −.402 (.238)
System (Dis)Affection
Politicians care about what
people like me think
−.058 (.258) −.061 (.158) .058 (.258) −.002 (.268)
Satisfied with democracy −.102 (.158) −.288** (.110) .102 (.158) −.186 (.173)
Party Images
DA Inclusive image .502 (.283) .245 (.202) −.502 (.283) −.258 (.306)
EFF inclusive image −.244 (.305) −.481*(.211) .244 (.305) −.237 (.322)
Opposition party more
competent than ANC
18.288 (.000) 1.601** (.599) −17.288*** (.599) −15.687 (.000)
Opposition leader more
favourable than Ramaphosa
2.630*** (.536) −.429 (.413) −2.630*** (.536) −3.060*** (.599)
Cases/respondents 300 300
Nagelkerke R
2
.496 .496
Cox & Snell R
2
.419 .419
Model Significance .001 .001
Loyalists Correctly Predicted 85% 85%
Switchers Correctly Predicted 68% 68%
Abstainers Correctly Predicted 41% 41%
*p= <.05, **p= <.01, ***p= <.001. Cells report multinomial logistic regression coefficients (with standard
errors in parentheses). Question items and response coding plus odds ratios and confidence intervals are avail-
able in the supplemental material.
14 C. SCHULZ-HERZENBERG AND R. B. MATTES

change after the country’s 2024 election, ranging from an ANC-led coalition
government necessitated by its failure to win 50% of the popular vote, to an opposi-
tion-led government resulting from an outright defeat.
56
However, the results of our
analysis suggest that unless opposition parties are able to do a better job convincing dis-
satisfied voters that they are a legitimate alternative, the ANC will continue to win a large
proportion of a progressively shrinking pool of election-day voters, and dissatisfaction
with government output will turn into dissatisfaction with the democratic regime.
57
More generally, our results suggest that scholars of voter behaviour in other countries
and regions need to consider the possibility that voter decisions about whether to vote,
and who to vote for, are not distinct processes based on different political, sociological
and psychological dynamics, but are connected and intertwined processes that are
both affected by voter evaluations of governing and especially opposition parties.
These connections may not be obvious or important in competitive party systems
where dissatisfied voters feel able to transfer their support to other political parties.
However, we think they are essential to understanding electoral trends in several
other types of systems: (1) where society is characterized by historic or structural con-
ditions that create sharp and asymmetrical cleavages between voters; (2) where parties
that led, or claim to have led the resistance to colonial or indigenous authoritarian
regimes win the inaugural democratic election and then subsequently portray opposition
parties as creatures of the ancien regime; (3) where governing parties claim to embody
the core of a new nation against oppositionparties who represent theethno-regional per-
iphery; or (4) poor societies where governing parties claim to be the champion of devel-
opment and modernization against the forces of tradition and reaction.
In such systems, opposition parties face daunting, though not insurmountable, chal-
lenges to presenting themselves as viable alternative homes for dissatisfied voters. But if
South Africa serves as any kind of example, opposition parties often make things even
worse through their selection of candidates, prioritization of issues, and conduct of the
campaign. Thus, scholars of one party dominant party systems need to widen their ana-
lytic lens to include the role of opposition parties. Without such a shift, we will continue
to interpret one-party dominance as the result either of governing party manipulation
of rules and resources, or of stubborn voters impervious to new information, and miss
the crucial role of the strategic choices of opposition parties.
Figure 4. Opposition party images, 2004–2019. (Source: Afrobarometer and SANES).
DEMOCRATIZATION 15

Notes
1. The toyi-toyi is a rhythmic, high-kneed, foot-stomping southern African dance, punctuated by
chanting and is a powerful statement of both protest and celebration. It is associated in South
Africa with marches and rallies held by the African National Congress or allied organizations.
2. Pempel, “Conclusion”; Giliomee and Simkins, The Awkward Embrace; Greene, “Dominant
Party Strategy”; De Jager and Du Toit, Friend or Foe; Southall, Liberation Movements.
3. While some scholars explore how the failings of African opposition parties might sustain one-
party dominance, opposition weakness is attributed to institutional and structural factors (see
Lindberg, “Opposition parties”; Rakner and van der Walle, “Democratization by elections?”).
Few studies explore vote switching in African elections (Lindberg and Morrison, “Exploring
voter alignments”) and fewer address how voter perceptions of opposition parties affect party
dominance (see Hanson, Post-Imperial Democracies;PlatasandRaffler, “Closing the Gap”).
4. Pauw, The President’s Keepers;Haffajee and Chipkin, Days of Zondo; Chipkin and Swilling,
Shadow State; Judicial Commission of Inquiry into State Capture Report: part 1 https://
www.gov.za/sites/default/files/gcis_document/202201/judicial-commission-inquiry-state-
capture-reportpart-1.pdf
5. Ramaphosa’s“new dawn”agenda of renewal and anti-corruption was an electoral asset for the
ANC without whom the party’s result may have been worse. Also see Schulz-Herzenberg, “The
2019 National Election”.
6. Horowitz, Ethnic Groups; Horowitz, A Democratic South Africa?; Johnson and Schlemmer,
Launching Democracy; Giliomee, “South Africa’s emerging”; Giliomee et al., “Dominant
Party Rule”; Harris, Everyday Identity.
7. Schulz-Herzenberg, “Trends in Voter Participation,”53.
8. Ibid., 56.
9. Bekker et al. “Beyond the Binary: Examining Dynamic Youth Voter Behaviour in South
Africa,”297–317. found evidence of limited vote switching among young voters. However,
the majority of young voters abstain.
10. See endnote 4.
11. Lieberman and Lekalake, “South Africa’s Resilient Democracy”.
12. In his 2021 Medium-Term Budget Policy Statement the country’sfinance minister Enoch God-
ongwana warned that rampant corruption was draining public finances: https://
theconversation.com/south-africas-corruption-busters-short-changed-on-funding-and-
political-commitment-173072. In September 2017, former Finance Minister Pravin Gordhan
estimated the cost of state capture at 250 billion rand (https://www.citizen.co.za/news/south-
africa/1651069/r250bn-lost-to-state-capture-in-the-last-three-years-says-gordhan/); The
Daily Maverick estimated that state capture cost roughly R1.5 trillion (https://www.
dailymaverick.co.za/article/2019-03-01-state-capture-wipes-out-third-of-sas-r4-9-trillion-
gdp-never-mind-lost-trust-confidence-opportunity/). The South African Reserve Bank found
state capture reduced GDP growth by 4% a year (https://www.news24.com/fin24/Economy/
damage-from-state-capture-worse-than-suspected-sarb-20190606).
13. Mattes et al., “South African Parties”.
14. Mattes, “The 2014 Election”; Africa, “Do Election Campaigns Matter”; Southall, “Opposition in
South Africa”; Maloka, “‘White’Political Parties.”
15. Africa, “Do Election Campaigns Matter.”
16. https://www.da.org.za/campaigns/manifesto.
17. https://www.sahistory.org.za/archive/eff-2019-election-manifesto. According to Afrobarom-
eter, land redistribution was seen as a key issue by 7% of South Africans in 2018. See
Nkomo, “Land Redistribution”.
18. Schulz-Herzenberg, “The 2019 National Election,”185.
19. Large proportions of the electorate either see opposition parties as exclusive but similar pro-
portions say they simply do not know enough about opposition parties to conclude one way
or the other (29 percent for the DA, 35 percent for the EFF and 56 percent for the IFP). For
parties that have been in existence for a long time (the EFF was six years old, while the DA,
IFP and FF+ were all at least 25 years old), these are damning indictments of their strategies.
20. A handful of scholars stress the importance of using single unified voter decision models
that simultaneously account for choice and participation and argue that the literature has
16 C. SCHULZ-HERZENBERG AND R. B. MATTES

largely neglected this relationship. See Weschle, “Two Types”;Söderlund,“Retrospective
Voting”;Tillman,“Economic Judgments”;Pierce,“Modelling Electoral”;Thurnerand
Eymann, “Policy-specificAlienation”;LacyandBurden,“The Vote-stealing”. Fewer still
argue that within this unified model the turnout decision may be dependent on voter
evaluations of the parties that stand for election (exceptions include Weschle, “Two
types,”52).
21. Campbell et al., The American Voter; Fiorina, Retrospective Voting; Dalton, Citizen Politics.
Scholars also identify a set of more distal factors such as parental and peer socialization, and
group membership, and political values.
22. Blais, “Turnout in Elections”; Feddersen, “Rational Choice Theory”; Norris, Democratic Deficit;
Rolfe, Voter Turnout.
23. Downs, An Economic Theory.
24. Riker and Ordeshook, “A Theory”.
25. Downs, An Economic Theory,39
26. Ibid., 260.
27. Lipset and Rokkan, Party Systems.
28. In South Africa, for instance, an average of 32 percent over successive post-election surveys
have told SANES interviewers that they do not use, or do not understand these terms. In
the 1994 South African National Election Study post-election survey, 40 percent of respondents
say they did not use or understand these terms; in 2004, 31 percent gave the same answer,
declining to 25 by the 2014 survey.
29. For evidence of how voter cleavages vary, even within industrialized states, depending on those
societies’social revolutions, see Rose and Urwin, “Social cohesion”.
30. Trilling, Party Image.
31. Rose and McAllister, The Loyalties of Voters, 132.
32. Richardson, “European Party Loyalties,”23.
33. Butler and Stokes, Political Change; Popkin, The Reasoning Voter; Rose and McAllister, The
Loyalties of Voters; Dalton, Citizen Politics.
34. Popkin, The Reasoning Voter; Chandra, Why Ethnic Parties; Green et al., Partisan Hearts;
Dalton, Citizen Politics.
35. Popkin, The Reasoning Voter.
36. Horowitz, Ethnic Groups in Conflict.
37. Ferree, Framing the Race.
38. Mattes, The Election Book;Mattesetal.,“Judgment and Choice”; Mattes and Piombo, “Opposi-
tion Parties”;Ferree“Explaining South Africa’s”;Ferree,Framing the Race; Schulz-Herzenberg,
“Trends”; Habib and Schulz-Herzenberg, “Democratization”;Gordon,“Racial Animosity”.
39. Mattes, The Election Book; Schulz-Herzenberg, “Trends in Party Support”.
40. For evidence of competence evaluations see Mattes et al., “Judgment and Choice”; Mattes and
Piombo, “Opposition Parties”; Southall, “Opposition in South Africa”; Maloka, “’White’Politi-
cal Parties”.
41. The picture of the presidential candidate also appears next to the party symbol on election
ballots. For evidence of leadership evaluations see Mattes et al., “Judgment and Choice”;
Mattes and Piombo, “Opposition Parties.”
42. CNEP is a cross-national project based at the Mershon Centre, Ohio State University which
conducts post-election surveys of democratic elections in 27 countries, including four elections
in South Africa (2004, 2009, 2014 and 2019). For datasets and reports see: https://u.osu.edu/
cnep/ and https://www.datafirst.uct.ac.za/.
43. Niemi and Weisberg, Classics in Voting; Teixeira, “Turnout in the 1992”; Blais, “Turnout in
Elections”.
44. Blais, “Turnout in Elections;”Norris, Democratic Phoenix.
45. Huckfeldt and Sprague, “Networks in Context”; Beck et al., “The Social Calculus”; Franklin,
Voter Turnout; Richardson and Beck, “The Flow of Political”; Magalhães et al., “Mobilisation”.
46. Farrell et al., “The Changing British Voter”; Blais et al., “Measuring Party Identification”;
Dalton and Wattenberg, Parties Without Partisans; Lachat, “Explaining Electoral Volatility”;
Rattinger & Wiegand, “Volatility on the Rise?”.
47. Campbell et al., The American Voter; Key, The Responsible Electorate; Fiorina, Retrospective
Voting; Popkin, The Reasoning Voter; Norris, Critical Citizens; Torcal and Montero Political
DEMOCRATIZATION 17

Disaffection; Berelson et al., Voting; Zelle, “Social Dealignment”; Dalton and Weldon, “Public
Images”; Dassonneville, Electoral Volatility; Söderlund, “Retrospective Voting”; Dassonneville
et al., “Staying With the Party”.
48. Campbell et al., The American Voter; Key, The Responsible Electorate; Fiorina, Retrospective
Voting; Popkin, The Reasoning Voter.
49. In other models, we included a rating of trustworthiness of opposition parties relative to gov-
ernment. This measure was highly statistically significant and overpowered the impact of other
opposition party ratings suggesting that trust in a party is itself an outcome of inclusiveness,
competence, and an attractive leader. We include these variables instead because they offer
more granular views of opposition party attributes.
50. The Odds Ratio provides the probability of exiting for each unit change in a respondent’s
approval of government performance, which has four ascending categories.
51. Giliomee et al., “Dominant Party Rule”; De Kadt and Lieberman, “Nuanced Accountability”.
52. The estimates in Step 3B Column 7 are the inverse of Step 3A Column 5 because the compari-
son groups are the same with the categories reversed.
53. Due to an empty cell, there are issues with the calculation of some statistical results for this
independent variable (see Table 2 in the supplemental material). Amongst dissatisfied 2014
ANC voters who did not see any opposition party as competent to govern, absolutely no
respondents switched their vote. Thus, while the model generates logistic regression coeffi-
cients, it does not calculate a standard error or p-value in Table 2, columns 1 and 4, and gen-
erates uninterpretable odds ratios in columns 1, 3 and 4.
54. Johnson and Schlemmer, Launching Democracy; Giliomee et al., “Dominant Party Rule”.
55. Voter behaviour may not necessarily follow the same trends at the local level. Mirroring
national trends, the 2021 municipal elections also witnessed a sharp decline in overall voter
turnout. But there was also some evidence of volatility in aggregate vote shares, and possible
vote switching in a limited number of metros such as the cities of Johannesburg and
Tshwane. We attribute this to the effect of the mixed electoral system in municipal elections
where voters are able to choose amongst individual candidates for a councillor in single
member wards, as well as choose amongst parties in a proportional representation ballot. In
Johannesburg, for example, most small parties won greater numbers of votes for their ward
candidates than for the party, suggesting that local candidates were able to use direct connec-
tion to voters to draw support away from the larger parties. However, this dynamic was evident
in only a few large urban councils.
56. Mashele and Qobo, The Fall of the ANC; Joubert, Who Will Rule;Haffajee “ANC’sCollapse”;
Mattisonn, “Who Will Eat”.
57. Mattes and Piombo, “Opposition Parties,”125.
Acknowledgements
We would like to thank Paul Beck, Russell Dalton, Narisong Huhe, Nicholas Kerr, Jaime Bleck, Robert
Nyenhuis, Matthias Kroenke, Thomas Schober, and Thomas Isbell and the panel audiences at the 2022
Annual Conference of the American Political Science Association, the 2022 Annual Conference of the
Election, Public Opinion and Parties Specialist Group of the British Political Studies Association, and
the 2020 Annual Conference of the World Association of Public Opinion Research, for insightful
comments and suggestions.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Funding
This work was supported by the South African National Research Foundation: [Grant Number
118512].
18 C. SCHULZ-HERZENBERG AND R. B. MATTES

Notes on contributors
Collette Schulz-Herzenberg is Associate Professor in the Department of Political Science at the Stel-
lenbosch University, South Africa.
Robert Mattes is Professor of Government and Public Policy at the University of Strathclyde, and
Adjunct Professor in the Nelson Mandela School for Public Governance at the University of Cape Town.
Bibliography
Africa, Cherrel. “Do Election Campaigns Matter in South Africa? An Examination of Fluctuations in
Support for the ANC, DA, IFP and NNP 1994–2019.”Politikon 46, no. 4 (2019): 371–389.
Beck, Paul, Russell Dalton, Steven Greene, and Robert Huckfeldt. “The Social Calculus of Voting:
Interpersonal, Media, and Organizational Influences on Presidential Choice.”American Political
Science Review 96, no. 1 (2002): 57–73.
Bekker, Martin, Carin Runciman, and Benjamin Roberts. “Beyond the Binary: Examining Dynamic
Youth Voter Behaviour in South Africa.”Politikon 49, no. 4 (2022): 297–317.
Berelson, Bernard, Paul Lazarsfeld, and William McPhee. Voting. A Study of Opinion Formation in a
Presidential Campaign. Chicago: The University of Chicago Press, 1954.
Blais, Andre. “Turnout in Elections.”In The Oxford Handbook of Political Behaviour, edited by Russell
Dalton, and Hans-Dieter Klingemann, 622–635. Oxford: Oxford University Press, 2009.
Blais, André, Elisabeth Gidengil, Richard Nadeau, and Neil Nevitte. “Measuring Party Identification:
Britain, Canada, and the United States.”Political Behavior 23, no. 1 (2001): 5–22.
Butler, David, and Donald Stokes. Political Change in Britain: Basis of Electoral Choice. New York: St.
Martin’s Press, 1969.
Campbell, Angus, Philip Converse, Warren Miller, and Donald Stokes. The American Voter. Chicago:
University of Chicago Press, 1960.
Chandra, Kanchan. Why Ethnic Parties Succeed. Patronage and Ethnic Head Counts in India.
Cambridge: Cambridge University Press, 2004.
Chipkin, Ivor, and Mark Swilling. Shadow State: The Politics of State Capture. Johannesburg: Wits
University Press, 2018.
Dalton, Russell. Citizen Politics: Public Opinion and Political Parties in Advanced Industrial
Democracies. Washington, DC: CQ Press, 2013.
Dalton, Russell. The Apartisan American: Dealignment and Changing Electoral Politics. Washington,
DC: CQ Press, 2013.
Dalton, Russell, Ian McAllister, and Martin Wattenberg. “The Consequences of Partisan
Dealignment.”In Parties Without Partisans: Political Change in Advanced Industrial
Democracies, edited by Russell Dalton, and Martin Wattenberg, 37–63. Oxford: Oxford
University Press, 2000.
Dalton, Russell, and Martin Wattenberg. In Parties Without Partisans. Political Change in Advanced
Industrial Democracies. Oxford: Oxford University Press, 2002.
Dalton, Russell, and Steven Weldon. “Public Images of Political Parties: A Necessary Evil?”West
European Politics 28, no. 5 (2005): 931–951.
Dassonneville, Ruth. “Electoral Volatility, Political Sophistication, Trust and Efficacy: A Study on
Changes in Voter Preferences During the Belgian Regional Elections of 2009.”Acta Politica 47,
no. 1 (2012): 18–41.
Dassonneville, Ruth, André Blais, and Yves Dejaeghere. “Staying With the Party, Switching or Exiting?
A Comparative Analysis of Determinants of Party Switching and Abstaining.”Journal of Elections,
Public Opinion and Parties 25, no. 3 (2015): 387–405.
De Jager, Nicola, and Pierre Du Toit, eds. Friend Or Foe? Dominant Party Systems in Southern Africa.
Cape Town: United Nations University Press, 2013.
De Kadt, Daniel, and Evan Lieberman. “Nuanced Accountability: Voter Responses to Service Delivery
in Southern Africa.”British Journal of Political Science 50, no. 1 (2020): 185–215.
Downs, Anthony. An Economic Theory of Democracy. New York: Harper, 1957.
Farrell, David, Ian McAllister, and David Broughton. “The Changing British Voter Revisited: Patterns
of Election Campaign Volatility Since 1964.”British Elections and Parties Yearbook 4, no. 1 (1994):
110–127.
DEMOCRATIZATION 19

Feddersen, Timothy. “Rational Choice Theory and the Paradox of Not Voting.”Journal of Economic
Perspectives 18, no. 1 (2004): 99–112.
Ferree, Karen. “Explaining South Africa’s Racial Census.”The Journal of Politics 68, no. 4 (2006): 803–
815.
Ferree, Karen. Framing the Race in South Africa: The Political Origins of Racial-Census Election.
Cambridge: Cambridge University Press, 2010.
Fiorina, Morris. Retrospective Voting in American Presidential Elections. New Haven: Yale University
Press, 1981.
Franklin, Mark. Voter Turnout and the Dynamics of Electoral Competition in Established Democracies
Since 1945. Cambridge: Cambridge University Press, 2004.
Giliomee, Hermann. “South Africa’s Emerging Dominant-Party Regime.”Journal of Democracy 9, no.
4 (1998): 128–142.
Giliomee, Hermann, James Myburgh, and Lawrence Schlemmer. “Dominant Party Rule, Opposition
Parties and Minorities in South Africa.”Democratization 8, no. 1 (2001): 161–182.
Giliomee, Hermann, and Charles Simkins. In The Awkward Embrace: One-Party Domination and
Democracy. Cape Town: Tafelberg, 1999.
Gordon, Steven. “Racial Animosity and Political Party Partisanship in South Africa: The Case of the
African National Congress and the Black African Majority.”Democratization 25, no. 2 (2018): 351–
368.
Green, Donald, Bradley Palmquist, and Eric Schickler. Partisan Hearts and Minds: Political Parties and
the Social Identities of Voters. New Haven: Yale University Press, 2002.
Greene, Kenneth. “Dominant Party Strategy and Democratization.”American Journal of Political
Science 52, no. 1 (2008): 16–31.
Habib, Adam, and Collette Schulz-Herzenberg. “Democratization and Parliamentary Opposition in
Contemporary South Africa: The 2009 National and Provincial Elections in Perspective.”
Politikon 38, no. 2 (2011): 191–210.
Haffajee, Ferial. “ANC’s Collapse as South Africa’s Majority Party Is Foretold in New Poll.”Daily
Maverick, August, 14 2022 https://www.dailymaverick.co.za/article/2022-08-14-ancs-collapse-as-
south-africas-majority-party-is-foretold-in-new-poll/.
Haffajee, Ferial, and Ivor Chipkin. Days of Zondo: The Fight for Freedom from Corruption.
Johannesburg: Daily Maverick, 2022.
Hanson, Stephen. Post-Imperial Democracies: Ideology and Party Formation in Third Republic France,
Weimar Germany, and Post-Soviet Russia. Cambridge: Cambridge University Press, 2010.
Harris, Adam. Everyday Identity and Electoral Politics: Race, Ethnicity, and the Bloc Vote in South
Africa and Beyond. Oxford: Oxford University Press, 2022.
Horowitz, Donald. A Democratic South Africa? Berkeley: University of California Press, 1991.
Horowitz, Donald. Ethnic Groups in Conflict. Berkeley: University of California Press, 1986.
Huckfeldt, Robert, and John Sprague. “Networks In Context: The Social Flow of Political
Information.”American Political Science Review 81, no. 4 (1987): 1197–1216.
Johnson, Richard, and Lawrence Schlemmer, eds. Launching Democracy in South Africa: The First
Open Election, April 1994. New Haven: Yale University Press, 1996.
Joubert, Jan-Jan. Who Will Rule in 2019? Johannesburg: Jonathan Ball Publishers, 2018.
Key, Valdimer O. The Responsible Electorate. Cambridge: Mass: Belknap Press, 1967.
Kittilson, Miki, and Christopher Anderson. “In “Electoral Supply and Voter Turnout”.”In Citizens,
Context and Choice: How Context Shapes Citizens’Electoral Choices, edited by Russell Dalton,
and Christopher Anderson, 33–54. Oxford: Oxford University Press, 2010.
Lachat, Romain. “Explaining Electoral Volatility: Predispositions’Strength, Heuristics, and Political
Sophistication.”In Annual Scientific Meeting of the International Society of Political Psychology,
Lund, Sweden, pp. 15-18. July 2004.
Lacy, Dean, and Barry Burden. “The Vote-Stealing and Turnout Effects of Ross Perot in the 1992 US
Presidential Election.”American Journal of Political Science 43, no. 1 (1999): 233–255.
Lieberman, Evan, and Rorisang Lekalake. “South Africa’s Resilient Democracy.”Journal of Democracy
33, no. 2 (2022): 103–117.
Lindberg, Staffan. “Opposition Parties and Democratisation in sub-Saharan Africa.”Journal of
Contemporary African Studies 24, no. 1 (2006): 123–138. doi:10.1080/02589000500514943.
Lindberg, Staffan, and Minion Morrison. “Exploring Voter Alignments in Africa: Core and Swing
Voters in Ghana.”The Journal of Modern African Studies 43, no. 4 (2005): 565–586.
20 C. SCHULZ-HERZENBERG AND R. B. MATTES

Lipset, Seymour Martin, and Stein Rokkan. Party Systems and Voter Alignments: Cross National
Perspectives. New York: Free Press, 1967.
Magalhães, Pedro, Paolo Segatti, and Tianjian Shi. “Mobilisation, Informal Networks, and the Social
Contexts of Turnout.”In Voting in Old and New Democracies, edited by Richard Gunther, Paul
Beck, Pedro Magalhães, and Alejandro Moreno, 64–98. New York: Routledge, 2016.
Maloka, Eddy. “‘White’Political Parties and Democratic Consolidation in South Africa.”
Democratization 8, no. 1 (2001): 227–236.
Mashele, Prince, and Mzukisi Qobo. The Fall of the ANC Continues: What Next? Johannesburg:
Picador Africa, 2014.
Mattes, Robert. The Election Book: Judgment and Choice in South Africa’s 1994 Election. Cape Town:
Idasa, 1995.
Mattes, Robert. “The 2014 Election and South African Democracy.”In Election 2014: The Campaigns,
Results and Future Prospects, edited by Collette Schulz-Herzenberg, and Roger Southall, 169–187.
Auckland Park: Jacana Media, 2014.
Mattes, Robert. “South Africa’s Emerging Black Middle Class: A Harbinger of Political Change?”
Journal of International Development 27, no. 5 (2015): 665–692.
Mattes, Robert, Cherrel Africa, and Helen Taylor. “Judgment and Choice in the 1999 South African
Election.”Politikon: The South African Journal of Political Studies 26, no. 2 (1999): 235–247.
Mattes, Robert, and Jessica Piombo. “Opposition Parties and the Voters in South Africa’s General
Election of 1999.”Democratization 8, no. 3 (2001): 101–128.
Mattes, Robert, Matthias Krönke, and Sarah Lockwood. “In “South African Parties and the
Development of Democracy”.”In Political Parties and the Crisis of Democracy, edited by
Thomas Poguntke, and Wilhelm Hofmeister. Oxford: Oxford University Press, forthcoming.
Mattisonn, John. “Who Will Eat the ANC’s Lunch come 2024?”Daily Maverick, August, 17 2022.
https://www.dailymaverick.co.za/article/2022-08-17-who-will-eat-the-ancs-lunch-come-2024/.
Nkomo, Sibusiso. “Land Redistribution: South Africans Prioritize Land Taken in Forced Removals,
Support ‘Willing Seller’Approach, Afrobarometer Dispatch No. 254, 2018.”https://www.
afrobarometer.org/wp-content/uploads/2022/02/ab_r7_dispatchno254_land_redistribution_in_
south_africa.pdf.
Niemi, Richard, and Herbert Weisberg, eds. Classics in Voting Behavior. Washington, DC: CQ Press,
1993.
Norris, Pippa. Critical Citizens: Global Support for Democratic Government. Oxford: Oxford
University Press, 1999.
Norris, Pippa. Democratic Phoenix: Reinventing Political Activism. Cambridge: Cambridge University
Press, 2002.
Norris, Pippa. Democratic Deficit: Critical Citizens Revisited. Cambridge: Cambridge University Press,
2011.
Pauws, Jacques. The President’s Keepers: Those Keeping Zuma in Power and out of Prison. Cape Town:
Tafelberg, 2017.
Pempel, T. J. “Conclusion. One-Party Dominance and the Creation of Regimes.”In Uncommon
Democracies, edited by Pempel, TJ, 333–360. Ithaca, NY: Cornell University Press, 2019.
Platas, Melina, and Pia Raffler. “Closing the Gap: Information and Mass Support in a Dominant Party
Regime.”The Journal of Politics 83, no. 4 (2021): 1619–1634.
Pierce, Roy. “Modelling Electoral Second Choices: Thwarted Voters in the United States, France, and
Russia.”Electoral Studies 22, no. 2 (2003): 265–285.
Popkin, Samuel. The Reasoning Voter. Chicago: University of Chicago Press, 1991.
Rakner, Lise, and Nicolas Van de Walle. “Democratization by Elections? Opposition Weakness in
Africa.”Journal of Democracy 20, no. 3 (2009): 108–121.
Rattinger, Hans, and Elena Wiegand. “Volatility on the Rise? Attitudinal Stability, Attitudinal Change,
and Voter Volatility.”In Voters on the Move or on the Run, edited by Bernhard Wessels, Hans
Rattinger, Sigfried Rossteutscher, Rudger Schmitt-Beck, 287–308. Oxford: Oxford University
Press, 2014.
Rolfe, Meredith. Voter Turnout: A Social Theory of Political Participation. Cambridge: Cambridge
University Press, 2012.
Rose, Richard, and Ian McAllister. The Loyalties of Voters. London: Sage publications, 1990.
Rose, Richard, and Derek Urwin. “Social Cohesion, Political Parties and Strains in Regimes.”
Comparative Political Studies 2, no. 1 (1969): 7–67.
DEMOCRATIZATION 21

Richardson, Bradley. “European Party Loyalties Revisited.”In Elections and Voting Behaviour: New
Challenges, New Perspectives, edited by Pippa Norris, 21–46. Aldershot: Dartmouth Publishing,
1998.
Richardson, Bradley, and Paul Beck. “In “The Flow of Political Information: Personal Discussants, the
Media, and Partisanship”.”In Democracy, Intermediation, and Voting on Four Continents Gunther,
edited by Richard Gunther, Jose Ramon Montero, and Hans-Jurgen Puhle, 183–207. Oxford:
Oxford University Press, 2007.
Riker, William, and Peter Ordeshook. “A Theory of the Calculus of Voting.”American Political
Science Review 62, no. 1 (1968): 25–42.
Schulz-Herzenberg, C. “Trends in Party Support and Voter Behaviour, 1994–2009.”In Zunami!: The
2009 South African Elections, edited by Roger Southall, and John Daniel, 23–46. Johannesburg:
Jacana, 2009.
Schulz-Herzenberg, Collette. “Trends in Voter Participation: Registration, Turnout and the
Disengaging Electorate.”In Election 2019: Change and Stability in South Africa’s Democracy,
edited by Collette Schulz-Herzenberg, and Roger Southall, 44–65. Auckland Park: Jacana Media,
2019.
Schulz-Herzenberg, Collette. “The 2019 National Election Results.”In Election 2019: Change and
Stability in South Africa’s Democracy, edited by Collette Schulz-Herzenberg, and Roger Southall,
170–189. Auckland Park: Jacana Media, 2019.
Söderlund, Peter. “Retrospective Voting and Electoral Volatility: A Nordic Perspective.”Scandinavian
Political Studies 31, no. 2 (2008): 217–240.
Southall, Roger. Liberation Movements in Power: Party and State in Southern Africa. Pietermaritzburg:
University of KwaZulu Natal Press / James Currey, 2013.
Southall, Roger. “Opposition in South Africa: Issues and Problems.”Democratization 8, no. 1 (2001):
1–24.
Southall, Roger. The New Black Middle Class in South Africa. Martlesham, UK: Boydell & Brewer,
2016.
Thurner, Paul, and Angelika Eymann. “Policy-specific Alienation and Indifference in the Calculus of
Voting: A Simultaneous Model of Party Choice and Abstention.”Public Choice 102, no. 1 (2000):
49–75.
Teixeira, Ruy. “Turnout in the 1992 Election.”Brookings Review 11, no. 2 (1993): 47–48.
Tillman, Erik. “Economic Judgments, Party Choice, and Voter Abstention in Cross-National
Perspective.”Comparative Political Studies 41, no. 9 (2008): 1290–1309.
Torcal, Mariano, and José Montero. Political Disaffection in Contemporary Democracies. London:
Routledge, 2006.
Trilling, Richard. Party Image and Electoral Behavior. New York: John Wiley, 1976.
Verba, Sidney, Kay Lehman Schlozman, and Henry Brady. Voice and Equality: Civic Voluntarism in
American Politics. Cambridge: Harvard University Press, 1995.
Weschle, Simon. “Two Types of Economic Voting: How Economic Conditions Jointly Affect Vote
Choice and Turnout.”Electoral Studies 34 (2014): 39–53.
Zelle, Carsten. “Social Dealignment Versus Political Frustration: Contrasting Explanations of the
Floating Vote in Germany.”European Journal of Political Research 27, no. 3 (1995): 319–345.
22 C. SCHULZ-HERZENBERG AND R. B. MATTES