Protest Blues Public opinion on the policing of protest in South Africa

The policing response to rising protest action in the country has received increased attention in the last decade. This is particularly owing to concerns over confrontations during which protesters have been arrested, injured and in some instances killed by the police. Despite the criticism voiced by various stakeholders about the manner in which the police manage crowd gatherings, relatively little is known about the views of South African adults on the policing of protest action and the factors that shape such attitudes. To provide some insight, this article draws on data from a specialised module on protest-related attitudes and behaviour that was fielded as part of the 2016 round of the Human Sciences Research Council’s South African Social Attitudes Survey (SASAS) series. This nationally representative survey included specific questions probing the public’s overall evaluation of the performance of the police in dealing with protests, and the justifiability of the use of force in policing protest action. The article will present a national picture of people’s views on the policing of protest, based on these measures, and then determine the extent to which there are distinct underlying socio-demographic cleavages in these data. A combination of bivariate and multivariate analysis is undertaken in order to understand how perceptions of effectiveness, acceptability and reported participation in protest (especially disruptive and violent actions) shape people’s views regarding policing of protest. The article concludes with a discussion that reflects on the implications of the research for the policing of protest action in future, given the appreciable rise in the incidence of protest since the mid-2000s and the mounting tensions between state institutions and communities over the political, moral and constitutional arguments for and against such actions.

The policing response to rising protest action in the country has received increased attention in the last decade.This is particularly owing to concerns over confrontations during which protesters have been arrested, injured and in some instances killed by the police.Despite the criticism voiced by various stakeholders about the manner in which the police manage crowd gatherings, relatively little is known about the views of South African adults on the policing of protest action and the factors that shape such attitudes.To provide some insight, this article draws on data from a specialised module on protest-related attitudes and behaviour that was fielded as part of the 2016 round of the Human Sciences Research Council's South African Social Attitudes Survey (SASAS) series.This nationally representative survey included specific questions probing the public's overall evaluation of the performance of the police in dealing with protests, and the justifiability of the use of force in policing protest action.The article will present a national picture of people's views on the policing of protest, based on these measures, and then determine the extent to which there are distinct underlying socio-demographic cleavages in these data.A combination of bivariate and multivariate analysis is undertaken in order to understand how perceptions of effectiveness, acceptability and reported participation in protest (especially disruptive and violent actions) shape people's views regarding policing of protest.The article concludes with a discussion that reflects on the implications of the research for the policing of protest action in future, given the appreciable rise in the incidence of protest since the mid-2000s and the mounting tensions between state institutions and communities over the political, moral and constitutional arguments for and against such actions.
The media and commentators have often referred to South Africa as 'the protest capital of the world'. 1 Indeed, the country has experienced a considerable increase in protest activity in the last 10 years, some of which has been quite violent. 2 The manner in which these protests have been handled from a policing perspective has placed law enforcement in South Africa under appreciable public scrutiny. 3rowd control of these protests by police and in particular the Public Order Police (POP) units has been called into question by academics as well as civil society. 4The death of Andries Tatane, who subsequently became a symbol of inadequate policing during protests, has regularly been cited as an example of police failure in this area.The August 2012 Marikana massacre also highlights the failure of policing during protests, and the lack of response from government. 5Despite the criticism voiced by civil society and other stakeholders about the manner in which the police control crowd gatherings, relatively little is known about South Africans' views on the policing of protest action and the factors that shape such attitudes.To provide some insight, this article draws on recent nationally representative public opinion data to examine attitudes about the policing of protest action.
The violent treatment of protesters at the hands of police officers is not a recent aberration but dates back to the apartheid era. 6The General Law Amendment Act 37 of 1963 and the Criminal Procedure Amendment Act 96 (180-Day Detention Law) of 1965 gave the South African Police (SAP) the power to arrest anyone suspected of acting against the state and hold them without charge for 90 days. 7These laws were used to suppress protests and arrest protesters.SAP officers often lacked proper crowd control training and were deployed to suppress public protests armed with shotguns, bullwhips and batons. 8The result was brutal and violent.Perhaps the most tragic example is the 1960 Sharpeville massacre, when police fired live rounds into a crowd of between 5 000 and 7 000 protesters, killing 69 and injuring hundreds.Similar incidents occurred in 1976 during the Soweto uprising as well as in Uitenhage in 1985, when 20 people were killed. 9During the apartheid period, the policing of protest action 'ensured that sustained brutality' was a dominant feature of a 'black South African experience'. 10One notable outcome of this history of authoritarian policing is a deep-seated lack of public confidence in the legitimacy of the police. 11th the transition to democracy in the early 1990s, the new government sought to restore public confidence in the authorities' ability to manage protests.Legislation, including the South African Police Service Act of 1995 and the Regulation of Gatherings Act of 1993, was introduced to reform how the police handled crowd control.The fragmented policing service that apartheid spatial planning had produced was swept away and a single, centralised South African Police Service (SAPS) was created.A new organisational transformation agenda aimed to alter 'police cultures, structure and symbols', and brought new emphasis on a community policing model. 12Unlike the former SAP, the new SAPS would no longer suppress popular will, but would work with communities to maintain order and law. 13POP units were created in 1996 to ensure prudent and judicious crowd control. 14In keeping with these commitments, the country became a member of the Peace and Security Council, which is an African Union organ concerned with stability and the resolution of conflict in Africa. 15 2002, POP units were restructured into Area Crime Combating Units (ACCUs), reflecting a strategic shift in focus from crowd management policing to crime reduction. 16POP units were further restructured in 2006 with the number of units cut from 43 in 2002 to 23; thereby significantly reducing the number of dedicated POP members. 17The restructuring of public order policing functions coincided with an increase in the number of crowd management incidents the ACCUs/POP units had to respond to, and the restructuring thus had a negative impact on the police's ability to deal with protest. 18This has placed a considerable burden on existing police resources and there has been an attempt to strengthen POP units by increasing the number of dedicated, trained POP officers and the number of POP units.In 2014 the SAPS reported that POP had 28 units and 4 175 officers, and requested R3.3 billion for further expansion. 19The government aims to employ 11 800 POP officers by 2020. 20e capacity of the SAPS to perform its crowd management duties is undermined by negative public sentiment towards the police.A small body of scholarship has attempted to understand antipathy towards the police in spite of the considerable policy change and experimentation post-1994.International scholarship on legitimacy and procedural justice has tended to demonstrate that public judgments about police fairness and effectiveness have a considerable influence on an individual's overall evaluations of police legitimacy. 21A number of recent studies have raised concern about the fairness with which the police treat ordinary South Africans. 22xisting research suggests that trust in the police is low, which undermines the legitimacy of this important institution. 23spite the widespread policing reforms since 1994, many challenges exist in relation to police legitimacy in present-day South Africa.The police's role during apartheid likely weighs heavily in evaluations of present-day policing for many people, and the resurgence of paramilitarism in policing practices, such as the deployment of Tactical Response Team (TRT) units at Marikana, likely produces ambivalent public responses.The use of excessive and lethal force, mounting issues of police corruption, lingering concerns over fair and equal treatment, as well as the perception of police incompetence in the face of high crime rates, further complicate the picture.This has resulted in a remarkable turn towards various forms of non-state policing, 24 including vigilantism, which in turn is likely to inform perceptions of police legitimacy.These factors have resulted in increasing calls for a form of minimalist policing in which police activity focuses on more effectively performing core functions such as criminal investigation and emergency response, with non-state actors taking strong roles in everyday policing and crime prevention. 25om an international perspective, it was not until the late 1980s and 1990s that the policing of protest became a subject of substantive interest within the social sciences, with early survey-based and qualitative research focusing on the repression of protest and on police actions in maintaining public order. 26In 1998 the concept of 'protest policing' was formally introduced through the influential volume edited by Donatella della Porta and Herbert Reiter titled Policing protest: the control of demonstrations in Western democracies.Defined simply as 'the police handling of protest events', protest policing within democratic societies was portrayed as involving a fine balance between protecting public law and order and defending individual freedoms and the citizen right to political participation and demonstration. 27The latter rights are regarded as quintessential elements of liberal democracy; consequently, the style of policing adopted in controlling protest, which has the potential to either polarise or win the favour of majoritarian public opinion, has come to receive much academic and policy scrutiny.The public order literature has charted how approaches to protest policing have evolved over the decades, from what was characteristically referred to as an 'escalated force' model, which predominated in the 1960s, to a 'negotiated management' approach in the 1980s and 1990s.The former involved a general disregard for the constitutional right to demonstrate and a failure to issue protest permits, tolerance only of 'comfortable' (most peaceful) forms of protest, nominal police-protester communication, a predisposition for forceful arrest of perceived agitators, and the use of force as a standard protest control method. 28y contrast, negotiated management entails respect for civic rights, tolerance of a certain level of disruptive behaviour, a strong emphasis on communication, reliance on arrests as a last resort, and adherence to minimum necessary force. 29Although there is recognition that the policing of protest has become less violent in Western democracies in recent decades with the rise of a softer, more tolerant and flexible approach, there are rising concerns that the pendulum may have begun to swing again towards repressive tendencies in the face of transnational, anti-globalisation protests and as a mounting response to terrorist threats. 30This, in turn, has led to renewed attention to the style of and explanations for protest policing.
In what remains the most widely applied theoretical model explaining styles of protest policing, Della Porter and Reiter argue that the prevailing approach to police handling of protest is informed by a two-tiered set of factors. 31At the first level, these determinants include: (1) the organisational structure and culture of policing, including the extent of police discretionary powers and the protestrelated stereotypes they hold; (2) the political context and culture of a country, including dominant norms about the role of the state and citizen rights; (3) public opinion and interests expressed by various collective actors, including government, social movements, political parties, trade unions, interest groups, civil society organisations and the media; and (4) the actual experiences of interaction between police and protesters. 32The extent and nature of the impact of these factors on protest policing approaches is ultimately mediated by their level of influence, at the second tier, on 'police knowledge'.This refers to the police's perceptions of external reality, both at the individual officer level and collectively.What is of particular theoretical relevance for this article is that public opinion is acknowledged as having a potential influence on trends in protest policing practice.However, this influence is conditional on such public preferences reaching and changing the way the police view the context into which they are sent to maintain public order.People's understanding of and response to protest dynamics are also likely to be informed by the media, which publishes and popularises the preferences of influential opinion leaders such as government, political parties and lobby groups.This, taken together with broader contextual events, may lead to a demand for either tougher or softer interventions in policing protest.
The next section of the article provides an outline of the survey data and measures used in our study.This leads into a presentation of our findings, which is structured in three parts.Firstly, we examine the extent to which the public on average expresses confidence in the way protest is being policed, and determine the extent to which distinct sociodemographic differences in perspective exist.Secondly, we cast attention on the use of force by police in managing protests in the country, focusing in particular on the perceived justifiability of such behaviour.Finally, we conduct multivariate regression analysis to discern which factors influence individual evaluations of the policing of protest.This analysis aims to provide an understanding of how various elements shape opinions regarding the policing of protest: the role of basic socio-demographic factors, the perceived effectiveness and acceptability of protest, reported participation in protest, as well as views on use of force and general trust in the police.The article concludes with a discussion that reflects on the implications of the survey results for the policing of protest action in future.

Data
This study employs quantitative data from the 2016 round of the SASAS, a repeat cross-sectional survey series that has been conducted annually since 2003 by the Human Sciences Research Council (HSRC).Each SASAS round has been designed to yield a nationally representative sample of adults aged 16 and older living in private residences.Statistics South Africa's 2011 Population Census Small Area Layers (SALs) were used as primary sampling units (PSUs).For each round of SASAS, 500 PSUs are drawn, with probability proportional to size, from a sampling frame containing all of the 2011 SALs. 33 each of these drawn PSUs, 21 dwelling units were selected and systematically grouped into three sub-samples of seven, each corresponding to the three SASAS questionnaire versions that are fielded.The relevant protest action questions were included in only one of the three instruments, and thus administered to seven visiting points in each PSU. 34The sample size of the study consisted of 3 079 interviews, which is equivalent to an 88% response rate.
The English base version of the research instruments was translated into the country's major official languages and the surveys were administered in the preferred language of the respondent.This was to ensure that all respondents in different provinces understood the questionnaire and that it was culturally equivalent and consistent across all languages.Pilot testing was conducted in an attempt to ensure the validity of the research instrument.Interviews were conducted by means of face-toface interviewing, using print questionnaires. 35

Measures on the policing of protest
The 2016 SASAS round included a specialised module on protest-related attitudes and behaviour.This was designed in conjunction with the University of Johannesburg's Centre for Social Change.The module included two items that address the policing of protest action in the country.The first measure addresses the perceived effectiveness with which the police are dealing with protest action.Specifically, respondents were asked: 'In your opinion, how well are the police dealing with protests in South Africa?' Responses were captured using a fourpoint scale, with the coded options labelled as 'very well', 'fairly well', 'not very well', and 'not at all well'.The second survey measure deals with the perceived legitimacy of the use of force by the police in responding to protests.The question was introduced with an explanation of use of force, followed by an example aimed to elicit a clear response by the public on whether they regard such police action as justifiable or not.The specific phrasing of the question is as follows: 'There are different views on the use of force by police during protest action.By force we mean the use of rubber bullets, stun grenades, tear gas and water cannons by the police.Please say whether the use of force by the police against protesters who throw stones at them is justified in all cases, is justified in some cases, or is never justified.'

Police performance in handling protest action
From Figure 1 it is apparent that barely a third (37%) of South Africans consider the police to be performing 'very' or 'fairly' well in handling protests in the country. 36By contrast, a majority (60%) believe that the police are faring poorly in their response to protest, with 35% stating they are not performing very well and a further 25% saying they are not performing well at all.The remaining 3% were uncertain as to how to evaluate this form of policing.
To better understand whether the South African public holds relatively uniform or discrepant views in relation to the policing of protest action, we examined the nature and extent of variance in perspective, based on various socio-demographic attributes.The findings show that there were no statistically significant differences in evaluation based on age, gender, race, marital status, educational attainment, employment status or standard of living level.Employment status has a modest effect, with unemployed adults providing more critical views than pensioners and others who were labour inactive. 37This suggests that demographic variables do not exert much influence over how the public views the way in which protest action is being policed in the country, and points to a fairly broad level of consistency in attitude.
There is, however, notable spatial variation underlying the national average.In terms of type of geographic location, we find that those residing in informal urban settlements tend to offer harsher views on police performance in handling protests than those based in formal urban areas, rural traditional authority areas and on rural farms.Provincially, those in Limpopo and the Northern Cape provide less critical assessments of the effectiveness of the policing of protest, although even in these instances the public remains quite ambivalent, with virtually equal shares adopting favourable and unfavourable positions.At the other extreme, the most negative evaluation comes from residents in the North West province, where approximately three-quarters (74%) indicated that the police were faring poorly in dealing with protest action.Unfortunately, given the absence of trend data on the measure, we cannot determine the extent to which this has been informed by events in Marikana five years ago, or as a result of other deaths that have occurred during protest in the North West, such as the water protests in Mothutlung that resulted in the death of four people.It is, however, plausible that these tragic events may have had an indelible effect on attitudes towards public order policing and the police more generally in the province.Bivariate testing reveals that those living in the North West, Gauteng and KwaZulu-Natal are more negative in outlook than those in Limpopo and the Northern Cape. 38

The justifiability of using force in policing protest
The use of force in the context of policing protest in the country has received increased attention over the last decade.This has been prompted in particular by specific high-profile events, including the killing of Andries Tatane and the Marikana massacre, as well as the manner in which the #FeesMustFall protests were handled.This raises the question as to whether the public favours or rejects the kinds of displays of force that have become an increasingly common response by public order police in cases of violent protest.In Figure 2 we present the national distribution, based on the measure regarding public views on the use of force in policing protest.Slightly more than a tenth (13%) regard a forceful policing response as unequivocally justifiable, with close to half of South Africans seeing such action as acceptable in certain instances.Only around a third (35%) expressly rejected the use of force in responding to protests, with a nominal share remaining uncertain in their views on this matter.This is quite a disconcerting finding, as it seems to suggest that the public has an appetite for a strong policing response (at least in certain contextual circumstances) in dealing with more violent forms of protest.It does nonetheless resonate with the public preferences about how criminality ought to be dealt with in general, which tends towards a demand for punitive actions. 39It is again important to understand how widely this general predisposition is shared among the adult public before we return to the issue of how this and other factors inform confidence in the policing of protest more broadly.
At the subgroup level, we find no significant differences in views on the use of force based on age, gender, educational attainment, employment status, marital status, or standard of living level.There are, however, notable population group and geographic differences that are apparent, as presented in Table 2.The findings show that white adults and, to a lesser extent, coloured adults are more inclined to favour the use of force than black African and Indian adults.The main basis of this distinction is due to a greater tendency among white and coloured adults to respond that the use of force is 'sometimes justifiable', while the opposite pattern is true in relation to the 'never justifiable' category.There is no significant variation in the shares responding 'always justifiable', though Indian adults were more likely to voice uncertainty (15% compared to 5-8% for the rest).Despite these differences, the predominant response in all cases is that police use of force is viewed as warranted in certain circumstances, even if the degree of support for this option varies.
The observed differences with respect to type of geographic location are only barely statistically significant.Those residing in informal urban settlements were less likely than formal urban dwellers to respond that the use of force is 'sometimes justifiable', while conversely, those in informal settlements were more likely to respond that it is 'never justifiable' than were those in formal urban areas.Those living on rural farms displayed greater uncertainty than those in informal settlements and rural traditional authority areas.

What factors influence evaluations of the policing of protest?
Apart from the descriptive analysis outlined above, we also conducted regression analysis  to provide a clearer sense of the significant predictors of public evaluations of the effectiveness of the policing of protest.In so doing, we aimed to ascertain whether the statistically significant findings identified earlier remained when we combined the variables into the multivariate models.Given that the dependent variable is an ordered categorical measure, we used ordered logistic regression techniques.For ease of interpretation, we reversed the scaling of the variable, so that a value of '0' was assigned to those reporting that the police are faring 'not at all well' in dealing with protest, a score of '1' to those answering 'not very well', '2' to those 'fairly well', and lastly a value of '3' to those responding that the police are doing 'very well'.A series of models was then generated, as presented in Table 3.
We begin with a base model containing only the socio-demographic attributes of respondents (Model 0).This is followed by five models that test the effect of including specific attitudinal or behavioural measures to the base model (Models I-V).Finally, we run a fully specified model that includes the socio-demographic and all the additional indictors (Model VI).In each of these ordered logistic models, we present the proportional Odds Ratios (OR). 40r base model (Model 0) confirms the earlier result that virtually none of the sociodemographic attributes is a statistically significant predictor of the way South Africans assess the policing of protest.Specifically, the model indicates that there is no evidence of an age, gender, race, marital status, employment status, or educational effect informing such evaluations.Political party identification was also included in the model.Using support for the ruling party as the reference group, supporting an opposition party was not found to be a significant determinant in this model.This finding holds true even after other variables are added in subsequent models in the table.Geography matters to some degree, with residents of informal settlements more inclined than those in formal urban areas to report lower policing effectiveness scores.This may partly be owing to a greater likelihood that respondents have participated in protest action, and by extension that they have more exposure on average to public order policing.We were also interested in determining whether respondents' views of the general image and perceived effectiveness of disruptive and violent protest action had any bearing on their evaluations of the policing of protest.These measures are more fully examined in their own right in the article by Bohler-Muller and colleagues in this special issue.The survey included separate measures on whether respondents tend to regard peaceful, disruptive and violent protest action in a positive or negative light, with responses captured on a 7-point scale ranging between 'extremely negative' and 'extremely positive'.For analytical purposes, we created an index focusing on the image of disruptive and violent action, which was constructed by averaging together the scores for the two indicators, which retains the original 1-7 negative to positive scaling.Similarly, the survey fielded questions on the effectiveness of the three types of protest, using a 7-point scale ranging from 'extremely unsuccessful' to 'extremely successful'.We constructed an index of the effectiveness of disruptive and violent actions by again averaging the two constituent items, with higher scores continuing to represent greater perceived effectiveness of these actions.The testing of these attitudinal measures as predictors of evaluations of public order policing is presented in models II and III respectively.Both the image and perceived effectiveness of disruptive and violent protest action are not significant factors in explaining public assessments of performance in policing protest, as was also observed with participation in protest action.
In Model IV, we concentrate on the relationship between views of the policing of protest and the perceived acceptability of the use of force by police in responding to protests.In this instance, we find that the justifiability of the use of force in policing protest emerges as a significant predictor.
Those who view the use of force as never or only sometimes justifiable tend to provide the SAPS with lower performance scores in terms of their handling of protests, compared to those who view the use of force as always justifiable.Even those respondents who were unsure about their position on the use of force tended to offer significantly lower evaluative scores relative to those viewing such force as always permissible when responding to protest.This remains the strongest single effect based on the various indicators that we tested in our analysis.In approaching the study, we were keen to examine the extent to which one's general confidence in the police as an authority has a bearing on appraisals of specific areas of performance, such as public order policing.
Our hypothesis was that those who exhibit distrust of the police would on average tend to voice more critical views on performance, and vice versa.Indeed, this proves to be the case, as demonstrated in Model V. Our measure of overall police confidence was initially designed as part of a European Social Survey module on confidence in the criminal justice system, which has been fielded in the SASAS series in recent years.The question is phrased as follows: 'Taking into account all the things the police are expected to do, would you say they are doing a good job or a bad job?', with responses captured using a five-point scale ranging from a 'very good job' to a 'very bad job'.For modelling purposes, we reversed the scale, so that higher values indicate greater confidence levels.The appeal of this item is that it is phrased in a similar way to our policing of protest item.We also tested the effect of an alternate police confidence measure that explicitly asks about levels of trust in police, using a standard fivepoint trust scale.Based on this specification, the finding remains the same.
Lastly, Model VI runs the analysis with all the different indicators included.The findings from the preceding models remain largely unchanged.The socio-demographic measures continue to be insignificant factors, with only minor geographic effects present.Limpopo residents continue to express higher than average performance ratings, although a similar pattern in the Northern Cape loses its salience once other attitudinal and behavioural variables are controlled for.South Africans living in informal settlements continue to exhibit a more disapproving stance than those in other geographic locales on how protests are being policed.The perceived justifiability of the use of force, in addressing protest, in addition to overall levels of confidence in the police retain their positive association with protest policing evaluations.Past participation in violent and disruptive protest actions, together with the image and perceived effectiveness of such protest, continues to register no discernible influence in appraising SAPS performance.

Discussion
Our examination of public attitudes towards protest policing has shown that, on the whole, performance evaluations tend to be fairly negative.This perspective is commonly shared across various demographic and class traits, though appreciable geographic variation is nonetheless apparent.These results confound expectations of lower levels of confidence in police crowd management activities among more vulnerable and marginalised segments of society, which indicates that the so-called 'rebellion of the poor' in protest behaviour is not resolutely manifest in the mind of the public. 41his is an interesting finding that will require further testing, using data on a broader set of concepts and constructs.
In considering other factors beyond sociodemographic markers that might help explain the way citizens appraise protest policing, the lack of statistical significance in relation to measures such as recent participation in protest action as well as support for and the perceived effectiveness of disruptive and violent protest actions, is particularly striking.It signifies that one's experience of engaging in protest action -and by extension first-hand exposure to the manner in which the police approach crowd management -does not exert a sizable influence on one's view of police performance in undertaking such duties.Furthermore, one's general predisposition towards disruptive and violent actions also does not play a role in structuring expressed levels of confidence in the policing of protest action.So, an aversion to more disruptive and violent forms of protest does not automatically translate into a more sanguine view of public order policing.
What clearly seems to matter, though, is the public's position in relation to the acceptability of the use of force in maintaining public order.
The more one deems it justifiable for officers to use violence in particular situations, the more inclined one is to provide a positive evaluation of the policing of protest.For approximately a third of South Africans, the use of force by the police in the context of protest is deemed to be wholly unacceptable.This is associated with acutely diminished confidence in the police's handling of protest.It may be that for this segment of society, the unfairness and brutality that have characterised the policing of protests have violated their notion of 'good' policing and the values of fair treatment, appropriate conduct and respect that maintain a sense of legitimacy, trust and confidence. 42By contrast, for the smaller minority (one in eight, or 13%) that considers the use of non-lethal physical force as always justifiable, levels of confidence in public order policing is more than four times higher.This suggests, somewhat controversially, that the use of force to control protesters may serve to promote or reinforce police legitimacy for some South Africans.This would imply that, for this group, a less aggressive or violent approach to public order policing might bring into question the legitimacy of, and confidence in, the police.Although our study does not provide a comprehensive account of the attitudes towards police use of force in protest situations, international evidence points to aggressive personality traits, a tendency towards rightwing authoritarianism, and a stronger social dominance orientation as possible factors associated with a more accepting stance on the excessive use of force. 43This may be due to a desire to control social threats, promote security and help maintain current power hierarchies. 44e dominant public response to the use of force question remains one that regards the violent policing of protest as justifiable in certain circumstances.Accounting for slightly less than half of the adult population, this position is associated with a more ambiguous position in respect of confidence in protest policing, with virtually equivalent shares expressing favourable and unfavourable views.The circumstances under which such tactics might be tolerable cannot be ascertained from our data, but the calculus is likely to involve a range of factors, from the behavioural repertoires of the protesters to whether the police response has firstly exhausted negotiation and all other options involving a minimal amount of force.The ambiguity in public order policing confidence ratings might also partially reflect a sense of unease about whether the police response in managing protests falls within the ambit of reasonable or justifiable use of force, or not.The former group is likely to view force as a constituent element of effective policing, but regard the application of force in crowd management incidents as highly conditional and contextual.In relation to the preceding points, it is worth noting that the definition and accepted normative limits of 'police violence' may tend to vary over time, context and ideological outlook.

Conclusion
The processes of transformation in public order policing in South Africa since the early 1990s have been complex and nonlinear.An initial political commitment to professional, democratic public order policing was subsequently followed by a period of organisational degradation and leadership problems.Together with the prioritisation afforded to the fight against crime, this led to the relative neglect of public order policing for a number of years.However, in response to the rising incidence of public protest in the country, the tide has turned and public order policing has received renewed attention.Concerns have nonetheless been expressed about whether this recent development has been accompanied by an ethos emphasising a 'hard-edged' approach involving more forceful policing practices, rather than the application of minimum force. 45The subsequent rise in reported cases involving excessive use of force and police fatalities during acts of demonstration, together with the events in Marikana, have raised fundamental questions about the manner in which protest is being policed in our constitutional democracy.
From a public opinion perspective, it has also led to questions about the implications of such developments on the perceived legitimacy of the police.
As a response to the policing failures in dealing with public protest, including the escalation in the number of protesters killed by police over the 2010-2014 period, 46 there have since 2014 been signs of a distinct retreat at the senior political and police level from the strong-arm public order policing approach that typified the early 2010s. 47This has involved something of a cyclical return to the priorities of the mid-tolate 1990s, a period characterised by deliberate attempts to move public order policing away from the apartheid state's repression of demonstration through brutally forceful policing.Developments include the return in name of the Public Order Policing (POP) unit with a primary emphasis on crowd management, a commitment to reinvesting in public order capacity in terms of both training and numbers of police members, and the introduction of a National Instruction on Crowd Management during public gatherings and demonstrations.The latter restates the importance of a welltrained, resourced and command-driven unit that displays utmost restraint, and adheres to strict guidelines governing the use of force as a tactic of last resort and in compliance with legislative and constitutional imperatives. 48e apparent political will that currently exists for a new organisational model of public order policing represents an opportune moment to critically engage with and shape the future approach to this specialised form of policing. 49he choices that are made in this regard will indelibly influence the next generation of policecitizen relations.Based on our survey results, we contend that a continued reliance by the police on disproportionate and excessive force, and a tendency to resort quickly to the use of rubber bullets and teargas as controlling tactics in dealing with protest, may provoke a further withdrawal of support for the use of force.This, in turn, would further diminish overall confidence in the ability to police protest actions.This is of concern, since public trust and confidence are generally recognised as a key component of ensuring effective, democratic policing. 50rganisational transformation is a necessary but insufficient part of promoting positive and enduring change.It also requires an appreciation of the socio-economic and political context in which protest action and public order policing are occurring. 51

Study limitations
This article has contributed to our knowledge of South African public opinion on police performance in handling protest action.
However, the analysis is not without limitations.
There is currently no available trend data on attitudes to the issues under discussion.
As a result, we do not know how stable or variant such attitudes are, and how sensitive these attitudes are to contextual events.In addition, we only have single-item measures of satisfaction with protest policing performance and the acceptability of use of force by the police.The use of single-item measures may fail to capture important nuances in public opinion on protest action.Consequently, it is not possible to say with confidence what motivates the observed link between attitudes towards the use of force during protests and evaluations of police performance in controlling protest.Other important questions also remain unresolved.
For example, what types of force used by the police to control protests would the public be comfortable with?Moreover, public attitudes towards the use of force by may vary, depending on the type of protesters under consideration, for instance students versus workers.Our use of force measure focused only on retaliatory responses to violent protest (i.e.protesters throwing stones at police) and we might arrive at a different or more nuanced set of results if a range of examples of excessive and reasonable use of force are provided. 53The role of the media in informing the understanding and preferences that the public has in relation to protest and the policing of such events has also not been examined in the article, owing again to the absence of relevant questions in the survey instrument.To address these limitations, future public opinion research needs to utilise a more comprehensive set of questions on police performance in handling protest action, as well as on other relevant contextual factors.34 Interviewers called at each visiting point selected and listed all those eligible for inclusion in the sample in terms of age and residential status criteria.The interviewer then selected one respondent using a random selection procedure based on a Kish grid.
35 The HSRC's Research Ethics Committee granted ethical approval for the instrumentation and research protocols for each round.
36 The lower and upper 95% confidence intervals for this estimate are 33.4% and 40.3% respectively.
37 This effect is present based on One-Way Analysis of Variance (ANOVA) post hoc testing, but it falls away when one combines the percentages opting for the two positive categories and compares this with the combined two negative categories.
38 Residents in the Western Cape also provide on average less favourable ratings of the handling of protest compared to those in Limpopo, but this is not true of comparisons between the Western and Northern Cape.
39 For instance, the SASAS series has found a strong demand for the reinstatement of the death penalty in cases of murder and broad-based tolerance of vigilantism, coupled with a positive response to the paramilitaristic turn in policing that characterised former police commissioner Bheki Cele's term of office.
40 For continuous variables in the models (e.g.age), an OR of greater than 1 signifies that for every unit increase in these predictors is associated with greater odds of rating the policing of protest as effective in character.Conversely, an OR of less than 1 denotes that for every one-unit increase, the likelihood of the policing of protest being regarded as effective diminishes.Odds ratios that are equal to (or approximate) 1 imply that the variable has no discernible effect on the effectiveness evaluations.For categorical or dichotomous variables in the models (e.g.gender, race), the odds are relative to the specified reference category.So, in the case of gender, we are comparing the proportional odds ratio of females to males on the effectiveness of the policing of protest, with all other variables in the model held constant.
41 There are glimpses that such an attitudinal patterning exists, such as the finding that significantly lower confidence ratings are evident among residents in informal urban settlements, although support for this hypothesis is overall fairly circumscribed.This is confirmed by the multivariate analysis.

Table 1 :
Spatial differences in the evaluations of how the police are handling protest action, 2016 (percentages and mean scores) Source: HSRC South African Social Attitudes Survey (SASAS) 2016.Note: The mean scores are based on a reversed scale, where 0='not at all well' and 3='very well'.'Do not know' responses are not presented in the table.The unweighted base number of observations are included in the final column based on the distributions with and without 'don't know' responses included.The percentages in the table are based on the former, and the mean scores the latter.

Table 2 :
Significant differences in views on the use of force in policing protest, 2016 (percentages) Source: HSRC South African Social Attitudes Survey (SASAS) 2016.

Table 3 :
Ordered logistic regression of the effectiveness of the policing of protest, 2016 Note: OR = odds ratio.The dependent variable is a reversed scaled version of the performance of the policing of protest measures, with 0='not at all well', 1='not very well', 2='fairly well' and 3='very well'.'Don't know' responses were omitted.Statistical significance is represented as follows: *p<0.05,**p<0.01,***p<0.001.
-massive-rebellion-of-the-poor (accessed24August 2017); Silvia Bianco, South Africa: the 'protest capital of the world', The South African, 20 June 2013, https://www.thesouthafrican.com/south-africa-the-protest-capital-of-theworld/(accessed 24 August 2017).33 The sampling frame is annually updated to coincide with StatsSA's mid-year population estimates in respect of the following variables: province, gender, population group and age group.The sample excludes special institutions (such as hospitals, military camps, old age homes, schools and university hostels), recreational areas, industrial areas and vacant areas.It focuses on dwelling units or visiting points as secondary sampling units (SSUs), which are separate (non-vacant) residential stands, addresses, structures, flats, homesteads and other similar structures.Three explicit stratification variables were used in selecting SALs, namely province, geographic type and majority population group.