The Lifelong Consequences of Protesting: A Longitudinal Analysis of the Gendered and Intergenerational Effects of Protest Participation on Individuals’ Life-Course Patterns

(1)

The Lifelong Consequences of Protesting

A Longitudinal Analysis of the Gendered and Intergenerational Effects of Protest Participation on Individuals’ Life-Course Patterns

Hannah Gisgård

Uppsala University, Autumn 2020 Department of Government Master’s Thesis

Supervisor: Sven Oskarsson Words: 15014

Pages: 39

(2)

Abstract

Protesting is a common tactic used by social movements and the outcomes are widely researched in social movement studies. This thesis examines the biographical consequences of protest participation on individuals’ life-course patterns from a gendered and intergenerational perspective. The study employs regression analysis and a longitudinal dataset collected from the Swedish Level of Living Survey, which includes six panel waves in total stretching from 1968 to 2010. It consists of a nationally representative sample of the Swedish population between the ages of 15–75, in which the last two panel waves include the respondents’ children in the ages of 10–

18. The results show that protest participants are likely to become more educated than non- participants and that they continue to remain active in political and union activities. Further, there is evidence of gender-based differences between protesters as women do not continue to uphold the same level of involvement in political organisations in comparison to men. No support is given to the expectation that protesting will have intergenerational effects. The results show that participating in protests may have long-lasting consequences for individuals and that there seems to be gender-based differences between protesters, which might have implications for individuals’

further involvement in political activities.

Keywords: social movements, political participation, protesting, biographical consequences, gender, intergenerational effects

(3)

All over the world, people come together to start campaigns, arrange public meetings, and to take to the streets to voice their opinions. In particular, protesting has become a common tactic used by social movements to influence policy decisions, as it is argued to constitute an important gateway for people’s participation in public politics (Quintelier and van Deth 2014; Snow et al.

2019; Tilly 2004; Verba et al. 1995). While the activity of protesting can be traced back to early revolutions, the protests in 1968 marked a turning point for the civil rights movement in the United States. These activities set the path for the following anti-war movement, the May 1968 protests in France, as well as the occupation of the Student Union Building in Sweden (Moore 1999; Peterson et al. 2018; Tarrow 1993). Lately, the world has witnessed large-scale protests such as the democracy movement in Hong Kong and the global Black Lives Matter Movement.

On an individual level, most people have come across the opportunity to attend a demonstration.

We might reflect on the short-term consequences of it, asking ourselves whether we have the time and the resources to participate. However, we rarely ponder over how attending a demonstration once might affect the rest of our lives a long time after having protested. This begs the question of what the long-term consequences of protesting might be. Previous literature has explored the consequences of protest participation and suggest that, not only does it affect the movement in itself, but also the participants’ family-life, education and career choices, and further involvement in political activities (Fendrich and Tarleau 1973; Giugni and Grasso 2016; McAdam 1989;

Sherkat and Blocker 1997). Moreover, it is likely that we will find gender-based differences in protesters’ life-course patterns (Cherniss 1972; Van Dyke et al. 2000), and that parents who protest are likely to transmit their behaviours to their children’s political lives (Campbell 2006;

Stoker and Jennings 1995; Walsh et al. 2004).

This paper argues that a common shortcoming in most of previous studies is the lack of controlling for socioeconomic background, as this is likely to affect people’s inclination to protest and to make certain life-choices to begin with. Drawing on a life-course perspective, the aim is to contribute to existing literature on the effects of protest participation on individuals’ political and lifestyle trajectories by further including the elements of gender and intergenerational transmission. Hence, the research question of this study is: how does protest participation affect individuals’ life-course patterns from a gendered and intergenerational perspective?

(5)

2 The relationship between protest participation and life-course patterns is studied with the method of regression analysis. The study employs longitudinal data from the Swedish Level of Living Survey, including six panel waves in total stretching from 1968 to 2010. A theoretical framework built on theories of protest participation and the gendered and intergenerational consequences is employed, which results in three hypotheses on suggested life-course outcomes, three hypotheses on the gendered effects, and a final hypothesis regarding the intergenerational effects. The first regression model attempts to recreate previous models and shows that protest participants are likely to become more educated than non-participants and that they continue to be involved in political and union activities, but finds no support for effects on family-life outcomes. The second regression model seeks to improve previous models and finds that the results in the recreated model are robust to the inclusion of controls for socioeconomic background. The third regression model tests the interaction effect of being a woman and a protester and shows that women participants do not attain the same high-level positions in political organisations as men participants. The fourth and last regression model explores the intergenerational effects of protest participation but finds no support for the claim that parent protest participants will influence their children to become more politically active. All of the analyses lend support to the hypotheses saying that protest participants are likely to become more educated than non-participants.

Moreover, the results give evidence to the assumption that the level of extended involvement in political activities is likely to differ between men and women protesters.

The study begins with a review of the relevant literature, which is concluded with the research question of the paper. This is followed by the assumptions and the hypotheses that are to be tested. The next section presents the research design, which includes the choice of method, a review of data selection, operationalisation of the variables, as well as the limitations of the study.

Next, along with the descriptive statistics, the results of the regression analyses are presented, analysed and discussed. Lastly, the conclusions are presented with a few final remarks and suggested improvements for future research on this topic.

(6)

3 2. Theoretical Framework

2.1. Literature Review

Key Concepts

Prior to presenting previous research, it is essential to outline the key concepts of this study.

Political participation can broadly be defined as the activities that citizens choose to become part of in order to be able to influence political decisions (Brady 1999; van Deth 2001). In addition, Verba et al. (1995, 38) argues that the aim of participating in these activities can be described as wanting to have an influence on government action, “either directly by effecting the making or implementation of public policy or indirectly by influencing the selection of people who make those policies”. In other words, to influence political decisions, people may voice their opinions through more official forms, such as voting in elections. Prior to the 1960s, political participation was mostly explained in terms of voting and campaigning. However, the emergence of social movement activities during the late 1960s and early 1970s forced a shift in the understanding of the concept of political participation. When people began organising demonstrations and arranging riots in Europe and the United States, scholars had to reassess the concept of political participation, i.e. to include other political activities than the formal ways of participating (Quintelier and van Deth 2014). Since then, taking to the streets to organise demonstrations have become a common tactic used by participants of social movements.¹

In order to explain the concept of social movements, this paper draws on the commonly used definition provided by Della Porta and Diani (1999), who argue that a social movement must consist of several mechanisms through which actors engage in collective action: first, they must be involved in conflictual relations with clearly identified opponents; second, they have to be linked by dense informal networks; third, they must share a distinct collective identity; and fourth, the goal should be to promote or oppose social change. A social movement may therefore be defined by the opinions and beliefs shared by its participants, and the interactions that these informal networks may lead to. Thus, it can involve either individuals or organisations and should strive to achieve a social or political goal. In order to reach these goals, the movement will therefore employ different tactics. As suggested by Tilly (2004), participating in social movements is an important gateway for people’s participation in public politics. The participants of a movement will express their opinions and concerns about issues such as human rights, welfare,

1 The concepts of protest/protesting and demonstration/demonstrating will be used interchangeably throughout this paper.

(7)

4 and the well-being of society overall. This is accomplished by using various repertoires, i.e. tactics, such as arranging protests, public meetings or petitions (Snow et al. 2019). Attending protests may especially be argued to constitute an accessible entry for citizens to participate politically. For this study, the main interest is therefore in the activity of protest participation and what effects it may have on individuals’ life-course patterns. Van Stekelenburg et al. (2019) further propose that participating in street demonstrations specifically should be regarded as the most typical protest activity that citizens in contemporary Western societies take part in. According to the authors, a street demonstration can be regarded as either ritualised, peaceful, or violent, and will usually be arranged by a coalition of organisers, such as a small political party, or a large social organisation.

The “coalition of the crowd” may vary as well, meaning that the people participating in the demonstration do not have to be part of the same establishment, or organisation, outside of said activity. Instead, what is of importance is that they share the same beliefs and opinions, which made them join the demonstration in the first place.

To summarise, people participate in political activities to influence political decisions (Brady 1999; van Deth 2001) and because they want to influence government action either directly or indirectly (Verba et al. 1995). During the last couple of decades, scholars have started to include more contentious forms of participation, such as protests and riots. Attending demonstrations is an important gateway for citizen’s political participation, and even though protest participation might be considered contentious in some respects, it may still be regarded as one of the more usual form of social movement activities (Tilly 2004; van Stekelenburg et al. 2019).

The Consequences of Participating

While most research on the outcomes and consequences of social movements has focused on the extent to which movements may influence policy outcomes, less attention has been paid to other consequences. Giugni (2013) points out three different types of outcomes, namely the political, cultural and biographical consequences. Political consequences are the effects that alter the movement’s political environment, or the effects it may have on legislative issues such as policy outcomes. Cultural consequences are simply the impact that mobilising activities may have on the movement’s cultural environment. Lastly, the biographical consequences can be regarded as the effects on the participants’ life-courses, which should at least partly be due to previous involvement in these types of activities. For instance, the biographical consequences of participating in social and political movements may refer to the way in which these consequences bring about further political participation, as well as the political socialisation aspects of it. While

(8)

5 political and cultural outcomes are important concepts to take into account when analysing the consequences of social movement activities, the aim of this study forces the focus to the personal and biographical consequences. Henceforth, the biographical consequences therefore refer to the political and personal lives of the participants.

Previous research has explored the personal effects of both strongly committed movement participants, as well as not-so-committed participants, that is, people who attend protests from time to time. Looking at the American context, McAdam (1989) studies the short- and long-term personal and political consequences of high-risk activism among members of the New Left. By examining the participants’ occupational and marital histories to assess whether participation has any effects on the subsequent lives of the participants, it was found that participants continued to voice and act on political values as well as remained active in movement activities. Moreover, because the New Left was argued to have become “the organising principle of their lives” (ibid, 758), their leftist attitudes were subsequently found to persist a long time after having participated.

On the other hand, a common belief in social science is that people tend to become more conservative the older they get, as their attitudes and values tend to become less affected by changes. As described by Glenn (1974, 176), if “each subsequent experience is a smaller proportion of the total background experiences”, it would mean that people’s attitudes and values are set on a determined path which should continue to be unaffected by any external interferences. To the contrary, protest participation may indeed have durable consequences, at least for the perseverance of leftist values. Scholars have found that activists in the New Left contention continued to maintain values of egalitarianism, libertarianism and support for diversity, as compared to non-activists (Whalen and Flacks 1984; Marwell, Aiken and Demerath 1987; McAdam 1989). Furthermore, former student civil rights continued to identify as more radical and liberal in their political orientation after having participated in mobilising activities (Fendrich and Tarleau 1973). Likewise, Giugni and Grasso (2016) show that Swiss participants’

lives are changed through earlier involvement in protests. In line with the findings of previous studies, the participants were found to espouse political values more leaning to the left, and that they became more engaged in institutional forms of participation.

Participating in activist activities thus seems to have an impact on the participant’s mind as well as their worldviews. However, it may be argued that those individuals who join organisations or participate in protests already hold such worldviews, meaning that it cannot be determined whether it was the activity in itself that changed the participants, or if they had already attained

(9)

6 such values prior to participating. Munson (2009) shows that in the case of pro-life activists and non-activists, people’s view on abortion changed once they joined pro-life groups. Passy and Monsch (2018) explore the Swiss context and examine if previous commitment may have an impact on participants’ social and political worldviews. While they discover that people who engage in various movement activities already possess values that are in symbiosis with said movement, they also find that a great part of those who join had other worldviews before joining;

values that are likely to have been affected and changed by movement participation.

Previous research present strong evidence to the claim that people who participate in protests do differ from those who do not participate, which may not be very surprising in regard to their political life. Individuals who participated in the New Left activities in the United States remained active and continued to espouse leftist as well as liberal values, as opposed to their non-participant counterparts (McAdam 1989; Whalen and Flacks 1984; Marwell, Aiken and Demerath 1987;

Fendrich and Tarleau 1973). The issue here may however be that the type of people who decide to attend a protest perhaps already had the resources and values that would make them participate to begin with. If these factors are considered constant, as proposed by Glenn (1974), it would mean that one’s values rarely change due to the activities one may take part in. Hence, it could be questioned why examining differences in life-course outcomes between participants and non- participants should make a contribution to the current literature. The counterargument may however be that attending these kinds of movement activities might actually change one’s values and attitudes (Munson 2009; Passy and Monsch 2018). If that is the case, participation in protests can be expected to have an effect on people’s political attitudes and behaviours.

Furthermore, individuals’ educational, employment, and marriage choices have been discovered to have become altered by their previous involvement in social and political activities. Sherkat and Blocker (1997) surveyed activists who participated in the anti-war, student and civil rights protests in the late 1960s and early 1970s to examine how activists differed from non-activists in regard to the individuals’ political and personal lives. Their findings show that protesters and non- protesters differed in terms of educational and occupational choices, in the sense that former protesters on average attained a higher level of education. In addition, activists were also found to enter the job market later and to change jobs more frequently compared to non-activists. In addition, previous activists are likely to earn less due to the work choices they tend to make, despite being more educated (McAdam 1989). Research suggests that activists and non-activists seem to differ in terms of type of occupational choices as individuals who previously in their lives

(10)

7 had participated in activist activities are more likely to be found in academic professions, social works, or creative jobs (Fendrich and Tarleau 1973; McAdam 1989; Sherkat and Blocker 1997).

The socialisation aspects of protest participation may therefore inspire people to make different choices regarding their education or career. Those who refrain from participating – which could be due to various reasons not addressed here – should consequently make other choices as a result of not having shared the same experiences.

The consequences of activism may also reach beyond political values and worldviews. Scholars have found that participation in movement activities can have an effect on people’s family life- spheres. For instance, Sherkat and Blocker (1997) show that former protesters tend to marry later in life and have fewer children. That previous participants’ marital status is affected by their involvement is further supported by McAdam (1989), who found that activists were less likely to be married in comparison to their non-activist counterparts, meaning that they either never got married, or their marriages failed. In this instance, if the marriages were built on shared assumptions of the world, any changes in worldviews due to movement activities might cause some sort of marital crisis, and consequently, divorce. Sherkat and Blocker (1997, 1066) further suggest that activists during the 1960s may have seen the “traditional family relations and childrearing orientations as supportive of the patriarchal culture they were fighting against”.

Hence, the argument that participating in such protests can alter one’s values in terms of domestic principles, and consequently choices about family and marriage should prove to be valid if we are to believe that it is the protest activity itself that is the main cause.

Thus, in terms of education and occupation, former protest participants may be more likely to obtain a higher level of education and continue to work in academic professions as well as creative and social jobs. Despite being more educated, former participants seem to earn less as compared to their counterparts. Protest participation may also affect their family life, in the sense that they are less likely to marry, and if they do, they marry later in life. In addition, participants are likely to have fewer children (Fendrich and Tarleau 1973; Sherkat and Blocker 1997; McAdam 1989).

Here it may be argued that all of these life-choices are affected by the values they attain from protest participation, or had before, which consequently can bring about different worldviews and values regarding the domestic sphere.

(11)

8 Gendered Effects

These studies do however seem to ignore an important aspect, as choices about family, education, occupation by all means could be considered a gendered issue. In terms of social movement participation, scholars have argued that gender may mediate the biographical effects of participating in protests. For instance, commitment among women have been shown to put constraints on their relationships, at least partly due to their involvement in social movement activities (Cherniss 1972). Furthermore, Van Dyke et al. (2000) propose an explanation as to why men and women might experience different consequences due to previous activism. While activists overall tend to deviate from the traditional paths in regard to marriage and employment choices, they argue that these differences will be more visible among women activists. This is due to the prevalence of gender-based roles, and the expectations that follow, which could result in diverging experiences between men and women activists. If the participants’ experiences are found to be unequal in the political arena, it is likely that these differences will spill over to the personal and domestic sphere, or vice versa. Furthermore, the authors show that the biographical impacts are not only confined to high-risk participation. Instead, low-risk, run-of-the-mill types of activities may also have a significant impact on the participants’ life-course patterns.

While gender should be considered an important factor when examining differences in activist experiences, the authors argue for the need of studying the way in which various aspects might intervene with the activist experience and the consequences that follow (ibid). It can thus be expected that women and men who participate in protest activities will differ from each other in terms of biographical outcomes. If gender differences between men and women participants correlate with the gender differences between men and women non-participants, it could be argued that any differences in outcomes are not due to the participation activity in itself, but to other underlying factors. On the other hand, if the life-course patterns of women participants differ from the life-course patterns of women non-participants, it could instead be suggested that the activity of participating perhaps altered the life of that person in some sense, which would therefore imply that participating in protests could have a gendered effect on biographical outcomes.

Intergenerational Outcomes

Assuming that protesting has enduring consequences on individuals’ life-course patterns, what ultimately begs the question is the impact this may have on the participants’ children. According to earlier literature, protest participation seems to have effects on the participants’ family life in

(12)

9 terms of when and how many children a person has, and thus, parents who are politically committed should therefore have some sort of influence on their children. Socialisation theory scholars suggest that in order for attitudes and behaviours to transfer across generations early childhood experiences and social learning aspects are of fundamental value, since children learn from their parents’ actual behaviours and attitudes. Because children may perceive their parents as role models, the opinions or ideas the parents express should have influence their children (Bandura 1982). To understand political socialisation in terms of parental influence, Beck and Jennings (1975, 83) describe it as “linking certain parental characteristics to certain observed outcomes in the offspring.” In other words, if a parent carries certain political traits, such as having more liberal values, political socialisation suggests that these liberal values should be visible in their children’s political attitudes later in life. It should however be mentioned that this process is not simply a two-generation phenomenon. Because the parents were once socialised by their parents, it should instead be understood as a process that takes place over several generations.

Regardless of the fact that this paper only has access to two generations, it should still be of importance to examine the intergenerational transmission of political attitudes and behaviours between parents and their children.

Previous studies have focused on the extent to which political ideology is transmissible from parent to child (Percheron and Jennings 1981; Westholm and Niemi 1992). However, individuals’ political characteristics may go beyond ideology and party affiliation – how one behaves politically may be transmissible too. The parental influence of political attitudes and behaviour remains relatively unexplored. For instance, scholars have found that the civic norms that are acquired at an early age can have effects on subsequent voting rates, that social class affects political participation, and that political attitudes and behaviour may transmit through marriage (Campbell 2006; Stoker and Jennings 1995; Walsh et al. 2004). In the case of mothers’ influence on their daughters’ gender attitudes, Moen et al. (1997) find support for the importance of childhood socialisation processes as well as the children’s own life-course experiences. In other words, parents do have an influence on children’s behaviours and attitudes in some senses, but it could also be the case that the children’s own experiences are equally as important.

Consequently, intergenerational effects in regard to political participation should be visible in this case. It is a long process; the parents were once raised by their own parents who might have come from different contexts and with various socioeconomic backgrounds, which may have affected their values. These values are then believed to have imparted on the soon-to-be participants (and

(13)

10 non-participants). It is then assumed that the experience of protesting had an impact on the individuals taking part in protests which influenced them to espouse certain values and to make different life choices in terms of education, occupation and their family-life sphere. And lastly, the participants’ children are consequently assumed to acquire these political attitudes and behaviours. If we were to find evidence of intergenerational transmission of such values, it should arguably prove to be useful for the understanding of political participation in general.

Recognising the Consequences of Protest Participation

As this study takes place in the domain of political science, it could be called into question why we would want to pay attention to possible changes in an individual’s life. Passy and Monsch (2019) argue that the biographical outcomes of social movement participation should be considered an important field of study for several reasons. For example, participation in mobilising activities may bring about socialisation processes, since participating in a protest may alter one’s worldview and behaviour. Moreover, such activities may have spill-over effects in the sense that they may pass on to other social spheres. That is, how one goes about making political choices will spill over into other life choices, such as one’s occupation, family, or religion (Sherkat and Blocker 1997). Consequently, biographical consequences on an individual level may have aggregate-level effects. Participants are likely to interact with non-participants, which in turn can influence the non-participants attitudes and behaviours. From a wider perspective, these activities and processes might have effects on society at large.

In regard to the impact of real-world events, Peterson et al. (2018) distinguish four protest waves which are visible in a global as well as a Swedish context. The first wave relates to the post-war peace protests and youth revolts of the 1960s when new social movement activities began to occur, in which people began taking to the streets in a way that the government could not ignore. The second wave appeared during the late 1970s and focused on women’s rights and environmental issues. These protests were followed by the third wave of the anti-racism, anti-austerity and global justice protests in the 1990s. The final and fourth wave has taken place since 2010 and focuses on climate issues and anti-racism. Since most of previous research on the biographical consequences were conducted during the peak years during the 60s and 70s, it may be argued that it could make it more difficult to assess whether the impact was a result of their participation or non-participation, or if it the explanations that were found might simply be contextually bound.

It may therefore be argued to be of fundamental value to look beyond the commonly exploited contexts to be able to assess the effect of protest participation on individuals’ life-course patterns.

(14)

11 2.2. Hypotheses

The hypotheses of this study will be divided into four categories: the first category concerns the domestic sphere; the second category aims to test occupation and educational choices; the third category relates to involvement in political organisations and trade unions; and the fourth and final category explores the intergenerational effects of protest participation. Hypotheses regarding the gender differences are added to the first three categories. Beginning with the first category, participating in protests may lead to various socialisation processes in which people’s values and worldviews change. Therefore, it is expected that protest participants are less likely to be married later in life and that they will have fewer children compared to non-participants (Fendrich and Tarleau 1973; McAdam 1989; Sherkat and Blocker 1997). Consequently, the first hypothesis reads as follows:

Hypothesis 1a. Protest participants will have fewer children and are less likely to be married.

Regarding education and work-life, scholars have found that individuals who decide to participate in protests differ from those who refrain in terms of educational choices, in the sense that protest participants have been shown to attain a higher level of education. Moreover, despite the finding that they study for a longer time compared to their non-participant counterparts, they seem to earn less – which may be a result of the fact that protest participants are more likely to be employed within creative and social works (ibid). Therefore, the second hypothesis is as follows:

Hypothesis 2a. Protest participants will attain higher levels of education and will at the same time have lower incomes.

The consistent argument in previous research is that protesting indeed affects whether and how individuals continue to participate in political activities. If protest participants continue to espouse certain values (McAdam 1989; Whalen and Flacks 1984; Marwell, Aiken and Demerath 1987;

Fendrich and Tarleau 1973), it may be believed that they differ from non-participants, and thus, they will continue to participate politically to express these opinions in other forms besides protesting (Giugni and Grasso 2016). In this case, trade union activity should be accounted for, since it is likely that participation in these activities will indicate a similar type of engagement as

(15)

12 participation in party organisations. Therefore, protest participants are expected to be more involved in extended political and union activities, compared to non-participants:

Hypothesis 3a. Protest participants will show higher levels of extended involvement in political organisations and trade unions.

More importantly, Cherniss (1972) and Van Dyke et al. (2000) have argued that the life-courses of men and women participants will differ from each other as a result of previous involvement in protest activities. Because participation may put constraints on one’s relationship, it is likely that this will affect women negatively due to gendered roles and the expectations that may follow.

Therefore, if we expect that having participated in protests will affect the individuals’ family life, it is likely that such differences will spill over to one’s work-life and educational choices as well because of the prevalence of structural gender roles. And because protesting has been proposed to be more popular among women, it is suspected that the predominance of women participants will be visible in other political activities. Hence, the hypotheses concerning gender-based differences are as follows:

Hypothesis 1b. The effect of protest participation on family-life choices will be stronger for women protesters.

Hypothesis 2b. The effect of protest participation on education and income will be stronger for women protesters.

Hypothesis 3b. The effect of protest participation on extended involvement in political organisations and trade unions will be stronger for women protesters.

Lastly, this thesis aims to explore whether protest participation has intergenerational effects. If we expect that parents’ political behaviours will transmit onto their children, it can be assumed that parents who have previously participated in protests will differ from parents who have not protested. Relying on insights from studies on political socialisation processes (Bandura 1982;

Beck and Jennings 1975; Campbell 2006; Percheron and Jennings 1981; Stoker and Jennings 1995; Walsh et al. 2004; Westholm and Niemi 1992), if the behaviour that separates participant from non-participant parents can be passed onto their offspring, there should be a visible effect in their children’s political behaviour. That is, if the parent of a child once participated in a

(16)

13 protest, there should be a visible effect in the political behaviour of the child, which is contingent on the assumption that protest participation has biographical consequences in the first place. For instance, this could be manifested through whether the children of protest participants ever participated in protests themselves, if they have expressed their opinions by for example wearing political symbols, or if they have written letters to the editor expressing their opinions on various matters. Such behaviour may therefore be regarded as being politically active. Therefore, the final hypothesis regarding the intergenerational effects is as follows:

Hypothesis 4. Children with participant parents will be more politically active than children with non-participant parents.

2.3. Case Selection

More contentious forms of political participation, such as protesting, have mostly been attributed to other contexts. Participating in protests have however become a more common tactic used by movements in Sweden since the late 1960s, despite the argument that the country’s inhabitants are of the more collaborative kind with little involvement in conflicts (Uba 2016). Sweden do seem to be a fairly protest-experienced country, since 27 percent of the population stated to have taken part in at least one demonstration in 2000, and 14 percent of the population in 1968 (SOFI 2016). Even though a relatively large proportion of the population has taken part in demonstrations and protests just once, not as many do it on a regular basis. Nonetheless, because many Swedes seem to be open-minded to taking part in demonstrations, Wennerhag (2017) suggests that demonstrating in Sweden may be considered a more routine, run-of-the-mill, form of participating. It is argued that this may be due to the prevalence of more “ritualised” forms of protests, such as the May Day demonstrations and the Pride parades of the LGBT movement.

Moreover, being active in trade unions is common in Sweden as they have been argued to be subject to organising demonstrations – and the reason for people’s decision to protest in the first place (Peterson et al. 2011). Another important characteristic may be the prevalence of “standby citizens”, which Amnå and Ekman (2012) suggest is much more common in the Scandinavian countries. The term refers to those citizens who are interested in politics and are willing and have the resources to participate if necessary. Swedish citizens in general may not participate in protests on a daily basis, but when it is called for, most people will show up. Protesting in Sweden can thus be argued to be fairly accessible for most people as participating may not involve such a large cost for the participants. Assuming that Swedish citizens generally are rather open-minded to

(17)

14 participating in protests, it should be of interest to explore how participants differ from non- participants and to scrutinise any possible differences in life-course patterns.

As shown in the theory section, previous studies on the biographical consequences of social movements have focused on the New Left and on participations in the US civil rights movements.

Moving beyond specific movements and already explored contexts, the Swedish case may offer additional insight into the long-term effects of protest participation. Furthermore, previous studies may be argued to suffer in the methodological sense that they lack a control group of non- participants and data collected prior to participation in protests (Giugni 2008; Giugni 2004;

McAdam 1989). Exploring the Swedish context therefore aims to bring about more valid findings, and to further add to the literature on the consequences of social movements. If participation in protests is, as suggested, relatively accessible for Swedish citizens, then it should be of interest to explore possible differences in life-courses between participants and non-participants. Any findings indicating that protest participation has enduring biographical consequences should hence be regarded as important, due the arguably high accessibility for citizens to participate politically as well as the country’s fairly high rate of protest participation.

(18)

15 3. Research Design

3.1. Choice of Method

Giugni (2013) highlights three types of shortcomings in previous research on the biographical consequences of movement participation. First, they lack data on activists prior to and after involvement since most of them focus on a specific time period. Second, most studies are delimited to analysing only two time periods. And third, they often focus on a narrow geographical area. This study further argues that a great part of previous studies does not account for confounders in the form of socioeconomic background. This demonstrates the need of a research design that allows for attaining data on activists during several different time periods and from a larger population, which could improve the generalisability of the results. While this study simply cannot solve all of the above-mentioned issues, the use of longitudinal data is a good starting point. To be able to draw any conclusions about the effect of protest participation on individuals’ life-course outcomes, it is crucial to include a control group of non-participant respondents to avoid the risk of finding spurious correlations. More importantly, it is essential to include confounding variables in the form of the socioeconomic background of the respondents’

parents, as these factors are expected to affect both the independent and dependent variables.

In order to examine the research question of the paper, the method of linear regression analysis will be employed. The aim is to explore and statistically test the effect of protest participation and the interaction with gender, and to further examine the intergenerational effects. The proposed relationships will be controlled for other factors that might explain why some individuals decide to participate in protests and why they tend to make certain life-choices. Regression analysis is hence a suitable method for investigating correlation between variables and to test whether it is sensitive to other possible explanations by adding control variables (Angrist and Pischke 2015).

In this paper, four multivariate regression models are estimated in order to contribute to our understanding of the effects of protest participation and to control for confounding variables. The first regression model attempts to recreate previous studies examining the effect of protest participation on life-course outcomes and does accordingly lack a control for the socioeconomic background of the respondents’ parents. The second regression model aims to resolve this issue by adding control variables in the form of parents’ nationality and level of education. The third regression model examines the effect of the interaction between protest participation and gender on life-course outcomes. And lastly, the fourth regression model analyses the intergenerational effects of protest participation on the respondents’ children.

(19)

16 The panel data includes six panel waves in total, stretching from 1968 to 2010. The independent variables are only measured in the first four waves, as we want to be able to estimate the effects in a later time period. In the first column in model one to four, the first four panel waves are merged into one in order to estimate the aggregate effect of protest participation in two time periods later. A potential problem in the estimated models is that the same individuals may be included several times in the data, since they can still be categorised as being young in 1974 if they were young in 1968. This is due to the fact that according to this study, a young respondent is coded as being between 15–25 years old. In addition, because some outcome variables are binary, it is necessary to adjust for heteroscedasticity, which by nature is present in these kinds of models (Lewis-Beck and Lewis-Beck 2016). To solve this issue, the standard errors are clustered on the individual level. In the remaining models, robust standard errors are used.

3.2. Data Selection

This study employs a longitudinal dataset collected from the Swedish Level of Living Survey (Levnadsnivåundersökningen; LNU), designed by the Swedish Institute for Social Research at Stockholm University (SOFI 2016; SOFI 2018; SOFI 2020). The LNU surveys have been conducted six times in total during the years of 1968, 1974, 1981, 1991, 2000, and 2010, and were performed by Statistics Sweden. The data consists of a nationally representative sample of the Swedish population, including respondents who are between 15–75 years old. The response rates were the highest in the first survey, and the lowest in the last one.² The respondents were visited at home and received questions on actual living conditions, concerning for example family, health, education, economy, political participation and employment. In conjunction with the fifth survey in 2000, children who were 10–18 years old and who were living in the main respondent’s household were interviewed regarding their living conditions as well as attitudes and behaviour.

These interviews were conducted simultaneously to the parents’ interviews. The children listened to the questions from a tape recorder and wrote down their answers in a separate response questionnaire. It was therefore not possible that anyone else could have had an impact on the children’s responses in any way. The children were interviewed a second time in 2010 in connection to the sixth and final wave. 1304 children participated in first survey in 2000, as compared to 920 participants in the second survey in 2010.

2 Response rates in the surveys: 90,8 percent in 1968; 85,2 percent in 1974; 82,4 percent in 1981; 79,1 percent in 1991; 76,6 percent in 2000; and 60,9 percent in 2010.

(20)

17 3.3. Operationalisation

3.3.1. Dependent and Independent Variables

Since the aim of this study is to explore how protesting affects individuals’ life-course patterns, the purpose of the main independent variable is to measure participation in protests. The assumption is that protesting during an early stage in life has an effect on individuals’ subsequent life-course choices. Thus, the sample of the main respondents is delimited to individuals who were between 15–25 years old in each panel wave and who stated that they either had or had not participated in protest during the last ten years. It should be reasonable to assume that being between the ages of 15–25 make up the formative years of one’s life, in the sense that it is during these years that people in general make the choices that will affect the rest of their lives and what life-courses they will embark upon. After all, that is when most people decide if they want to attend university, perhaps they meet a partner, or take a job opportunity that will set them on a certain path. Evidently, it is not plausible to measure the effect of a 70-year-old participating in a protest in 1968, as this person is likely to have already made his or her life-choices. Participation in protests is hence argued to work as an indicator for something that separates participants from non-participants. The main independent variable protest participation therefore measures whether the respondent had participated in a public protest or not during the last 10 years. It is constructed as a dichotomous variable, where being young and having participated in a public protest is coded as one and being young and not having participated is coded as zero. Therefore, in each panel the respondents who were 15–25 years old are defined as being young at the time of the survey, from which participation in or non-participation in protests will be estimated. The respondents are then followed during each panel to estimate the long-term effects of protest participation. To account for the use of a smaller sample, a sensitivity analysis is performed.

These models, which are available in the appendix, include a larger sample and intend to test whether the results are robust to changes in the assumptions of this study.

An apparent limitation is that the variable does not imply what type of protest the individual participated in. It can therefore not be determined if it was a more routine, run-of-the-mill, type of protest, or if it was of a more disruptive manner. Likewise, the protest participation variable does not imply what kind of protest it was in terms of aim. If the respondent stated that he or she had participated in protest during the last ten years, this study can neither conclude the ideological foundation of the protest nor its aim. Nonetheless, it is reasonable to assume that individuals who decide to participate in one of these types of protests will differ from those who do not protest,

(21)

18 since attending a protest may require certain resources of the individual regardless of the protest type. This is further supported by Van Dyke et al. (2000, 175), who found that “the life-course impact of individual activism varied very little across movements”. Thus, since the aim of this study is to explore the effect of protest participation on life-course outcomes overall it should not constitute a major issue, but should nonetheless be taken into account when reading the rest of the study.

The second independent variable gender is constructed as a dummy variable, where men constitute the reference category with the value of zero and women take the value of one. An obvious shortcoming here is the fact that the variable measuring the respondents’ gender in the panel does not account for any other gender identities, thus it is possible that a part of the population is excluded in the data. This study is hence limited to exploring the binary gender identities of being either a woman or a man. While it may restrict the analysis in some senses, the data as well as the aim of the paper should still allow for a sufficient analysis of any gender differences in individuals’ life-courses.

All outcome variables are measured two panel waves after having protested or not. The first dependent variable measures whether the respondent is married or not, in which unmarried respondents constitute the reference group and married respondents are coded as one. Being married may include partners who are living together as well as not living together. The second dependent variable measures the respondent’s number of children. It is constructed as an interval variable measuring the respondent’s total number of children, which includes those who are still living at home with their parents and those who have left home. For instance, they may have become of the age where they move to their own place. The variable only includes the children that the respondent had and does not include the children of a new partner. The third dependent variable education measures the number of years the respondent has been enrolled in educational institutions subsequent to having protested but does not allow for determining the type of education. However, as this paper seeks to determine any differences in level of education between participants and non-participants, the number of years should be sufficient in order to be able to assess whether protest participants differ from non-participants in terms of educational level. The same assumption is applied to the fourth dependent variable measuring the respondent’s total amount of work years. The number of work years the respondent has undertaken is constructed as a numeric variable of years in employment, ranging from 1–67 years.

The variable should hence work as a proxy for any differences in employment status among the

(22)

19 respondents. For instance, a higher number of work years might indicate that the respondent has had a consistent number of jobs or has stayed within the same job throughout the years, whereas a lower number of work years could mean that the respondent has been unemployed in some periods, possibly due to a frequent change in jobs, as suggested by McAdam (1989). The fifth dependent variable simply measures the respondent’s monthly net income in Swedish Crowns.

Because income distributions in general are skewed, the log of income is used which also allows for easier interpretations of the outcome variable. For instance, a coefficient of 0.05 is interpreted as a 5 % increase in income. The final dependent variables measure the respondent’s further involvement in political as well as union activities. The variables political activity and union activity are constructed as interval variables. One means that the respondent is not a member in neither a political party nor a trade union. Two shows that the respondent is a passive member. Three indicates that the respondent is an active member. And four means that he or she holds a position of trust within the organisation.

In order to measure the intergenerational effects of protest participation, the study uses panel data on the children of the respondents in the main LNU-survey. In this study, the children of the main respondents are matched accordingly to the delimited sample. As described earlier, the children were interviewed the first time in 2000 and the second time in 2010. To assess whether parents’ previous involvement in protests affect their children’s political attitudes and behaviour, this study will employ variables on these issues. The children were asked a number of questions to find out if they had done anything to influence or express their opinions during the last twelve months. All variables are constructed as dummy variables, where answers stating that the children did not do what the question asked is constructed as the reference category, and any answers saying that they did do what the question asked is coded as one. The first variable protest participation examines if the child has attended a protest. The variable meet politician asks whether the child have attended a political party. The variable asking if the child have met with a non-political organisation, meet organisation, is also used to measure any other type of engagement. The variable symbols ask whether the child have worn political symbols, labels or stickers of any kind. The variable eco goods measures if the child has chosen products or goods that meet the requirements for eco-labelling, i.e. green labelling. The variable letter asks if the child have written a letter to the editor. And finally, the variable no activity that the child has participated in none of the above-mentioned activities.

(23)

20 3.3.2. Control Variables

When testing the correlation between the independent variable protest participation and the dependent variables measuring certain life-course outcomes, it is of fundamental value to control for confounding variables to avoid omitted variable bias. In other words, we want to isolate the relationship from other factors that might affect the dependent and independent variables simultaneously, and to hold constant the most apparent differences between protest participants and non-participants (Angrist and Pischke 2015). In the case of this study, the aspects that might affect individuals’ tendency to protest may be reliant on the resources they attained during their upbringing. Resources in terms of money, time and knowledge may affect whether one decides to protest or not, while aspects concerning one’s political values may have been acquired through socialisation processes (Bandura 1982; Beck and Jennings 1975). Despite the fact that we have included the respondent’s age in the independent variable it is still necessary to control for age by adding their year of birth to the analyses as age may still be highly correlated with proclivity to participate in protests as well as various life-course outcomes.

It is further likely that the socioeconomic background of the respondent’s parents will affect the respondents’ life-course patterns. In order to further avoid omitted variable bias, these factors must therefore be controlled for. Drawing from the classification proposed by Statistics Sweden, the educational level of the parents should work as a proxy for socioeconomic status (Haldorson 2008; SCB 1982). While the parents’ income level could be argued to be a suitable indicator for socioeconomic status, it is likely that such a variable would already be highly correlated with level of education. In addition, it is likely that people with immigrant backgrounds face other obstacles than those born in Sweden during their lives, hence, it should likewise be considered a determinant of socioeconomic background. In line with Wennerhag’s (2017) approach, the nationality of the respondents’ parents is constructed as dummy variable, where all countries but Sweden constitute the references category and one equals that the mother or the father of the respondent was born in Sweden. The variable measuring the respondent educational level of the respondent’s mother and father respectively is constructed as follows: the value of zero means that the parent does not have an education; values up to one that the parent attended elementary school; values up to four that the parent attended high school; and values up to six that the parent attended university.

(24)

21 3.4. Limitations

Because this paper delimits the LNU data to only consist of young protest participants and non- participants between the ages of 15–25 in each panel wave, it results in much smaller sample, compared to if we were to include protesters and non-protesters of all ages. Hence, it may be argued that this study consists of a fairly small number of observations. This is arguably a limitation in regression analysis as a small number of observations simply leads to larger standard errors, which ultimately might hinder the chance of finding statistically significant results. As noted earlier, a robustness check is performed to test the sensitivity of the results. Thus, the fairly small sample in the main regression analyses should not constitute a major issue but should be taken into account when interpreting the results of the study.

Moreover, the design of the study does evidently not allow for making causal claims, as regression analysis only allows for finding correlation between variables. For instance, in order to be able to find the causal mechanism explaining the relationship between protest participation and life- course outcomes, it would be required to complement the study with qualitative methods as well as quantitative methods of more experimental nature. Therefore, this paper can only assess whether there might be a correlation between the independent and dependent variables, which may still be considered an important result.

(25)

22 4. Results and Analysis

4.1. Descriptive Statistics

When comparing different groups in regression analysis, it is necessary to describe the characteristics of the groups to assure that they are comparable to begin with. The tables below thus present the relevant descriptive statistics. As a reminder, the dependent variables are measured two panel waves after the independent variables.

Table 1. Descriptive statistics of main respondents.

Independent Variables Participant Women

Non-participant Women

Participant Men

Non-participant Men

85.6 % 14.4 % 85.7 % 14.3 %

Dependent Variables

Married 0.79 0.78 0.77 0.77

Children 1.7 1.6 1.5 1.6

Education 12.1 11.6 12 11.6

Work 15 15.7 16.7 15.8

Income 9,151 9,178 10,619 9,356

Political activity 1.2 1.2 1.2 1.2

Union activity 2.3 3 2.4 2.3

Control Variables

Age 21 21 21 21

Father’s nationality 0.9 0.9 0.9 0.9

Father’s education 2 1.9 2 1.9

Mother’s nationality 0.9 0.9 0.9 0.9

Mother’s education 1.9 1.8 1.9 1.8

Note: Dependent and control variables are reported in means.

The independent variables describe the distribution of respondents in each group, from which it can be concluded that the proportion of non-participant men and women correspond with each other. The same can be applied to participant men and women. In line with the regression analyses in the next section, the dependent variables are measured two panel waves after the independent variables. Because the independent variable is constructed as a dummy variable, the

(26)

23 dummy variable married thus shows that almost 80 % of all respondents in the sample are married subsequent to having, or not having, protested. Reviewing the table, it shows that the mean number of children for the sample is around having one to two children. The respondents have studied for almost 12 years on average and have generally been employed for 15–16 years. The income variable shows that participant men earn more compared to both non-participant men, and women in general. The level of political activity does not differ much between the groups, as most of the groups display a mean value of 1.2, meaning that most respondents are not members of a political party. As for union activity, non-participant women are on average members of a trade union, whereas in the rest of the groups the respondents are passive trade union members.

Comparable characteristics are of importance as we want the groups to be as similar as possible in order to be able to estimate any differences in life-course outcomes. Looking at the control variables, this should not constitute a problem as the means for all variables are as good as similar in all groups. For instance, the mean age for the respondents is 21 years old, which is to be expected since the sample is comprised of respondents between the ages of 15–25.

Graph 1. Mean share of young protest participants.

The graph presents the percentages in means of young protest participants during the first four panel waves. In 1968, the amount of 15–25-year-olds that stated to have protested were almost 6 percent. In 1991, this number had risen to approximately 20 percent. From this it can be concluded that, since the global emergence of protest activities in the late 1960s, there has been a consistent increase in the amount of young protest participants in Sweden.

0 5 10 15 20 25

1968 1974 1981 1991

(27)

24 Table 2. Descriptive statistics of child respondents.

Independent Variables Participant Parents Non-participant Parents

16.7 % 83.3 %

Dependent Variables

Protest participation 0.03 0.05

Meet politician 0.03 0.06

Meet organisation 0.13 0.16

Symbols 0.21 0.15

Eco goods 0.31 0.27

Letter 0.06 0.08

No activity 0.47 0.47

Control Variables

Age 14 14

Gender 0.5 0.5

Parent’s education 13.6 11.4

Parent’s nationality 0.99 0.99

Note: Dependent and control variables are reported in means.

Table 2 reviews the variable means for the child respondents in the sample. Because all dependent variables are constructed as dummy variables, the means are interpreted in percentages. For instance, only 3 % of the children with participant parents stated to have participated in a protest themselves, whereas 5 % of the children with non-participant parents reported to have protested. It becomes clear here that almost half of the children stated to not have engaged in any political activities at all. Looking at the control variables, the children were on average 14 years old, and equal parts of the sample stated to be a boy or a girl. Most of the children’s parents were of Swedish nationality. And lastly, the table shows that participant parents tend to study for longer than non-participant parents on average.

(28)

25 4.2. Regression Analyses

In this section, four multivariate regression analyses are carried out to examine the relationship between protest participation and life-course outcomes. The main independent variable estimates protest participation among the youngest respondents in the four first waves. The sample in the regression analyses thus consists of participant and non-participant respondents between the ages of 15–25, where young non-participants form the reference group and young participants are coded as one. The dependent variables measuring life-course outcomes include the respondents’

marital status, number of children, years of education, years in employment, monthly income, as well as involvement in political organisations and trade unions. As shown in the theory section, in order to estimate the effect of protest participation on life-course patterns, scholars have previously measured life-course outcomes in terms of marital status, number of children, education, occupation, income, as well as further involvement in political activities. Table 3 therefore aims to replicate the regression analyses used in previous studies which examine the consequences of protest participation. Table 4 seeks to address the issue of omitted variable bias by controlling for the socioeconomic background of the main respondents’ parents. Table 5 tests the interaction between protest participation and gender and its effect of life-course outcomes.

Table 6 estimates the intergenerational effects of protest participation.

This study will begin with a replication of the regression analyses that have been carried out in previous studies. In model 1 in the first three regressions, the first four panel waves are merged into one in order to be able to assess the combined effect of protest participation during all panel waves. The outcome variables are measured two panel waves after having or not having participated in a protest. For instance, if protest participation is measured in 1968, then the life- course outcomes are measured in 1981. And if protest participation is measured in 1974, outcomes are measured in 1991, and so on. Hence, model 2 to 5 seek to capture any possible generational differences between protest participants.

(29)

26 Table 3. The effect of protest participation on life-course outcomes.

(1) Protest

(2) Protest 1968

(3) Protest 1974

(4) Protest 1981

(5) Protest 1991

Married 0.0343

(0.0221)

0.0134 (0.0529)

0.0302 (0.0496)

0.113**

(0.0344)

-0.0456 (0.0461)

Children -0.0415

(0.0647)

-0.0597 (0.140)

0.07 (0.148)

-0.185 (0.108)

0.03 (0.125)

Education 0.76***

(0.152)

0.96*

(0.482)

1.3***

(0.374)

0.635**

(0.216)

0.494*

(0.249)

Work years -0.383

(0.209)

0.01 (0.685)

-0.0266 (0.495)

-0.39 (0.291)

-0.166 (0.339)

Log income -0.0563

(0.108)

-0.433 (0.403)

0.238 (0.364)

-0.0929*

(0.0408)

0.0332 (0.0407)

Political activity 0.131**

(0.0416)

0.282*

(0.139)

0.178 (0.111)

0.139*

(0.0597)

0.0527 (0.0586)

Union activity 0.122*

(0.0597)

0.01 (0.135)

0.213 (0.141)

0.121 (0.103)

0.0573 (0.125) Parents’

background

N 2,237 812 399 620 406

Notes: *** p<0.01, ** p<0.05, * p<0.1. In model 1, the standard errors reported in parentheses are clustered on the individual level. In model 2–5, robust standard errors are reported in parentheses.

Model 1 shows that protest participation is positively correlated with having studied for a longer time period. We see that education is statistically significant at 99 % confidence level, which means that it is fairly certain that if the respondent stated to have participated in a protest prior to the survey in 1968, protest participants are on average likely to study for 0.76 years longer than non-participants. The value of the coefficient could in some senses be interpreted as rather weak.

However, considering how many years people study in general, having studied for one year longer on average could indeed be argued to make a difference. For instance, if a bachelor’s degree is attainable within three years, one year is certainly of substantial value. Hence, the finding that protest participants are likely to study more than their non-participant counterparts may be interpreted as an important result.

The Lifelong Consequences of Protesting: A Longitudinal Analysis of the Gendered and Intergenerational Effects of Protest Participation on Individuals’ Life-Course Patterns