This report has been enabled
through the generous support of
Telkom
Copyright © 2024
Inclusive Society Institute
PO Box 12609
Mill Street
Cape Town, 8010
South Africa
235-515 NPO
All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means without the permission in writing from the Inclusive Society Institute
DISCLAIMER
Views expressed in this report do not necessarily represent the views of the Inclusive Society Institute or its Board or Council members.
November 2024
Author: Georgi Dragolov and Klaus Boehnke
Constructor University, Bremen, Germany
Editor: Daryl Swanepoel
Table of Contents
Table of Contents
List of Tables
List of Figures
Executive Summary
1. Introduction
2. Measuring social cohesion
2.1 Data
2.2 Analytical approach
3. Level and trend of social cohesion
3.1 Social cohesion in South Africa
3.2 Social cohesion in the nine provinces
4. Structural influences on social cohesion
4.1 Data and method
4.2 Results
5. Individual experiences of social cohesion
5.1 Data and method
5.2 Four classes of experience
5.3 Socio-demographics of the four classes
6. Social cohesion and subjective well-being
6.1 Provinces
6.2 Individuals
7. Discussion and conclusion
References
Appendices
Appendix A: Indicators of cohesion across time
Appendix B: Dimensions of cohesion in the provinces over time
Appendix C: Correlations of social cohesion on the province level
Appendix D: Latent class analyses
Cover photo: istock.com - Stock photo ID:1440750455
List of Tables
Table 2.1 Sample sizes of Khayabus – Waves 1
Table 2.2 Factor loadings of items for dimensions within Domain 1, “Social relations”
Table 2.3 Factor loadings of items for dimensions within Domain 2 “Connectedness”
Table 2.4 Factor loadings of items for dimensions within Domain 3 “Focus on the common good”
Table 3.1 Social cohesion and its dimensions in South Africa across time
Table 3.2 The overall index of social cohesion in South African provinces across time
Table 4.1 Structural characteristics and social cohesion in South African provinces
Table 5.1 Social cohesion and its dimensions in the four classes
Table 5.2 Socio-demographic and economic characteristics of the four classes of respondents
Table 6.1 Social cohesion and subjective well-being in the provinces
Table 6.2 Subjective well-being in the four classes of respondents
Table 0.1 Indicators of Dimension 1.1 “Social networks” across time
Table 0.2 Indicators of Dimension 1.2 “Trust in people” across time
Table 0.3 Indicators of Dimension 1.3 “Acceptance of diversity” across time
Table 0.4 Indicators of Dimension 2.1 “Identification” across time
Table 0.5 Indicators of Dimension 2.2 “Trust in institutions” across time
Table 0.6 Indicators of Dimension 2.3 “Perception of fairness” across time
Table 0.7 Indicators of Dimension 3.2 “Solidarity and helpfulness” across time
Table 0.8 Indicators of Dimensions 3.2 “Respect for social rules” across time
Table 0.9 Indicators of Dimension 3.3 “Civic participation” across time
Table 0.10 Dimension 1.1 “Social networks” in the provinces across time
Table 0.11 Dimension 1.2 “Trust in people” in the provinces across time
Table 0.12 Dimension 1.3 “Acceptance of diversity” in the provinces across time
Table 0.13 Dimension 2.1 “Identification” in the provinces across time
Table 0.14 Dimension 2.2 “Trust in institutions” in the provinces across time
Table 0.15 Dimension 2.3 “Perception of fairness” in the provinces across time
Table 0.16 Dimension 3.1 “Solidarity and helpfulness” in the provinces across time
Table 0.17 Dimension 3.2 “Respect for social rules” in the provinces across time
Table 0.18 Dimension 3.3 “Civic participation“ in the provinces across time
Table 0.19 Structural characteristics and social cohesion in South African provinces
Table 0.20 Social cohesion and subjective well-being in the provinces
Table 0.21 Goodness-of-fit indices of LCA solutions
Table 0.22 Relative class sizes for LCA solutions
List of Figures
Figure 1.1 Constitutive elements of social cohesion (Leininger et al., 2021)
Figure 1.2 Measurement concept of the Bertelsmann Social Cohesion Radar
Figure 2.1 Provinces of South Africa
Figure 3.1 Overall index of social cohesion in South African provinces (2023)
Figure 3.2 Overall index of social cohesion in South African provinces across time
Figure 5.1 Average scores of dimensions in the four classes
Executive Summary
This report provides a comprehensive assessment of social cohesion in South Africa, focusing on its development over the past three years. The study uses data from the Khayabus Survey conducted by Ipsos South Africa. Following the conceptualization developed by the authors and their colleagues for Bertelsmann Stiftung, the study assesses social cohesion in three domains: social relations, connectedness, and focus on the common good. Each of these domains encompasses three dimensions of cohesiveness, namely the intactness of social networks, general trust in people, and acceptance of diversity for the domain ‘social relations,’ identification with one’s place of residence, trust in institutions, and perception of fairness for the ‘connectedness’ domain, and solidarity and helpfulness, respect for social rules, and civic participation for the domain ‘focus on the common good.’
The Bertelsmann concept of defining a country’s social cohesion level allows scores between 0 (no cohesion) and 100 (maximal social cohesion). Which score to evaluate as a sufficiently high score of social cohesion is a normative, not to say political decision. It has become customary to designate scores above 60 as speaking for a high level of social cohesion. Scores between 40 and 60 are designated as moderate, below 40 as low and below 20 as very low. Scores above 80 speak for a very high level of social cohesion, which has, however, not been reported for any country or other type of geopolitical entity yet.
The overall social cohesion index in South Africa is moderately high, neither specifically high nor low, but has seen a stable decline in the past three years. The highest scores for a single dimension were found for identification, followed by solidarity and helpfulness and social networks. The lowest scores were found for 'Perception of Fairness' and 'Respect for Social Rules.'
The study reveals that the most significant decline occurred in the cohesion dimension, respect for social rules. Other weakened dimensions include trust in institutions and perception of fairness. The level of solidarity and helpfulness remained stable, whereas civic participation and general trust in people have become slightly stronger.
The current level of cohesion in South Africa, 51.7, is almost identical to that found in Germany in 2023, where the index score was 52. As for the South African Provinces, social cohesion was found lowest in KwaZulu Natal (46.1), second lowest in Free State (49.4), and third lowest in Gauteng (50.5) in 2023.
Focusing on the relationship between the level of social cohesion in nine provinces and the characteristics of the provinces reveals surprising facts. The study shows that social cohesion in the South African provinces is negatively correlated with GDP and assets, i.e., higher assets resulting in lower social cohesion. This is mirrored by the finding that poverty positively correlates with social cohesion, with more poor people living in areas with higher social cohesion. The percentage of people living in rural areas is a good predictor of social cohesion, whereas city dwellers report lower levels of subjectively experienced social cohesion. High levels of language fractionalization correlate positively with social cohesion, and social cohesion tends to be lower when the median age is higher. In contrast, social cohesion elsewhere in the world is higher in countries with a high median age. Language and religious fractionalization results for South Africa also deviate from what had been reported for, e.g., Asia.
The study also found that social cohesion is perceived as lower among intense Internet users. These findings contradict the findings of the Bremen Social Cohesion Radar, which suggested that Internet use fosters social cohesion. Cohesion is high where many Whites and Coloureds live and low where many Indians and Blacks live. As one would expect, low percentages of singles in a geopolitical entity and high numbers of married people are related to high social cohesion.
Finally, the study also presents the results of a grouping procedure called latent class analysis (LCA), which isolates subgroups of South Africans who experience different deficits in social cohesion in their immediate life context. The results show that identification with the country of South Africa is not a variable that differentiates the classes: All South Africans are highly identified with their country.
LCAs identify four classes of South Africans: Class 1 (Critics), which is characterized by low tolerance levels and little respect for social rules; Class 2 (Integrated Sceptics), which is characterized by well-knit social networks, high levels of general trust and tolerance; Class 3 (Middle South Africa), which is characterized by an exceptionally high level of loyalty to the country and below-average acceptance of otherness, and Class 4 (Cohesive Communities), which is characterized by a strongly felt social cohesion in their typically rural life context.
Ultimately, the report examines the relationship between social cohesion and subjective well-being among South Africans. The data support the OECD countries' finding that high levels of social cohesion are related to well-being. Results show that higher levels of social cohesion in a geopolitical entity lead to a more positive outlook on life, with optimism and subjective well-being being considerably higher in provinces with high levels of social cohesion.
The study also examines subjective well-being in the four classes of respondents. The results show that in Class 4, the Cohesive Communities, families are seen as better off than a year ago, children have a bright future ahead, and life satisfaction has improved. Furthermore, the study found that the higher the social cohesion in a geopolitical entity, the more positive people's outlook on life is. Findings suggest that higher levels of social cohesion in a geopolitical entity will likely lead to increased life satisfaction and overall well-being. Exactly that finding underscores the necessity of political action to improve South Africa’s level of social cohesion. Whereas conceptual academic work on social cohesion sometimes claims that too high a level of social cohesion can cement the societal status quo and prevent progress, all empirical studies have shown that high levels of cohesion foster peaceful coexistence of various societal groups in respect, dignity, trust, and cooperation. Cohesion translates the social and economic structures (performance and output of the economy, living conditions) into quality of life (happiness, life satisfaction) directly experienced by individual members of society. In case, cohesion is neglected, one can expect societal polarization and political instability.
1. Introduction
Since the French Revolution with its famous motto ‘liberté, égalité, fraternité,’ discourse on the cohesion of geopolitical entities (countries/provinces/neighbourhoods) has seen waves of greater and lesser intensity, but one thing is clear: A healthy social entity needs fraternité or, in modern terminology, ‘social cohesion’ among its members. Social cohesion stands for the ability of societies to stick together or, as Leininger and colleagues put it, “the glue that holds society together” (Leininger et al., 2021: 2).
In recent years, when social cohesion has been discussed in South Africa, it has been with an increasingly critical undertone. The sentiment that the self-declared Rainbow Nation (Tutu, Mandela) is drifting apart rather than growing together has become stronger. However, research – mainly empirical – on social cohesion in South Africa is scarce. Our search for any available scholarly literature within the past 10 years yielded seven publications, two of which are reviews of a book by Ballantine et al. (2017) included in the count. The book by Ballantine and colleagues is a collection of essays by local academics and public figures about issues related to, amongst others, inequality, xenophobia, safety, gender-based abuse, political leadership, law, education, identity, sport, arts, and South Africa’s position in the world. A paper by Abrahams (2016) tracks the evolution of social cohesion over twenty years in South African politics, criticizing the instrumentalization of cohesion as a social policy concept exclusively towards a form of nation-building that seeks to solidify the hegemony of the ruling party.
A brief report by the South African Institute of International Affairs (2021) reviews the status quo and progress in religion, nationality, race and ethnicity, and LGBTQ+ rights. The report offers recommendations for improving the situation in these spheres and promotes the role of young people in fostering social cohesion. A paper by Burns and colleagues from the South African Labour and Development Research Unit emphasizes the importance of social cohesion as a social policy concept, also referring to studies conducted by the authors of the present report, critically reviews existing concepts of social cohesion proposed in the academic and policy discourse, and formulates a definition for its assessment in the South African society based on theoretical considerations: “Social Cohesion is the extent to which people are co-operative, within and across group boundaries, without coercion or purely self-interested motivation” (Burns et al., 2018, p. 10). Interestingly, Burns et al. (2018) identify the overlap between social cohesion and ubuntu, arguing that the two have become synonymous regarding nation-building and efforts to close South African society's cultural and racial divides.
To our knowledge, two existing empirical studies have defined and measured social cohesion in South Africa. Langer et al. (2017) define social cohesion in an African context as the interplay of three salient aspects: perceived inequalities, trust (interpersonal and institutional), and identity (national vs ethnic). Their measurement draws on data from 19 countries, including South Africa, from Round 3 (2005 – 2006), Round 4 (2008 – 2009), and Round 5 (2011 – 2013) of the Afrobarometer survey. For each aspect of a country, the authors calculate the proportion of respondents who provide those answers to the selected survey items that point to a stronger expression of cohesion. The resulting proportions for each aspect are then averaged by taking their arithmetic mean into a social cohesion index. According to the findings, overall cohesion in South Africa and its three aspects have recorded only minor ups or downs in the period studied. The level of identification was found to range from 0.612 (2005 – 2006) to 0.700 (2011 – 2013) and can be considered moderately high. The perception of equality was found in the range from 0.328 (2008 – 2009) to 0.469 (2011 – 2013) and can be qualified as low to moderately low. Trust was found in the range from 0.239 (2011 – 2013) to 0.293 (2005 – 2006) can be qualified as low.
From a comparative perspective, South Africa emerged in the middle of the country ranking on the overall level of cohesion and the perceived level of equality, in the lower half of the ranking on trust, and among the top countries on identification.
The second available empirical study (Leininger et al., 2021) similarly compares African countries. According to its authors, “cohesion is characterised by a set of attitudes and behavioural manifestations that includes trust, an inclusive identity and cooperation for the common good” (Leininger et al., 2021, p. 3). These three attributes unfold into two elements, encompassing horizontal or vertical relations among citizens and the state (see Figure 1.1): social trust and institutional trust, group identity and national identity, intergroup cooperation, and state-society cooperation. Leininger and colleagues operationalize their concept with items from the Afrobarometer, covering a varying set of African societies depending on the data availability for the particular element of cohesion: 17 to 18 in Round 3 (2005 – 2006), 20 in Round 4 (2008 – 2009), 28 to 34 in Round 5 (2011 – 2013), 32 to 36 in Round 6 (2014 – 2015). Their methodological approach measures the three attributes on a scale from 0 (low) to 1 (high). South Africa achieved scores in the range from 0.44 (2015) to 0.51 (2011) on trust, 0.3 (2015) to 0.43 (2006) on cooperation, and 0.42 (2015) to 0.74 (2011) on identity. These scores point to a downward trend in cohesion in South Africa for each attribute. In comparing African countries, South Africa ranked in the middle on trust, in the lower half on cooperation, and among the top countries on identity (except for the last year of observation, 2015). The work by Leininger et al. (2021), however, does not produce an overall index of cohesion, does not offer insights for more recent years since 2015, and does not venture into exploring the determinants of the country scores on the cohesion attributes or outcomes of cohesion.
Figure 1.1 Constitutive elements of social cohesion (Leininger et al., 2021)
The present report attempts to close the gap in the existing research on South Africa. It aims to provide an all-around theoretically founded and methodologically sound empirical assessment of social cohesion in South African society. In particular, our study attempts to:
measure the current degree of social cohesion in South Africa and its nine constituent provinces;
track how cohesion has developed in the period from 2021 to 2023;
identify structural characteristics from the thematic fields of economic situation, inequality and poverty, demography, diversity, and modernization that promote or hinder social cohesion;
explore which social groups experience a high or low level of cohesion;
investigate how social cohesion, i.e., the quality of society, relates to citizens’ well-being, i.e., quality of life.
To achieve these aims, we apply the measurement concept of the Bertelsmann Social Cohesion Radar, which was informed by a comprehensive literature review (Schiefer & van der Noll, 2017) and the input of experts on the topic. It defines cohesion as the “quality of social cooperation and togetherness of a collective, defined in geopolitical terms, which is expressed in the attitudes and behaviours of its members. A cohesive society is characterized by resilient social relations, a positive emotional connectedness between its members and the community, and a pronounced focus on the common good” (Dragolov et al., 2016: 6). These three domains unfold into three dimensions. The domain ‘Social relations’ measures the strength and resilience of individuals’ social ties (Dimension 1.1 – Social networks), the degree to which people trust others (Dimension 1.2 – Trust in people), and the extent to which people accept individuals of different background, lifestyle, and values as equal members of society (Dimension 1.3 – Acceptance of diversity). The domain ‘Connectedness’ measures the strength of individuals’ identification with the geopolitical entity (Dimension 2.1 – Identification), the degree to which individuals trust the entity’s institutions (Dimension 2.2 – Trust in institutions), and individuals’ perception that they are treated fairly and that material resources are fairly distributed (Dimension 2.3 – Perception of fairness). The domain ‘Focus on the common good’ captures the extent to which people feel and demonstrate responsibility for weak others (Dimension 3.1 – Solidarity and helpfulness), people’s willingness to abide by the rules of society (Dimension 3.2 – Respect for social rules), and their participation in society and political life (Dimension 3.3 – Civic participation). Figure 1.2 depicts the measurement concept.
Figure 1.2 Measurement concept of the Bertelsmann Social Cohesion Radar
Source: Dragolov et al. (2016)
A comparison of the Social Cohesion Radar to the approaches of Langer et al. (2017) and Leininger et al. (2021) shows that the three concepts overlap in several regards. First, social cohesion is a quantifiable quality of collectives, not individuals. Second, they all cover horizontal and vertical ties. Third, they all suggest that social cohesion should be assessed via a perception-based index, not based on objective socio-demographic indicators. Next, there is a considerable conceptual overlap in the emphasis on specific dimensions (aspects) of social cohesion. As already mentioned above, Langer et al. suggest including the extent of perceived inequalities (‘Perception of fairness’ in the SCR), the societal level of trust (‘Trust in people’ in the SCR), and the strength of people’s adherence to their national identity (‘Identification’ in the SCR). Leininger and colleagues also mention trust and identity, whereas their dimension of cooperation is called ‘Solidarity and helpfulness’ in the SCR.
The main difference between the three approaches lies in their conceptual scope. Whereas the SCR comprehensively describes the components necessary for a full-fledged assessment of the level of social cohesion in a given society, the two concepts based on the Afrobarometer remain somewhat piecemeal. In addition, one of the main advantages of the SCR approach is its leanness, a necessity also underscored by Leininger et al. (2021). On the one hand, the concept covers the essential components of social cohesion. At the same time, it leaves room for a systematic exploration of determinants (e.g., state of the economy, socio-economic exclusion) and outcomes (e.g., population well-being). For a critical review of the advantages and disadvantages of measurement concepts of cohesion that have been applied in empirical research, we refer readers to Delhey, Dragolov, and Boehnke (2023).
On a final note, the SCR has been utilized to assess social cohesion in 34 Western (EU and OECD) countries (Dragolov et al., 2016), 22 Asian countries (Delhey & Boehnke, 2018), the 16 federal states of Germany (Dragolov et al., 2016), 79 spatial planning regions of Germany (Arant et al., 2017; Boehnke et al., 2024), 78 neighbourhoods of the Free Hanseatic City of Bremen (Arant et al., 2016), the 32 federal entities of Mexico (Boehnke et al., 2019), and the seven regions of Kyrgyzstan (Larsen & Boehnke, 2016).
2. Measuring social cohesion
This section details the data and methodological approach employed for measuring social cohesion in South Africa.
2.1 Data
The current report offers empirical evidence from analyses performed on data from the Khayabus Survey. The data were collected by Ipsos South Africa and provided by the Inclusive Society Institute. The Khayabus Survey is a population-representative survey on various topics related to society and politics in South Africa. The survey initially included the sections Socio-Political Trends (SPT), Government Performance Barometer (GPB), and Party Image (PI). A fourth section, GovDemPol, was added in 2021. It has been fielded annually in at least two waves since 2019 among respondents aged 15 and above.
The analyses performed for this report draw on Waves 1 of the Khayabus survey, typically conducted from late May/early June to mid-July. Because the core set of indicators needed for assessing social cohesion along the Bertelsmann concept is part of the GovDemPol section, our analyses can only start with 2021. Data on three items crucial for the measurement concept, each belonging to the SPT section, had to be taken from the 2020 Khayabus, as they were not included in the 2021 survey. The most recent survey data available to us refer to the year 2023. Our report concentrates on survey respondents aged 18 and above, as several of the needed items were not included in the questionnaire for under-aged South Africans. The overall samples in the employed data encompass 3758 respondents in 2020, 3402 respondents in 2021, 3459 respondents in 2022, and 3519 respondents in 2023.
Table 2.1.1 Sample sizes of Khayabus – Waves 1
Table 2.1 offers detailed information on sample sizes achieved in Waves 1 of the Khayabus surveys from 2020, 2021, 2022, and 2023. The table provides a breakdown of the samples by province, as the present report also aims to measure social cohesion in the nine constituent provinces of South Africa. We refer readers unfamiliar with the nine provinces' geographic location and administrative borders to Figure 2.1. It is important to note that population sizes vary widely between the mostly urban Gauteng province, where well over a quarter of South Africa's adult population lives, and the mostly desert Northern Cape province, which encompasses less than 3%. The uneven distribution of the South African population across the provinces is reflected in the achieved sample sizes for the provinces, as evident from Table 2.1. The low sample sizes for the least populated provinces, e.g., Northern Cape, with only 64 respondents in 2023, do not necessarily reduce the representativity of the data concerning core socio-demographic characteristics of their population. We address this issue by calibrating the survey data with the population weights provided by Ipsos South Africa. Smaller sample sizes do, however, involve a larger standard error for sample statistics like percentages and means. In practical terms, this means that the precision of the measurements for Gauteng (NGP,2023 = 1168) is about four times higher than that for Northern Cape (NNC,2023 = 64) at the same variability in the data. Caution is, therefore, required when interpreting such statistics as estimates of the ‘true’ situation or opinion in the population of provinces for which low sample sizes are available.
Figure 2.1.1 Provinces of South Africa
Source: Apraku et al. (2018)
In addition to the above-addressed statistical issues, population sizes, and population density are closely related: In the Gauteng province, more than 800 people live per square kilometer, whereas in the Northern Cape province, the density figure is below 4 per square kilometer (Statistics South Africa, 2024a). Considering these stark differences is essential when evaluating our findings on levels and trends of social cohesion in South Africa.
2.2 Analytical approach
Below, we elaborate on the methodological approach for assessing social cohesion. We begin with the strategy for selecting Khayabus survey questions, also referred to here as items or indicators, to measure the nine dimensions of social cohesion in line with the Bertelsmann concept. We then turn to the approach for computing scores for the nine dimensions and the overall social cohesion index.
Item selection was conducted using a multi-step procedure. First, members of the research team – independent of each other – identified potential items for measuring the nine cohesion dimensions from the Khayabus questionnaire according to face validity. Members of the research team then jointly prepared a pool of items according to face validity. In the third step, items from the pool were subjected to an exploratory factor analysis for each dimension. Factor analysis is a statistical sorting procedure that analyses the matrix of item intercorrelations to separate items with a highly similar response pattern from items with a different response pattern and then sort them into distinct subgroups. The various subgroups of items (called factors) allow us to assess whether or not the items were selected appropriately according to their face validity as per the different dimensions of social cohesion. An important selection criterion is factor loading, which reflects how strongly an item is correlated with the other items sorted into the given factor. Item loadings should typically exceed .40 to be seen as sufficiently high. Items exhibiting sufficiently high factor loadings were retained. In the final step, we assessed the internal consistency of the scales formed by the selected items to measure a pertinent dimension. Cronbach’s α consistency coefficients should reach .90 for an excellent scale, .80 for a very good scale, .70 for a satisfactory scale, and minimally .30, or, in case of short scales, at least .10 times the number of items in the scale.
Several data preparation steps had to be taken before performing factor analyses. Where needed, the response options of the items were reverse coded so that a higher numerical value stands for a more vital expression of the pertinent aspect of cohesion. The response options of all items were rescaled to range from 0 (weakest expression of cohesion) to 100 (strongest expression of cohesion). If present, missing values on an item were substituted with the sample mean as the missingness rate was very low. Tables 2.2 to 2.4 document the selected items' factor loadings and the internal consistencies of the scales these items form for measuring the nine dimensions of social cohesion.
Table 2.2.1 Factor loadings of items for dimensions within Domain 1, “Social relations”
Table 2.2.2 Factor loadings of items for dimensions within Domain 2 “Connectedness”
Table 2.2.3 Factor loadings of items for dimensions within Domain 3 “Focus on the common good”
Readers will note that not all dimensions of social cohesion were measured equally well. This had several reasons. To begin with, the questionnaire offered a limited choice of indicators for some dimensions. This is why not all dimensions could be measured with at least three items. This pertained to Dimension 2.1 (Identification) and Dimension 3.1 (Solidarity and helpfulness), for which only two items could be included. Moreover, for Dimension 2.1, an item from the 2020 Khayabus had to be included, even with two items, to assess citizens’ identification with South Africa in 2021. Second, not all scales exhibit a high degree of homogeneity (level of intercorrelation) of the included items. This is particularly true for Dimension 2.1 and Dimension 3.3 (Civic participation).
After sorting items via factor analyses, the nine dimension scores were computed by calculating the arithmetic mean of the items determined to belong to a given factor. The overall cohesion index was calculated as the arithmetic mean of the nine dimension scores. Dimension and index scores for the provinces and South Africa were calculated by aggregating the individual-level data to the respective level via the population-weighted arithmetic mean. Scores for the dimensions and the overall index range from 0 (very low cohesion) to 100 (very high cohesion), where scores from 0 to 19.99 can be interpreted as pointing to a very low level of cohesion, 20 to 39.99 – low, 40 to 59.99 – medium, 60 to 79.99 – high, and 80 to 100 – very high.
3. Level and trend of social cohesion
In this section, we report findings on the level and trend of social cohesion in South Africa and its nine constituent provinces from 2021 to 2023.
3.1 Social cohesion in South Africa
Table 3.1 documents the annual level and trend of social cohesion in South Africa since 2021. We first present the findings for 2023, the most recent year for which data are available, and then proceed to the changes observed over time.
Level in 2023
In 2023, the overall social cohesion index for South Africa was 51.7, slightly above the theoretical midpoint of the measurement scale of 50. As such, the strength of social cohesion in South Africa can be qualified as moderate—neither high nor low. What is behind this result? A look at the single dimensions reveals the strong and weak spots of cohesion in South Africa that jointly produce its moderate overall level.
Table 3.1.1 Social cohesion and its dimensions in South Africa across time
Dimension 1.1, ‘Social networks,’ scored 58.8 in 2023. The intactness of citizens’ social networks is currently the third strongest dimension in South Africa. The strength of this dimension can be qualified as moderate, but it should be noted that the result is very close to the lower bound of the interval of high scores (60). To exemplify with one indicator for this dimension: About 52 % of the respondents agreed with the statement “I entertain friends from different population groups at home or another place,” 25 % disagreed, and 20 % positioned themselves in between (see Table A.1 of the Appendix).
Dimension 1.2, ‘Trust in people,’ can also be found in 2023 in the upper half of the measurement scale, with a score of 54.1. The level of trust that South Africans place in others qualifies, thereby, as moderate. To exemplify the finding with one indicator for this dimension: 19 % of the respondents trust people in their community completely, 62 % only somewhat or not very much, and 18 % not at all (see Table A.2).
Dimension 1.3, ‘Acceptance of diversity,’ achieved 2023 a score of 46.8. The numeric result qualifies the tolerance level in South African society still as moderate. However, it should be noted that it falls within the lower half of the measurement scale, unlike the previous two dimensions from the Domain ‘Social relations.’ To exemplify with one indicator for ‘Acceptance of diversity’: 14 % of the respondents trust coloured South Africans completely, 58 % only somewhat or not very much, and 26 % not at all (see Table A.3).
Dimension 2.1, ‘Identification,’ scored 71.2, the highest among all dimensions in 2023. Identification is the most pronounced aspect of social cohesion in South Africa and the only dimension that can be qualified as high. This is manifested, for example, in the responses to the statement “I am proud to be South African”: 70 % agreed, 17 % disagreed, and 11 % positioned themselves in between (see Table A.4).
Dimension 2.2, ‘Trust in institutions,’ achieved in 2023 a score of 46.6. Just like ‘Acceptance of diversity’, the extent of trust citizens have in the country's institutions can be considered moderate. However, it falls within the lower half of the measurement scale. To exemplify with one indicator: 46 % of the respondents agreed with the statement “Elections are free and fair,” 33 % disagreed, and 17 % positioned themselves in between (see Table A.5).
Dimension 2.3, ‘Perception of fairness,’ scored in 2023 at 40.5. It is the second weakest aspect of social cohesion in South Africa. The extent to which people perceive the distribution of material resources as fair can be qualified as moderate. However, it should be noted that the result is very close to the upper bound of the interval of low scores (40). To exemplify with one indicator for this dimension: 27 % of the respondents stated the government is doing very well or fairly well at narrowing the income gap between races, whereas 68 % stated the government is handling this issue not very well or not at all well (see Table A.6).
Dimension 3.1, ‘Solidarity and helpfulness,’ achieved in 2023 a score of 59.1. With this result, it is the second most vital aspect of social cohesion in South Africa. The extent to which people help the weak members of society can be qualified as moderate, but it should be noted that it is very close to the lower bound of the interval of high scores (60). To exemplify with one indicator for this dimension: 58 % of the respondents agreed with the statement “I actively look for ways in which I can support people who are less fortunate than I am”, 21 % disagreed, and 20 % positioned themselves in between (see Table A.7).
Dimension 3.2, ‘Respect for social rules,’ achieved in 2023 a score of 33. This is the weakest aspect of social cohesion in South Africa. The extent to which people perceive that rules are observed is low. To exemplify with one indicator for this dimension: 21 % of the respondents stated that the government is doing very well or fairly well at reducing the crime rate, whereas 78 % stated not very well or not at all well (see Table A.8).
Dimension 3.3, ‘Civic participation,’ scored at 55.5 in 2023. Just like ‘Trust in people’, the involvement of citizens in society and political life can be qualified as moderate. One indicator for this dimension: 47 % of the respondents agreed with “I actively work for the welfare of my community”, 31 % disagreed, and 20 % positioned themselves in between (see Table A.9).
The results for the dimensions form a nuanced profile of cohesion. None of the three domains of social cohesion (‘Social relations,’ ‘Connectedness,’ and ‘Focus on the common good’) exhibits only deficits or strengths for all dimensions. Interestingly, the three top-scoring dimensions (‘Identification,’ ‘Solidarity and helpfulness,’ and ‘Social networks’) stand out as strong to moderately strong anchors of their respective domains. However, two domains are imbalanced: ‘Connectedness’ by the moderately low ‘Perception of fairness’ and ‘Focus on the common good’ by the low ‘Respect for social rules.’ If one should point out the glue that holds South African society together, this would undoubtedly be citizens’ strong identification with the country, their solidarity with their weaker fellow citizens, and the functioning of their social networks. On the other hand, what could destabilize South African society is the perceived lack of respect for rules and the perceived lack of distributional fairness.
Is the level and profile of cohesion in South Africa unique? A comparison to Germany (Boehnke, Dragolov, Arant & Unzicker, 2024) reveals that the current level of cohesion in South Africa is almost identical to that found for Germany in 2023, where the index score was 52. Nevertheless, despite the similar overall levels, the two countries have distinct patterns of strengths and weaknesses in the single dimensions. In 2023, the strongest dimension in Germany was ‘Acceptance of Diversity’ (69), followed by ‘Respect for Social Rules’ (67), whereas the weakest dimensions were ‘Solidarity and helpfulness’ (34) and ‘Perception of Fairness’ (35). Extending the scope to the comparison of social cohesion in 34 EU and OECD member states (Dragolov et al., 2016), we find considerable similarities in the pattern of strengths and weakness between South Africa and Israel. The society of Israel exhibits similarly strong identification, moderate solidarity, social networks, and trust in people, as well as pronounced deficits
in the perception of fairness and respect for social rules. No country among the 22 Asian studied (Bertelsmann Stiftung, 2018) exhibits this profile.
Trend over time
Although the overall social cohesion index in South Africa emerges as moderate, with scores in the upper half of the measurement scale, it has been on a stable decline (see Table 3.1). Over the past three years, it declined from 53.5 in 2021 to 52.4 in 2022 and 51.7 in 2023, thereby losing 1.8 points.
The downward trend is evident for most of the dimensions. The steepest decline was observed for ‘Respect for social rules.’ In 2021, this dimension still ranked as moderate with a borderline score of 40.3 but lost 7.3 points over time to qualify in 2023 as low. Other dimensions of cohesion that have weakened from 2021 to 2023 are, in this order, ‘Trust in institutions’ (-4.1 points), ‘Perception of fairness’ (-2.9 points), ‘Social networks’ (-2.4 points), ‘Identification (-2.1 points), and ‘Acceptance of diversity’ (-0.3 points). Besides ‘Respect for social rules,’ social networks are the only dimension that has experienced a downgrade from a previously higher to a lower category. In 2021, ‘Identification’ was not the sole dimension that ranked as high, but also ‘Social networks’ with a score of 61.3. In contrast, ‘Solidarity and helpfulness’ has remained strikingly stable, whereas ‘Civic participation’ (+1.0 points) and ‘Trust in people’ (+1.6 points) have become slightly stronger. We refer readers to Tables A.1 to A.9 of the Appendix for changes over time in the responses to the indicators of the respective dimensions.
Is the trend of cohesion in South Africa unique? In this regard, we can only compare to Germany (Boehnke et al., 2024). Despite their identical overall levels of social cohesion in 2023, South Africa and Germany do not have the same starting positions. While the decline in the overall index for South Africa amounts to only 1.8 points, the one observed for Germany within the same three-year period is alerting: The strength of social cohesion declined from a stable value of 61 in 2017 and 2020 by 10 points in 2023. Cohesion in Germany slid from a high down to a moderate level. Moreover, between 2020 and 2023, every dimension of cohesion in Germany weakened considerably. The steepest declines were observed for ‘Solidarity and helpfulness’ (-14 points), ‘Acceptance of diversity’ (-13 points), ‘Identification’ (-11 points), and ‘Social networks’ (-10 points).
3.2 Social cohesion in the nine provinces
Level in 2023
Zooming into the provinces, we find some variation across these administrative units. Figure 3.1 maps the strength of the overall social cohesion index in the nine provinces. Table 3.2 documents the annual level and trend over the three years examined here. In 2023, social cohesion was found lowest in KwaZulu Natal (46.1), second lowest in Free State (49.4), and third lowest in Gauteng (50.5). Social cohesion was slightly above the country average of 51.7 in all other provinces. Limpopo (58.3) emerged as a province with a level of social cohesion considerably above the country average. However, it should be noted that all provinces rank in the interval of the measurement scale, referring to a moderate level of cohesion.
Tables A.10 to A.18 of Appendix A document the provinces' performance on the single dimensions of cohesion. Interestingly, the leader Limpopo achieved only middle positions on the dimensions ‘Social networks’, ‘Trust in people’, and ‘Acceptance of diversity’ but consistently ranked highest or second highest on all other dimensions of cohesion. In contrast, KwaZulu-Natal ranked consistently lowest or second lowest on all nine dimensions.
Figure 3.2.1 Overall index of social cohesion in South African provinces (2023)
Note: The map applies the colour scheme displayed above to visualize the strength of social cohesion in 2023 across the nine provinces.
Table 3.2.1 The overall index of social cohesion in South African provinces across time
Trend over time
As evident from Table 3.2, social cohesion has declined from 2021 to 2023 in six of the provinces, most noticeably in Gauteng (-3.6 points), Free State (-4.2), and KwaZulu-Natal (-5.1 points). These three provinces ranked last in social cohesion in 2023. Cohesion has remained, by and large, stable only in the Western Cape (-0.8 points). In contrast, social cohesion has improved in the Eastern Cape (+3.8 points), Limpopo (+1.5 points), and North West (+1.4 points). Figure 3.2 depicts these developments in the overall cohesion index across the provinces.
Figure 3.2.2 Overall index of social cohesion in South African provinces across time
Note: The figure shows the scores of the nine provinces on the overall index of social cohesion in 2021, 2022, and 2023.
As to the trend in the single dimensions, Tables A.10 to A.18 reveal that ‘Trust in people’ and ‘Civic participation’ are the only dimensions that improvements for almost all provinces can characterize. ‘Trust in people’ has weakened only in Limpopo (-4.4 points) and more pronouncedly in Northern Cape (-11.6 points). ‘Civic participation’ has declined only in KwaZulu-Natal (-5.9 points) and Northern Cape (-10.4 points). The picture is reversed for the dimensions of ‘Trust in institutions’, ‘Perception of fairness,’ and ‘Respect for social rules,’ in which almost all provinces experienced declines. ‘Trust in institutions’ has increased only in Western Cape (+3.3 points), ‘Perception of fairness’ – only in Western Cape (+3.2 points) and Limpopo (+6.0 points), and ‘Respect for social rules’ – only in Western Cape (+3.4 points). Interestingly, although ‘Identification’ emerged as the glue that holds the South African society together, all provinces but three experienced declines in identification from -3.0 points (Gauteng) to -10.3 points (Mpumalanga). Identification has improved only Limpopo (+6.3 points), Eastern Cape (+8.2 points), and North West (+8.5 points).
In the subsequent section, we investigate which structural characteristics of the provinces may be at play in promoting or hindering social cohesion.
4. Structural influences on social cohesion
This section aims to find evidence on structural determinants of social cohesion. For this purpose, we explore the relationship between the level of social cohesion in the nine provinces and selected characteristics of the provinces from the following thematic fields: economic situation, inequality and poverty, demography, diversity, and modernization. The focus on these aspects is not arbitrary: Our studies on Western and Asian societies as well as the federal states and spatial planning regions of Germany demonstrated empirically that aspects from these thematic fields act as determinants rather than outcomes of social cohesion (Dragolov et al., 2016; Bertelsmann Stiftung, 2018; Arant, Dragolov & Boehnke, 2017; Boehnke, Dragolov, Arant & Unzicker, 2024).
4.1 Data and method
To touch on the economic situation in the provinces, we use data on the gross domestic product per capita[1] in Rand (Statistics South Africa, 2024a), Human Development Index (Global Data Lab, 2024), and unemployment rates – official and expanded (Statistics South Africa, 2023b). We measure poverty using one subjective indicator – the share of households in a province who perceive themselves as poor (Statistics South Africa, 2024c), and objective indicators concerning three definitions of the poverty line[2] – the share of the population below the food poverty line, the lower-bound poverty line, and the upper-bound poverty line (own calculations based on CRA, 2023). We employ the Gini index of income inequality and the P90/P10 ratio (own calculations based on CRA, 2023) to measure inequality[3]. We draw on data from Census 2022 (Statistics South Africa, 2023a) for the remaining thematic fields. In particular, as to demographics, we consider population density, the share of urban and rural population, the share of singles and married citizens, and the population's median age. To tap into diversity, we use the share of Blacks, Whites, Coloured, Indian/Asian, and Other races; the share of immigrants; as well as ethnic, linguistic, and religious fractionalization[4] (own calculations based on Statistics South Africa, 2023a). To touch on modernization, we use the share of citizens with completed primary, secondary, and post-secondary education, the share of citizens owning a computer and a cell phone, and the share of the population without access to the internet.
All indicators refer to 2021 or 2022, preceding the most recent measurement of social cohesion from 2023. The intentional time lag—earlier measurement of the structural characteristics of the provinces and later measurement of social cohesion—introduces a certain degree of temporal order in the analyses. It cannot prove the existence of a causal relationship, but it can increase the plausibility of attributing causality.
Each of the above-listed structural characteristics of the provinces was subjected to a correlation test with the level of social cohesion. Two variables are correlated when changes in one are (closely) followed by changes in the other. A correlation can be positive (the more of Variable A, the more of Variable B) or negative (the more of Variable A, the less of Variable B). The strength of the association is reflected in the correlation coefficient, which can range from 0 (no correlation) to ±1 (perfect correlation). Typically, a correlation of size below |.10| is very weak and not worth interpreting, between |.10| and |.30| – weak, between |.30| and |.50| – moderate, and above |.50| – strong.
A considerable obstacle arises from the sample size for the correlations on the level of provinces. Of course, the overall sample size in each survey wave is large and has sufficient statistical power. Power (in mathematical-statistical theory) means that a given sample is large enough to corroborate a particular effect as likely ‘true’ about the population from which the sample was drawn – South Africa in this case. On the level of provinces, however, showing that a specific correlation is sizable enough to conclude that it is significant (i.e., likely true in the nine provinces) is problematic due to the number of provinces – only nine, unlike the number of surveyed individuals – over three thousand in each year. This may be surprising at first glance because more people have been surveyed in the provinces, with a minimum of 64 in Northern Cape and a maximum of 1168 in Gauteng in 2023. However, social cohesion is not a characteristic of individuals but of geopolitical entities, and the latter, we only have nine—the provinces—in South Africa.
The low number of cases (provinces) means that only extremely high correlations can reach statistical significance. Mathematically, significance is a function of sample size (the higher, the more likely it is for a particular coefficient to be significant) and data variability (the higher the variance of the included data, the less likely it is that a specific coefficient is significant). Given these constraints, we disregard the significance of the correlation coefficients in our reporting and interpretation, focusing instead on the tendency in the data. Thus, we follow the appeal of a respectable number of scientific community members to ditch p-values (Wasserstein, Schirm & Lazar, 2019). The results from additionally performed bivariate biserial Pearson correlations, for which the province-level characteristics were disaggregated to the individual-level data set for 2023 (N = 3519), show that only 6 of altogether 31 associations may not be considered significant (see Table A.19 of Appendix C). This finding supports our decision to focus on tendencies instead of discarding associations because they do not meet a criterion for statistical significance.
Besides significance, the number of provinces is critically low for performing Pearson correlations. As a parametric test, the Pearson correlation involves assumptions that cannot be fulfilled with the data on the level of provinces. We, therefore, resort to Spearman correlations as a non-parametric, assumption-free alternative. A Spearman correlation is, in essence, a Pearson correlation performed on ranked data. The significant distinction between the two approaches is that a Pearson correlation considers the exact distances among the observations on each variable. In contrast, a Spearman correlation considers only whether there are differences, disregarding their size. For example, in 2022, the wealthiest province per capita GDP was Gauteng, with 96,252 Rand, and the poorest was Eastern Cape, with 54,805 Rand. The Pearson correlation will be influenced by the numeric difference of 41,447 Rand between the two provinces, whereas the Spearman correlation will only consider which province has the higher value.
One implication is that the Spearman method is not sensitive to outliers in the data – observations (provinces) with an extremely large or extremely low value on a characteristic of interest. Given the critically low sample size of nine provinces, the Spearman method is advantageous compared to a Pearson correlation which may be quickly and heavily biased by an outlier. Because in all our previous studies on cohesion, the data allowed us to apply the Pearson correlation method, we performed Pearson correlations for the present report. Interested readers can find those in Table A.19 of Appendix C. In a few instances, we observe great discrepancies – mostly in size but also in direction. As discussed at length above, to stay on the safe side, we report and interpret the findings from the Spearman correlation tests.
On a final note, we performed both bivariate correlations and partial correlations for GDP, because both in our 34-country OECD study (Dragolov et al., 2016) and our study of 22 Asian countries (Bertelsmann Stiftung, 2018), GDP was highly positively related to social cohesion: The more prosperous a society, the more cohesive it is. Partialing GDP out of a relationship removes the influence of GDP on both variables involved. This makes it possible to speak of associations between a given structural characteristic of the provinces and social cohesion, independent of their economic prosperity. In the section below, we report and interpret the partial correlations.
4.2 Results
Table 4.1 documents the relationships between the structural characteristics of the nine provinces and social cohesion per thematic field. The empirical findings for South Africa present several surprises concerning what has been previously found for Western and Asian societies.
Economic situation
The bivariate correlation between per capita GDP (in a province) and social cohesion (in the same province) was found at ρ = -.38. The relationship is negative and moderate in size. It informs that social cohesion tends to be lower in more economically affluent provinces. This result is striking as it goes against the positive association between GDP and cohesion that was consistently found in all our previous studies on Western and Asian societies.
The correlation between the Human Development Index and cohesion emerged positive and weak in size (ρ = .25). People-centered economic progress (Gross National Income coupled with mean years of schooling and life expectancy) appears conducive to social cohesion.
Both measures of unemployment – the official and expanded unemployment rates – exhibited negative, though only weak, associations with social cohesion (ρ = -.15 and ρ = -.21, respectively). Provinces in which more unemployed people reside tend to have weaker social cohesion.
Poverty and inequality
The evidence presents mixed findings on poverty. Whereas the subjective indicator exhibited a moderately negative correlation with social cohesion (ρ = -.40), the three objective indicators were found to correlate positively and moderately in the range from ρ = .37 to ρ = .40. Social cohesion tends to be lower in provinces where more households perceive themselves as poor. However, it tends to be higher in provinces where more people objectively fall below the poverty lines. Though at first glance puzzling, the results for objective poverty can be explained with the social welfare programs targeted at supporting poor citizens.
The correlations of the social cohesion index with both measures of income inequality were found to be consistently negative. Social cohesion tends to be lower in provinces with larger inequality in income. Interestingly, the correlation with the Gini index is much weaker (ρ = -.19) than that with the P90/P10 ratio (ρ = -.52). The Gini index considers the entire income distribution. In contrast, the P90/P10 ratio contrasts the income at the top of the distribution (90th percentile) to the income at the bottom (10th percentile). The latter focuses on inequality, which is more visible and more accessible for ordinary citizens to perceive. The top-to-bottom income ratio is 38 in the Free State and 11 in the Northern Cape. These values inform that the top earners' income is 38 times higher than that of poor citizens in the Free State; in Northern Cape – ‘only’ 11 times higher. Vast discrepancies in income tend to be detrimental to social cohesion.
Table 4.2.1 Structural characteristics and social cohesion in South African provinces
Demography
Population density exhibited a weak negative relationship with social cohesion (ρ = -.27). Social cohesion tends to be lower in more densely populated provinces. This finding corresponds with the associations of the cohesion index with the shares of urban (ρ = -.26) and rural population (ρ = .26). Though the relationships are only weak in size, they indicate that social cohesion tends to be lower in more urbanized provinces and, in contrast, higher in provinces with a larger share of rural population.
Marital status was found to correlate strongly with social cohesion. Social cohesion tends to be lower in provinces with a large share of singles (ρ = -.69) and higher in provinces with a larger share of married citizens (ρ = .55). This finding suggests that families contribute strongly to cohesion in the South African society.
We found a moderately negative association (ρ = -.36) with median age. Social cohesion tends to be lower where the population’s median age is higher. The association was reversed in Asia: Social cohesion was higher in Asian countries with a higher median age.
Diversity
The composition of the provinces’ population regarding race, migration background, language, and religion seems to be weekly to moderately related to social cohesion. Provinces with larger shares of Blacks (ρ = -.12) and Indians/Asians (ρ = -.15) tend to have weaker social cohesion. In contrast, cohesion tends to be higher in provinces with larger shares of Whites (ρ = .27), Coloured (ρ = .42), and other races (ρ = .19). Provinces with more immigrants were also found to have stronger levels of cohesion (ρ = .50).
The fractionalization measures offer findings that generally follow the tendencies mentioned above. Ethnic (racial) fractionalization exhibited a positive, though only weak, correlation with social cohesion (ρ = .12). The relationship with linguistic fractionalization emerged as positive and moderate (ρ = .41), whereas that with religious fractionalization was found to be weak and negative (ρ = -.14). Racial and linguistic diversity in the provinces seem to contribute to social cohesion, whereas religious diversity appears to harm it.
Modernization
The evidence is puzzling concerning educational attainment. Whereas the share of citizens with completed primary education exhibited a positive and moderate correlation with social cohesion (ρ = .34), the shares of citizens with completed secondary education exhibited a negative and very strong correlation (ρ = -.76) – in fact, the strongest of all associations explored. The correlation with the share of citizens with completed post-school education also emerged negative, though at the border of being negligible (ρ = -.10). Provinces with better and more highly educated citizens tend to have lower levels of cohesion. A possible explanation for these surprising results could be the economy of the country which does not deliver jobs up to the expectations of the better educated citizens.
Access to modern information and communication technology (computers, cell phones, and the internet) weakens social cohesion. The social cohesion index correlation is weakly negative with the share of computer owners (ρ = -.22) and cell phone owners (ρ = -.20) and strongly positive with the share of the population without access to the internet (ρ = .61).
What brings social cohesion in the South African provinces forward is people-centered economic progress, rural population, marriages, racial and linguistic diversity, and immigration. As hazards to social cohesion emerged: unemployment, felt poverty, income inequality, high population density and urbanization, single life, older population, religious diversity, and the penetration of modern information and communication technology.
5. Individual experiences of social cohesion
The previous sections of this report examined the levels and trends of social cohesion in South Africa and its provinces. Correlational analyses on the level of the provinces offered insights into potential structural characteristics that determine the local level of cohesion. In this section, we go down to the level of individual respondents to explore which population groups are at risk of experiencing low cohesion in South Africa.
5.1 Data and method
There are several methodological approaches for identifying groups at risk of experiencing low cohesion, each involving different assumptions. One possibility is to perform separate analyses relating the individual scores on the overall cohesion index and its nine dimensions to the respondents' relevant socio-demographic and economic characteristics. This approach will likely lead to many difficult results to systematize. In order to reduce the complexity without a significant loss of information, we prefer to identify classes (distinct groups) of respondents based on the pattern of their scores on the nine dimensions of social cohesion. The resulting classes are characterized by similarities within and dissimilarities across the classes concerning the experience of the nine aspects of cohesion by the respondents who belong to them. In a second step, we relate class membership to socio-demographic and socio-economic characteristics. In simpler terms, we investigate how the experience of social cohesion is related to individual characteristics. An example could be rich and poor citizens experiencing different levels of social cohesion.
To classify respondents into groups with distinct experiences of cohesion, we employ the Latent Class Analysis statistical procedure. To cite from the abstract of a recent overview paper (Weller et al., 2020): “Latent class analysis (LCA) is a statistical procedure used to identify qualitatively different subgroups within populations who often share certain outward characteristics. The assumption underlying LCA is that membership in unobserved groups (or classes) can be explained by patterns of scores across survey questions, assessment indicators, or scales.” We take respondents’ scores on the nine dimensions of social cohesion in 2023 as the basis for the LCAs performed here.
LCAs are typically undertaken sequentially. The statistical procedure is programmed so to come up with different numbers of groups, usually starting with two distinct groups (classes) and continuing until several groups (classes) are distinguished that offer plausible pathways of interpretation: Who are the people grouped into Class 1, Class 2, …, Class k? It is customary to summarize interpretations by labelling the different classes in a way that best characterizes their response patterns. Next to interpretability, specific indices of goodness-of-fit aid the decision of how many groups are most plausible to extract from the available data (Weller et al., 2020). These include the Akaike Information Criterion (AIC) and the (sample-size-adjusted) Bayesian Information Criterion (BIC, saBIC), which do not have pre-defined thresholds but inform comparisons of solutions: The solution with the lower AIC and (sa)BIC would be deemed better. Goodness-of-fit indices with pre-defined thresholds include the Entropy coefficient H and the Average Probability of Class Membership (APCM): Each should be greater than .90 for excellent fit or .80 for acceptable fit. Table A.21 of Appendix D documents the goodness-of-fit indices of six different LCA solutions. Table A.22 of Appendix D shows the population-weighted relative sizes of the classes in the total sample (N = 3519) from the various LCA solutions.
Based on the resulting goodness-of-fit indices of the LCA models we specified and considering the classes' interpretability, we selected the LCA model producing four classes.
5.2 Four classes of experience
Table 5.1 provides an overview of respondents’ average scores on the nine dimensions of cohesion, which served as the basis for the LCA and the overall index of social cohesion within each class. Class 1 encompasses 16.7 % of the respondents. It is characterized by low cohesion with an average score of 32.9 out of 100 points. Classes 2 and 3 encompass 32.8 % and 29.8 % of the respondents, respectively. Both exhibit moderate social cohesion with an average score of 51.0 in Class 2 and 51.6 in Class 3. Class 4, encompassing 20.7 % of the respondents, exhibits a high level of social cohesion with an average score on the overall index of 68.3 points.
Table 5.2.1 Social cohesion and its dimensions in the four classes
A closer look at the average scores on the nine dimensions reveals the class differences (see Table 5.1 and Figure 5.1). Likely due to the large sample sizes of the four classes, a series of Scheffe pairwise comparisons from one-way analyses of variance for each of the dimensions informs of significant differences (p ≤ 0.05) among all classes on all dimensions except for: Dimension 1.2 ‘Trust in people’ for Class 2 (M = 69.3) vs Class 4 (M = 71.1), Dimension 3.1 ‘Solidarity and helpfulness’ for Class 2 (M =60.4) vs Class 3 (M = 59.7), and Dimension 3.3 ‘Civic participation’ for Class 2 (M = 54.8) vs Class 3 (M = 54.5).
Respondents within Class 1 (low cohesion) exhibit only a high level of identification (62.4). The strength of their social networks (43.8), their solidarity with weak others (41.5), and their involvement in civic life (42.8) are moderate, yet tending toward weak. Members of Class 1 place low trust in others (25.4) and in institutions (27.2) and have a low perception of fairness (21.9). Their tolerance for diversity is very low (16.5). These respondents also have a very low perception that social rules are respected (14.4). Class 1 can be described as ‘Critics’.
Figure 5.2.1 Average scores of dimensions in the four classes
Note: The figure visualizes the average scores of the social cohesion dimensions in each of the four classes of respondents.
Members of Class 2 (moderate cohesion) exhibit a mixed pattern. They are well-networked socially (61), place high trust in others (69.3), and have high levels of tolerance for diversity (63.0), identification (67.4), and solidarity with weak others (60.4). However, the trust they place in institutions is low (35.6), and they perceive low levels of fairness (27.4) and respect for rules (20.1). Their involvement in civic life is moderate (54.8). Class 2 can be described as ‘Integrated sceptics.’
Members of Class 3 (moderate cohesion) rate most social cohesion aspects on the middle level. The strength of their social networks (57) and solidarity with others (59.7) are moderate to high. Moderate are the levels of trust they place in institutions (54.9), perceived fairness (49.3), and civic participation (54.5). These respondents place moderate to low trust in others (41.5) and have a moderate to low perception that rules are observed (40.1). Their tolerance for diversity is, however, low (33.0). Their identification with the country is the second highest (74.3) among all four classes. If these were findings on the US society, we would label this class ‘Middle America,’ with strong loyalty to the country and below-average acceptance of otherness. Class 3 can be described as ‘Middle South Africa.’
Members of Class 4 express high levels (60 to 80 points) of social cohesion in all aspects, but ‘Respect for social rules’ has a moderate expression, though tending towards a high one (58.3). It should be noted that the strength of identification with the country tends to be very high (79.8). Class 4 can be described as ‘cohesive communities’.
Across all four classes, Dimension 2.1, ‘Identification,’ was found to be consistently high. In contrast, the deficits in Dimension 3.2, ‘Respect for social rules,’ emerged yet again as an antithesis to citizens’ loyalty to the country.
5.3 Socio-demographics of the four classes
In this section, we explore which individual characteristics are typical for the four classes with distinct experiences of social cohesion. We do so using the following socio-demographic and socio-economic characteristics: biological sex (male, female), age group (18 to 24, 25 to 44, 45 to 64, 65 years and above), race (White, Black, Indian/Asian, Other), primary language (English, Afrikaans, Indigenous), marital status (single, married or living as married, widowed or divorced), community size (rural or village, town or city, metropolitan area), education (post-secondary, secondary, lower or none), employment status (employed, not in workforce, unemployed), and income class[5] (low, lower-middle, upper-middle, high, missing). Data on these characteristics stem from Wave 1 of Khayabus 2023.
The four classes were characterized in the framework of separate chi-square tests of independence between respondents’ class membership and the respective individual characteristics of interest. Table 5.2 documents the population-weighted relative frequencies (%) of the socio-demographic and socio-economic categories in the total sample and each of the four classes and the respective result from the chi-square test of independence and Cramer’s V coefficient of effect size. Due to the large sample size, all tests but one emerged as statistically significant, whereas effect sizes were consistently small. We, therefore, focus on the tendencies in the data.
Class 1, the Critics, is characterized by an overproportional representation of men (51.9 %), speakers of indigenous languages (76.5 %), dwellers in metropolitan areas (54.9 %), persons with completed secondary education (58.4 %), unemployed (38.3 %), and persons who have refused to report their household income (46.1 %). In addition, the shares of respondents from the age group 18-24 years (21.2 %), Blacks (80.1 %), singles (62.7 %), and members of the high-income class (14.9 %) tend to be slightly larger than in the total population.
Class 2, the Integrated skeptics, is characterized by an overproportional representation of respondents from the age groups 25-44 years (54.1 %) and 65+ years (4.7 %), non-Black races (Whites – 13.6 %, Indian/Asian – 3.6 %, Other – 9.7 %), speakers of English (13 %) and Afrikaans (18.1 %), widowed or divorced (10.6 %), respondents who have completed post-secondary education (18.4 %), respondents who are not in the workforce (19.8 %), and members of the high-income class (15.2 %). In addition, the shares of women (53.6 %) and members of the upper-middle income class (15.2 %) tend to be slightly larger than in the total population.
Class 3, Middle South Africa, is characterized by an overproportional representation of women (55.1 %), Blacks (81.8 %), singles (63.6 %), dwellers in towns or cities (26 %), and members of the lower-middle income class (14.3 %). In addition, the shares of respondents from the age group 18-24 years (21.6 %), speakers of indigenous languages (75.4 %), dwellers in rural areas or villages (30.5 %), respondents with lower than secondary or no formal education (33.4 %), unemployed (37 %), but also members of the upper-middle income class (16.8 %) tend to be somewhat larger than in the total population.
Table 5.3.1 Socio-demographic and economic characteristics of the four classes of respondents
Class 4, the Cohesive Communities, is characterized by an overproportional representation of respondents from the age group 45-64 years (28 %), married or living together as married (32.9 %), dwellers in rural areas or villages (33.2 %), respondents with lower than secondary or no formal education (34.4 %), employed (47.8 %), and members of the low-income class (19.2 %) but also of the upper-middle income class (17.8 %). In addition, the shares of respondents from the age group 18-24 years (21.6 %), Blacks (79.8 %), and respondents not in the workforce (20.6 %) tend to be larger than in the total population.
6. Social cohesion and subjective well-being
In the present section, we inspect the data to determine whether they support the consistent finding from our previous studies on the topic that high levels of social cohesion are related to greater (subjective) well-being. We employ the following five items from the Khayabus survey as proxies of well-being:
Think of the way your family lives, would you say that your family is… better off than a year ago / about the same / worse off than a year ago?
And how do you think your family’s lives will be in a year’s time? Do you think your family will be… better off than today / about the same / worse off than today?
Please think about your children or the children of family or friends. What do you think the future holds for these children? Do you think that… they have a bright future ahead of them / they have a bleak future ahead of them?
And your satisfaction with life? Has it… improved/staying the same/worse compared to a few months ago?
On a scale from 1 to 5 please indicate whether you (1) strongly disagree, (2) disagree, (3) neither agree nor disagree, (4) agree or (5) strongly agree with the following statement: I am seriously considering emigrating to another country in the next year or so.
We perform analyses both on the level of provinces and of individuals.
6.1 Provinces
We aggregate the individual responses to the five items to measure well-being on the level of provinces. For each province, we take the respective share of the positive response option to each of the four items with categorically scaled answers (better off, bright future, improved) and the arithmetic mean of the individual responses to the Likert-scale item tapping on emigration.
We apply the same methodological approach as in Section 4, which explored associations between several structural characteristics of the provinces and the index of social cohesion. Table A.20 of Appendix C documents the biserial bivariate correlations on the individual level and the bivariate and partial Pearson correlations on the level of provinces. As in Section 4, we report and interpret the results from the Spearman correlations partialled for GDP (see Table 6.1).
Table 6.1.1 Social cohesion and subjective well-being in the provinces
The share of respondents evaluating their family’s life as better off today than a year ago correlated strongly and positively with the level of social cohesion in respondents’ province (ρ = .60). The same was found for the share of respondents evaluating their life satisfaction as improved in comparison to a few months ago (ρ = .59). Further, in more cohesive provinces, there are larger shares of respondents believing that their family’s lives will be better off than today in a year (ρ = .67) and that their children will have a bright future ahead (ρ = .68). The share of respondents who consider emigrating to another country was found lower in the more cohesive provinces (ρ = -.43).
The provinces' results indicate that social cohesion is conducive to a positive life evaluation, higher life satisfaction, and greater optimism. They are entirely in line with our findings from other continents: The higher the level of social cohesion in a geopolitical entity, the more positive people’s outlook on life (see Dragolov et al., 2016; Bertelsmann Stiftung, 2018; Arant et al., 2017; Boehnke et al., 2024).
6.2 Individuals
To investigate the association between social cohesion and subjective well-being on the individual level, we relate the individual responses to the well-being items to respondents’ membership in the four distinct classes of experiencing social cohesion. As most of the items on well-being are of categorical measurement quality, we apply the approach from Section 5 to describe the four classes based on respondents’ socio-demographic and socio-economic characteristics. Table 6.2 documents the results. Unlike the very weak relationships found in Section 5, the individual experience of social cohesion exhibits somewhat stronger, though still weak, associations with four of the indicators of subjective well-being.
Table 6.2.1 Subjective well-being in the four classes of respondents
The evidence presents a clear picture. The greatest share of respondents who evaluate their family’s life as better off than a year ago was found in Class 4, Cohesive Communities (29.5 %). The second largest share was found in Class 3, Middle South Africa (22.3 %). The lowest shares were found in Class 2, Integrated skeptics (13.3 %), and Class 1, Critics (12.7 %). The largest shares of respondents evaluating their family’s life as worse off than a year ago were found in Class 1 (43.8 %) and Class 2 (43 %).
Moving on to the two indicators of optimism, we find similar evidence. The largest share of respondents believed their family’s life would be better off than today in a year was found again in Class 4 (44 %). Class 3 appears somewhat reserved, given that most believed life would be about the same (43.2 %). Again, in Class 1 (43.6 %) and Class 2 (39.8 %), we find the largest shares of respondents believe their family’s life will be worse off than today. Optimism regarding children’s future was found highest in Class 4: 62.3 % believe children would have a bright future ahead of them. Pessimism prevails in the other classes, even in Class 3 (53.8 %), but yet again, we find the largest shares of respondents believe children would have a bleak future ahead of them in Class 1 (77 %) and Class 2 (69.2 %).
The above-described pattern holds for life satisfaction, too. The largest share of respondents evaluating their life satisfaction as improved was found in Class 4 (35.7 %), followed by Class 3 (20.8 %). The majority in both classes, though, report no change in life satisfaction: 50.8 % in Class 3 and 44.9 % in Class 4. Life satisfaction has worsened for the majority in Class 1 (50.4 %) and Class 2 (45.8 %).
Finally, the association between the experience of social cohesion and emigration is unclear or instead reversed. An overwhelming majority in each class (over 70 %) disagreed or strongly disagreed with the statement, “I am seriously considering emigrating to another country in the next year or so.” Interestingly, the disagreement rate is highest in Class 1 (81.6 %). Class 2 has the relatively highest rate of agreement with the statement (10.5 %), followed by Class 3 (9 %) and Class 4 (8.1 %), whereas only 7.4 % of Class 1 consider emigrating. High social cohesion does not suffice to discourage people from emigrating. The relationship is, however, very weak in terms of effect size.
7. Discussion and conclusion
We deliberately keep the discussion of our results relatively brief and mostly leave the conclusions to the political bodies working with the study results. South Africa has debated social cohesion ever since the end of the apartheid regime, especially in light of the Rainbow Nation concept. However, only two empirical studies have assessed South Africa's social cohesion during all these years. According to that research, South Africa scores middle on overall cohesion, lower on trust, and high on identification with the country.
The current study assesses social cohesion in South Africa, tracks its development, identifies structural factors that promote or impede it, explores social groups with high or low subjectively perceived cohesion, and examines its relationship to citizens' well-being. The Bertelsmann Social Cohesion Radar has been used as a model to quantify societal cohesiveness in many countries, including now South Africa. South Africa's social cohesion in 2023 (51.7 of 100 possible points) is similar to Germany's (52), although with different strengths and shortcomings in specific aspects.
The 2020–2023 Khayabus surveys provided by IPSOS South Africa also allowed us to examine cohesiveness in the nine South African provinces. Results demonstrated that social cohesion was predominantly determined by intact social networks, trust in institutions, perceived fairness, and respect for social rules fostering the common good. KwaZulu Natal had the lowest social cohesion (46.1), followed by Free State (49.4) and Gauteng (50.5). Limpopo (58.3) had greater social cohesion than the rest of the country.
As for trends across the 2020s, one can clearly state that South Africa's social cohesion is decreasing. However, the downward trend’s 'speed' is not devastating. A similar downward trend was also found for Germany between 2020 and 2022. The COVID pandemic is often seen as one of the reasons for dwindling social cohesion. In South Africa, high levels of identification with the country seem to be the decisive glue of society. Political attention must, however, be paid to the perception of fairness and respect for social rules, as those can spark turmoil. Trust in institutions and acceptance of diversity also need to be strengthened.
When one looks at people’s subjective experience with social cohesion, our study points to the core driver of deteriorating social cohesion: the urban-rural split. Cohesion is considerably higher in rurally dominated South Africa, while metropolitan areas are hotbeds of an intra-societal split.
Finally, several surprising results must be highlighted again. Previous studies elsewhere have always yielded that geopolitical entities that enjoy higher levels of per capita GDP also exhibit higher social cohesion, Scandinavia being the most prominent example. Findings for Asian countries corroborate that result. In South Africa, poorer regions of the country exhibit higher levels of cohesion. The study furthermore found conflicting results on South African poverty and inequality. Social cohesion is lower in provinces where more households view themselves as poor but higher in those where more people objectively live in poverty. A negative link with median age (the younger people are in an area, the higher social cohesion) and a positive correlation with ethnic and linguistic fractionalization were also discovered, which appear counterintuitive at first glance.
However, what emerges from the South African Social Cohesion Index (SASCI) study as it did for essentially all other studies that set out to measure the effects of social cohesion on people’s well-being? Social cohesion is ‘good’ for people, or in more technical terms, the higher people perceive social cohesion to be, the more positive their subjective well-being.
There are certainly also shortcomings to the study presented here—as there probably are to all empirical studies: The assessment of the level of social cohesion had to be based on very few questions from the Khayabus studies. Before the SASCI can be fielded as a regular endeavour in providing data for the further development of South Africa’s social cohesion, it may, after all, be advisable to conduct one more extensive representative study devoted exclusively to the topic of social cohesion and its economic (pre-)conditions as has been the case in Germany with the series of Bertelsmann studies.
References
Abrahams, C. (2016). Twenty years of social cohesion and nation-building in South Africa. Journal of Southern African Studies, 42(1), 95–107. http://www.jstor.org/stable/43900558
Alesina, A., Devleeschauwer, A., Easterly, W., Kurlat, S., & Wacziarg, R. (2003). Fractionalization. Journal of Economic Growth 8, 155–194. https://doi.org/10.1023/A:1024471506938
Apraku, A., Moyo, P., & Akpan, W. (2018). Coping with climate change in Africa: an analysis of local interpretations in Eastern Cape, SA. Development Southern Africa, 36(3), 295–308. https://doi.org/10.1080/0376835X.2018.1482199
Arant, R., Dragolov, G., & Boehnke, K. (2017). Sozialer Zusammenhalt in Deutschland 2017. Gütersloh: Bertelsmann Stiftung.
Arant, R., Larsen, M., & Boehnke, K. (2016). Sozialer Zusammenhalt in Bremen. Gütersloh: Bertelsmann Stiftung.
Ballantine, C., Chapman, M., Erwin, K., & Maré, G. (Eds.) (2017). Living together, living apart? Social cohesion in a future South Africa. University of Kwa-Zulu Natal Press. http://www.ukznpress.co.za/?class=bb_ukzn_books&method=view_books&global%5Bfields%5D%5B_id%5D=496
Bertelsmann Stiftung (Ed.). (2018). What holds Asian societies together: Insights from the Social Cohesion Radar. Gütersloh: Bertelsmann-Stiftung.
Boehnke, K., Berrueto, A., Dragolov, G., & Ocampo Villegas, P. (2019). Are value preferences and social cohesion interconnected? The case of Mexico. Acta de Investigación, 9(2). http://dx.doi.org/10.22201/fpsi.20074719e.2019.2.262
Boehnke, K., Dragolov, G., Arant, R., & Unzicker, K. (2024). Gesellschaftlicher Zusammenhalt in Deutschland 2023. Gütersloh: Bertelsmann Stiftung.
Burns, J., Hull, G., Lefko-Everett K., & Njozela, L. (2018). Defining social cohesion (SALDRU Working Paper 216). Cape Town: SALDRU, UCT.
Center for Risk Analysis. (2023). Assets and Incomes. November 2023. Johannesburg: Center for Risk Analysis.
Delhey, J. & Boehnke, K. (2018). Conceptualizing social cohesion in Asia, In Bertelsmann Stiftung (Ed.), What holds Asian societies together: Insights from the Social Cohesion Radar (pp. 29-48). Gütersloh: Bertelsmann-Stiftung.
Delhey, J., Dragolov, G. & Boehnke, K. (2023). Social Cohesion in International Comparison: A Review of Key Measures and Findings. Kölner Zeitschrift für Soziologie und Sozialpsychologie, 75, 95-120. https://doi.org/10.1007/s11577-023-00891-6
Dragolov, G., Ignácz, Z. S., Lorenz, J., Delhey, J., Boehnke, K., & Unzicker, K. (2016). Social cohesion in the Western world. What holds societies together: Insights from the Social Cohesion Radar. Springer International. https://doi.org/10.1007/978-3-319-32464-7
Global Data Lab (2024). Subnational HDI (v8.0). Retrieved on September 24, 2024 from https://globaldatalab.org/shdi/table/shdi/ZAF/
Langer, A., Stewart, F., Smedts, K., & Demarest, L. (2017). Conceptualising and measuring social cohe-sion in Africa: Towards a perceptions-based index. Social Indicators Research, 131, 321-343. https://doi.org/10.1007/s11205-016-1250-4
Larsen, M. M., & Boehnke, K. (2016). Measuring Social Cohesion in the Kyrgyz Republic. Social Cohesion Index. University of Central Asia's Institute of Public Policy and Administration Working Paper No. 36, Bishkek.
Leininger, J., Burchi, F., Fiedler, C., Mross, K., Nowack, D., von Schiller, A., Sommer, C., Strupat, C., & Ziaja, S. (2021). Social cohesion: A new definition and a proposal for its measurement in Africa (Discussion Paper 31/2021). Bonn: Deutsches Institut für Entwicklungspolitik (DIE). https://doi.org/10.23661/dp31.2021.v1.1
OECD (2021). Old-age income inequality. In Pensions at a Glance 2021: OECD and G20 Indicators. OECD Publishing, Paris. doi: https://doi.org/10.1787/d1a5a309-en
Schiefer, D. & van der Noll, J. (2017). The essentials of social cohesion: A literature review. Social Indicators Research, 132(2), 579-603. https://doi.org/10.1007/s11205-016-1314-5
Statistics South Africa (2023a). Census 2022. Pretoria: Stats SA.
Statistics South Africa (2023b). Quarterly Labour Force Survey. Quarter 4: 2022. Pretoria: Stats SA.
Statistics South Africa (2024a). Estimation of regional gross domestic product for South Africa: Experimental estimates. Pretoria: Stats SA.
Statistics South Africa (2024b). Population Estimates. Pretoria: Stats SA
Statistics South Africa (2024c). Subjective poverty in South Africa. Findings from the General Household Surveys 2019 and 2022. Pretoria: Stats SA.
Wasserstein, R. L., Schirm, A. L., & Lazar, N. A. (2019). Moving to a World Beyond “p < 0.05.” The American Statistician, 73(sup1), 1–19. https://doi.org/10.1080/00031305.2019.1583913
Weller, B. E., Bowen, N. K., & Faubert, S. J. (2020). Latent Class Analysis: A guide to best practice. Journal of Black Psychology, 46(4), 287-311. https://doi.org/10.1177/0095798420930932
[1] In line with the customary practice in economic research and our previous studies, we transform the raw values by taking their natural logarithm (ln).
[2] Individuals below the food poverty line cannot afford enough food to obtain the minimum daily energy requirement for adequate health. Individuals below the lower-bound poverty line are unable to afford both adequate food and non-food items and have to sacrifice food for essential non-food items. Individuals below the upper-bound poverty line can afford adequate food and essential non-food items. In 2022, the food poverty line was at 663 Rand, the lower-bound poverty line at 945 Rand, and the upper-bound poverty line at 1417 Rand, according to the report of the Center for Risk Analysis (CRA, 2023).
[3] The Gini index measures income inequality in the population as a whole. It ranges from 0 (perfect equality among all individuals) to 1 (perfect inequality, where one individual has all income). The P90/P10 ratio contrasts the income at the 90th percentile of the income distribution to the income at its 10th percentile (OECD, 2021).
[4] Fractionalization is the probability that two randomly selected individuals are not from the same group (ethnic, linguistic, religious, etc.; Alesina et al., 2003). The corresponding indices for ethnic/linguistic/religious fractionalization range from 0 (all individuals are from the same ethnic group/speak the same language/are from the same religious group) to 1 (each individual belongs to a separate ethnic/linguistic/religious group).
[5] Income classes were derived from the reported total household income. The latter was equivalized concerning household size using the modified OECD equivalence scale. Respondents with equivalized household income lower than 60 % of the median belong to the low-income class, from 60 % to 100 % - to the lower-middle income class, from 100 % to 200 % - to the upper-middle income class, greater than 200 % - to the high-income class. Respondents with a missing value on household income are treated as a separate group due to the large share of non-response (41.5 %).
Appendices
Appendix A: Indicators of cohesion across time
This Appendix documents the population-weighted relative frequencies of the response categories of the indicators used to calculate the social cohesion scores in this report. The reported values pertain to the respective percentage distributions in the total sample for each year of data collection.
Table 0.1 Indicators of Dimension 1.1 “Social networks” across time
Table 0.2 Indicators of Dimension 1.2 “Trust in people” across time
Table 0.3 Indicators of Dimension 1.3 “Acceptance of diversity” across time
Table 0.4 Indicators of Dimension 2.1 “Identification” across time
Table 0.5 Indicators of Dimension 2.2 “Trust in institutions” across time
Table 0.6 Indicators of Dimension 2.3 “Perception of fairness” across time
Table 0.7 Indicators of Dimension 3.2 “Solidarity and helpfulness” across time
Table 0.8 Indicators of Dimensions 3.2 “Respect for social rules” across time
Table 0.9 Indicators of Dimension 3.3 “Civic participation” across time
Appendix B: Dimensions of cohesion in the provinces over time
This Appendix contains results on the level and change over time in the nine single dimensions of social cohesion across the provinces of South Africa.
Tables A.10 to A.18 reveal that Gauteng is the province with the weakest level of social cohesion in Domain 1, “Social Relations.” Gauteng ranks last on Dimensions 1.1, ‘Social Networks,’ 1.2, ‘Trust in People,’ and 1.3, ‘Acceptance of Diversity.’ KwaZulu Natal ranks last on five of the remaining six dimensions. Only for Dimension 3.3, ‘Civic Participation,’ Northern Cape ranks last, likely because of lacking infrastructure in this least densely populated province.
Domain “Social Relations”
Table 0.10 Dimension 1.1 “Social networks” in the provinces across time
Table 0.11 Dimension 1.2 “Trust in people” in the provinces across time
Table 0.12 Dimension 1.3 “Acceptance of diversity” in the provinces across time
Domain “Connectedness”
Table 0.13 Dimension 2.1 “Identification” in the provinces across time
Table 0.14 Dimension 2.2 “Trust in institutions” in the provinces across time
Table 0.15 Dimension 2.3 “Perception of fairness” in the provinces across time
Domain “Focus on the Common Good”
Table 0.16 Dimension 3.1 “Solidarity and helpfulness” in the provinces across time
Table 0.17 Dimension 3.2 “Respect for social rules” in the provinces across time
Table 0.18 Dimension 3.3 “Civic participation“ in the provinces across time
Appendix C: Correlations of social cohesion on the province level
Table 0.19 Structural characteristics and social cohesion in South African provinces
Table 0.20 Social cohesion and subjective well-being in the provinces
Appendix D: Latent class analyses
This Appendix documents goodness-of-fit indices for the various LCA models specified.
Table 0.21 Goodness-of-fit indices of LCA solutions
Table 0.22 Relative class sizes for LCA solutions
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
This report has been published by the Inclusive Society Institute
The Inclusive Society Institute (ISI) is an autonomous and independent institution that functions independently from any other entity. It is founded for the purpose of supporting and further deepening multi-party democracy. The ISI’s work is motivated by its desire to achieve non-racialism, non-sexism, social justice and cohesion, economic development and equality in South Africa, through a value system that embodies the social and national democratic principles associated with a developmental state. It recognises that a well-functioning democracy requires well-functioning political formations that are suitably equipped and capacitated. It further acknowledges that South Africa is inextricably linked to the ever transforming and interdependent global world, which necessitates international and multilateral cooperation. As such, the ISI also seeks to achieve its ideals at a global level through cooperation with like-minded parties and organs of civil society who share its basic values. In South Africa, ISI’s ideological positioning is aligned with that of the current ruling party and others in broader society with similar ideals.
Email: info@inclusivesociety.org.za
Phone: +27 (0) 21 201 1589
Comments