Over the last eight years, contentious actions such as street protests and sit-ins have been a constant presence in news reports from the MENA region. While a significant number of academic, journalistic, and think-tank articles have focused on the causes of social discontent and contentious actions in the region since 2011, few works have used a quantitative approach to investigate the determinants of protest participation. This paper contributes to this literature by offering a comparative analysis of protest participation in 2011 and 2018–2019 in four Arab countries (Egypt, Tunisia, Jordan, and Algeria), using the data provided by the second and fifth wave of the Arab Barometer. It finds a persistent and increasing level of protest participation throughout the two waves, even in countries such as Egypt, where repressive policies have worsened over the last years. Moreover, gender, education level, income, and level of religiosity emerge as the most consistent determinants of participation in both waves. Overall, being male, having higher education, being from the upper half of the income distribution, and being non-religious are the characteristics significantly linked to protest participation in most cases. However, considerable differences persist from country to country.The picture emerging from the analysis shows a persistent and increasing level of contention, which can be interpreted as the sign of an ongoing process of social and political change throughout the region.
Since 2011, words such as “protest,” “demonstration,” and “uprising” have featured increasingly in academic and journalistic pieces on the MENA region. The wave of street protests that hit most Arab countries in 2011—commonly labeled as the “the Arab Spring” or the “Arab Awakening” —has left a lasting impression in public opinion and dramatically changed the field of Middle Eastern Studies. A significant number of works published over the last eight years have focused on the root causes of the uprisings and on the (few) successes and (many) disappointments brought by the transitional phases that followed.
Some eight years after the 2011 movements, a new wave of huge street demonstrations erupted between the end of 2018 and the first half of 2019. These led to the end of Omar Al-Bashir’s 30-year long presidency in Sudan and the 20-year long tenure of Algeria’s President Abdelaziz Bouteflika, marking the beginning of a period of political transition in both countries. Over the following months, new protests hit several other Arab countries, such as Jordan, Lebanon, and Iraq, some of which are ongoing at the time of writing.
This new wave of protests has led some observers to claim that a new Arab Spring is in the making. However, many have also stressed the significant differences between the 2011 movements and the present ones. For instance, the latter originated in countries, such as Algeria and Sudan, which were perceived—somewhat mistakenly—as having been mostly insulated from the contentious episodes of 2011. Second, the new protest movements have been far more durable compared to those in 2011: while eight years ago most movements dissolved after few weeks (usually after the president’s resignation or after major concessions) or quickly transformed into armed confrontations (such as in Libya and Syria), the current movements have lasted for months—and, in the cases of Sudan and Algeria, for over half a year. Third, while in 2011, Islamist parties quickly emerged as the most organized and effective components of the protest movements—often gaining the upper hand during the transitional periods—they have been mostly absent from the current movements.
This paper aims to contribute to the study of these movements by providing a comparative analysis of the demographic determinants of protest participation in both 2011 and 2018–2019. It utilizes data provided by the second and the fifth waves of the Arab Barometer surveys conducted in Egypt, Tunisia, Algeria, and Jordan. It highlights analogies and differences between the protest movements in 2011 and those in 2018–2019 and contributes to isolating those root factors that have been crucial in pushing people in the streets over the last decade.
The paper proceeds as follows. The first section includes some notes regarding the methodology utilized. The second section draws on the aggregated data from the four countries considered in order to provide a general overview of the determinants of protest participation and their evolution between 2011 and 2018-2019. The third part analyses the data from each country separately, to underscore the differences among the four countries considered. Finally, the fourth part provides some concluding remarks that present the findings of this paper regarding the role of each demographic characteristic and their overall evolution over the last eight years.
This paper considers four countries: Egypt, Tunisia, Algeria, and Jordan. Two criteria guided the choice of these case-studies. First, data from these countries are present in both the second and the fifth waves of Arab Barometer surveys. Second, they all saw the coagulation of protest movements —either in 2011 or in 2018–2019 (or both)—before the collection of the surveys was conducted. In countries such as Sudan, Iraq, and Lebanon, the most recent protests occurred after the last wave’s surveys had been already collected. Therefore, their inclusion in the analysis would not have allowed consideration of the results regarding participation in the last protests.
The analysis is restricted to the demographic determinants of protest participation; the respondents’ ideological features and perceptions—which the Arab Barometer surveys also investigate—are not considered in this study. In fact, this paper aims to depict the profile of the average protester in each country based on their objective demographic characteristics. However, the influence of ideological features and individual perception in protest participation should be the subject of future research.
The determinants of protest participation were analyzed using logit regressions, with participation in protests (at least once) as the dependent variable, and demographic variables (scale ordered) included in the surveys as independent variables. The ones included in the Arab Barometer surveys are gender, age, income level, education level, and urban or rural place of residence. Only for the regression conducted on the 2011 Algeria sample was the income variable not included due to the large amount of missing data (more than half of the sample). Furthermore, another variable—corresponding to the survey question “Do you pray daily?”—was added as a proxy for the individual level of religiosity, analogous to the approach taken by Beissinger and colleagues in their study on the 2011 uprisings in Egypt and Tunisia.
To determine whether a respondent participated in one or more demonstrations during the considered period, I utilized the specific question included in both survey waves: “During the last three years, did you participate in a protest, march, or sit-in?” Four possible answers are available: “Once”, “More than once”, “I have never participated,” and “I don’t know”. To avoid distortions in the results, I considered only the respondents that selected one of the first three possible answers and transformed the question into a binary variable, merging in the same category those who selected “Once” or “More than Once”. In this way, I obtained a binary variable dividing those who participated at least once in street protests over the previous three years and those who certainly did not. It is important to note that for the 2011 wave, the Arab Barometer survey asked slightly different questions in Egypt and Tunisia. In Egypt, the question was: “Did you participate in the protests against former president Hosni Mubarak between January 25 and February 11, 2011”? In Tunisia, it was “Did you participate in the protests against former president Zain Al-Abdeen Ben Ali between December 17th, 2010 and January 14th, 2011”? The level of specificity entailed in the two questions—especially the narrow time period they consider—may generate slight distortions in the comparison with the data collected in the other countries or with the data from the 2018–2019 wave, where the question asked in the survey referred to a time period of three years. Even if such distortions should not be exaggerated—since it is plausible that most people who participated in protests over the three previous years also participated in the protests against Mubarak and Ben Ali—it is important to keep account of them.
In the 2011 and 2018–2019 Arab Barometer waves, respondents were asked to report their level of income in two different ways: while in 2011 they were asked to write the exact amount corresponding to their household income (in both local currency and US dollars), in 2018–2019 they were asked to indicate whether their income was above a certain amount expressed in the local currency, representing the median income level of their country. In order to harmonize the two waves, the income variable in the 2011 wave was also transformed into a binary one: the median income for each country was calculated using the respondents’ declared incomes, and then respondents were grouped according to whether their income was above or beneath that calculated median.
General overview: analysis of the determinants of street protests in the region
This section analyses the aggregated data from all four countries, comparing data from the 2011 Arab Barometer wave and the 2018–2019 wave. Figure 1 shows the proportion of people who in 2011 and 2018–2019 affirmed having participated in a street protest at least once over the previous three years.
This first comparison is already interesting: in 2018–2019, the share of people who participated in protests is larger than that of 2011. Hence, although the public emphasis put by international media on the so-called “Arab Spring” in 2011 was significantly higher, in 2018–2019, a greater share of citizens from the four countries affirmed having participated in street protests at least once.
Figure 2 shows participation shares disaggregated for each country. Also, in this case, the figures are rather surprising: Algeria shows the highest rates of participation, with almost identical shares for both the waves. Egypt saw an increase of people who affirmed having participated in at least one street protest, despite the strict rules and controls against contentious actions introduced by the new Egyptian regime led by Abdel Fattah Al-Sisi. Tunisia, more predictably, saw a decrease, as did Jordan, despite the mass protests that occurred between 2018 and 2019, especially in the capital Amman.
Table 1 shows the main demographic determinants of protest participation as they emerged from the logit regression carried out using the data from the 2011 wave. Gender, education, income, and religiosity emerge as significant determinants of participation. Hence, being male, with higher education and an income above the median, and being less religious, are the characteristics making participation more likely. Table 2 shows the results of the regressions for the 2018-2019 wave. Gender, education level, and income are still significant, while this time the religiosity level is only weakly significant (0.1). In addition, in this wave age also emerges as significantly and inversely linked to participation, which means that younger individuals were more likely to participate.
Table 1 – Demographic determinants of participation in Algeria, Egypt, Tunisia, and Jordan (aggregated data - 2011 wave)
Table 2 - Demographic determinants of participation in Algeria, Egypt, Tunisia, and Jordan (aggregated data - 2018-2019 wave)
Overall, the analysis results of the aggregated data are rather puzzling for at least five reasons. First, although Algeria was not in the spotlight in 2011, it shows the largest share of people affirming their participating in protests in both waves. Second, Egypt showed a higher rate of participation in the second wave despite the strict measures introduced by the Al-Sisi government and the scarce presence of news regarding protest movements in the country over the last three years. In direct contrast, while Jordan saw significant demonstrations in 2018 and 2019, the participation rate was slightly higher in 2011. Third, while the protest movements in 2011 were commonly labelled as “youth revolutions,” age appears not to have been a significant determinant of protest participation in the first wave. Contrarily, it emerges as significant in the second wave. Fourth, while many commentaries have depicted the protests as the reaction of the most deprived social groups against increasing inequality and poverty, income emerges as directly related to participation in both waves, meaning that individuals belonging to the upper part of the income distribution are more likely to have participated in demonstrations. It is noteworthy, however, that the coefficient of the income variable decreased considerably between the two waves, indicating an increased share of participants from the lower part of the distribution in the second wave. Finally, despite that Islamist parties gained the upper hand during the transitional periods that followed most of the 2011 protest movements, protest participation emerges as inversely related to the respondents’ level of religiosity in both waves (although the significance of religiosity level in the second wave is weaker). This result may confirm the perception that Islamist parties benefited from the power vacuum created in countries like Tunisia and Egypt during the transitional period but did not constitute a significant share of those who took part in the protests.
In the next section, the analysis is repeated with data from each country in order to highlight equivalences and differences among the societies considered.
Protest determinants by country
Egypt saw major street protests in 2011 that led to the end of Hosni Mubarak’s presidency, which was followed by a transition process that culminated in a military coup in 2013 that brought former military general Abdel Fattah Al-Sisi to power. The new regime introduced several new measures to tighten access to public spaces and limit public participation in order to avoid the repetition of another uprising like in 2011. However, despite there being no news reports of significant demonstrations since 2013, the share of people affirming having participated in at least one street protest in 2018–2019 increased significantly in comparison with 2011, from 8% to 14.3%. To account for this mismatch between perception and data, we might hypothesize that over the last years, protests were sparser and more local, focused on specific issues, and not directed against the ruling regime. For these reasons, they may have been less controllable and more tolerated by the government. These results, in any case, show the persistence of—and even growth in— the level of contentious politics within Egyptian society.
Tables 3 and 4 show the determinants of protest participation in Egypt in 2011 and 2018–2019, respectively. In both cases, gender, urban/rural divide, and level of education emerge as significant determinants. The main difference is the fact that age was the fourth-strongest determinant in 2018–2019—in 2011, it was weakly significant (0.1) and with a small coefficient. Hence, in 2011 being male, having high education, and being from an urban area were the characteristics that made participation in demonstrations more likely. In 2018–2019, the same characteristics emerge as significant with the addition of being younger than average. This is consistent with what was observed in September 2019 when a short-lived, yet significant, protest movement emerged in a few Egyptian cities against the continuous deterioration of living conditions. Local and international witnesses stressed the rather young age of the protesters, many of whom were too young to have been part of the 2011 uprising.
Table 3 - Demographic determinants of participation in Egypt (2011 wave)
Table 4 - Demographic determinants of participation in Egypt (2018-2019 wave)
In 2011, the Tunisian uprising generated a wave of protests that hit most Arab countries in the following weeks. In Tunisia, they resulted in the end of the 34-year old presidency of Zine El Abidine Ben Ali. The transitional period that followed was marked by the electoral victory of the Islamist party Ennahda, which, contrary to what had occurred in other contexts such as Egypt, adopted from the start a strategy of forming coalition governments with secularist parties. A democratic, semi-presidential constitution was ratified by the elected Constitutional Assembly in 2014, and new legislative elections were successfully held the same year and again in 2019. Tunisia is considered the only successful case of democratization emerging from the 2011 Arab Spring. The share of participants in the 2011 protests in the country was remarkably high at 16%, while a smaller share of people (11.8%) affirmed having participated in demonstrations in the 2018–2019 survey.
Table 5 shows the results of the regressions for the 2011 wave. Gender, age, and education level emerge as the strongest determinants of protest participation in 2011. The level of religiosity shows only a weak explanatory power for participation. Hence, being younger than average, more educated, and male made it more likely for an individual to take part in the demonstrations that toppled president Ben Ali. Also, being slightly less religious than average emerged as a weak but still significant determinant. Opposite to observations for the case of Egypt, where being from an urban area was strongly linked to participation, in Tunisia, the urban/rural divide was not as significant. That may reflect the fact that while in Egypt protests sparked and developed in urban areas, in Tunisia, they originated in the small centers of the country’s impoverished south and then spread toward the larger and more prominent urban centers of the coast.
Table 6 shows the regression results for the 2018–2019 wave. They are mostly consistent with those of 2011. Gender, age, and education level remain the most significant determinants, with the coefficient of the latter being considerably higher. Also, in this wave, the level of religiosity gains a 0.1 significance with a negative coefficient, meaning that participation over the previous three years was linked to more religiosity.
Table 5 - Demographic determinants of participation in Tunisia (2011 wave)
Table 6 - Demographic determinants of participation in Tunisia (2018-2019 wave)
Although in 2011, many commentators expected Jordan to follow the path of Tunisia and Egypt, the country saw only limited protests. People in the streets refrained from targeting the monarchy, directing their calls to the ruling government. To appease the street, the king sacked three prime ministers in 18 months. Promises of reforms and temporary measures to relieve the economic situation managed to bring the movement gradually to an end. However, over the following years, new harsh austerity measures were introduced to help the national budget cope with the deterioration in the country’s financial position. This exacerbated popular discontent, leading to the emergence, at first, of sparse protests and then to the coagulation of new major demonstrations that for several weeks took place especially in the capital Amman. The government reacted by sacking the government led by Hani Mulki and introducing new coercive measures targeting the use of social media. This appeared to extinguish the movement, especially the demonstrations organized by the teacher worker unions. At the time of writing, no more major protests had occurred since mid-2019, when the government accepted the protesters’ requests for salary increases.
At first glance, the data from Jordan show somewhat surprising figures: despite the more significant media coverage of the 2018–2019 the movement in had a and overall lasted significantly longer than the one in 2011, the share of people that affirmed of having taken part in demonstrations in 2018-2019 (5.1%) is significantly smaller than the one in 2011 (7.9%). Also, the regression results from the two waves show some significant differences.
Table 7 and Table 8 show the results from the 2011 and the 2018-2019 waves, respectively. In 2011 only income and religiosity emerge as strong determinants, while gender emerges as a weak one (0.1). The income coefficient is particularly high, meaning that being from a household significantly more affluent than the average was linked to participation. Despite Jordan’s long history of peaceful Islamist militancy—especially embodied by the local Muslim Brotherhood—protest participation in 2011 was inversely related to religiosity (i.e., the less religious the individual, the more likely her participation). Finally, participation is also weakly linked to being male.
The data from 2018–2019 show some interesting developments. While income, religiosity, and gender are still significant determinants (with gender gaining stronger significance), in the second wave also age, education, and the urban/rural divide emerge as linked to participation. In 2018–2019, the protesters appear to have also been younger, on average, more educated, and from an urban area. This last characteristic may help explain why the latest protests have been perceived as more considerable than those in 2011, despite the share of participants appearing to have been considerably greater in the first wave. The coagulation of the protest movement in urban areas, especially the capital Amman, may have helped to make demonstrations larger and more visible than in the past.
Table 7 - Demographic determinants of participation in Jordan (2011 wave)
Table 8 - Demographic determinants of participation in Jordan (2018-2019 wave)
In 2011 Algeria was among the first countries to be hit by the wave of protests that originated in Tunisia. After a rather intense beginning, demonstrations became gradually sparser and more intermittent, lasting until the first months of 2012. According to local and international observers, the movement was quelled by a mix of repression and concessions in terms of welfare and public jobs, in addition to the widespread fear of recreating the kind of situation that had led to the Algeria civil war through the 1990s. Almost eight years later, a new protest movement erupted in February 2019 after president Bouteflika announced his intention to run for a fifth term in office. The protesters lamented the absence of any political reform and development in the face of the continuing deterioration of the living conditions of the population, made harsher by the austerity measures introduced over the previous years to cope with the dramatic decrease in oil prices. The new movement caused Bouteflika to withdraw his candidacy and to announce the beginning of a transitional period that would lead the country to new elections, scheduled on December 12th, under the supervision of a provisional government and the army’s leadership. At the time of writing, these concessions had failed to quell the movement that over the last months has engaged in new massive protests to obtain more concessions in terms of institutional renovation and power redistribution away from the traditional élite.
Overall, Algeria shows very high shares of protest participation—significantly higher than any other country analyzed in this paper—in both the 2011 (18%) and the 2018-2019 waves (18.4%). This can be in part attributed to a widespread culture of political contention in addition to the long-term cumulation of socioeconomic grievances over the last decades. As explained previously, the regression results for the first wave do not include the income variable due to the significant amount of missing data. Table 9 shows the results for the 2011 wave. Gender and education levels are the only two significant determinants of participation. Males with higher education were more likely to take part in the protests. Table 10 shows the results for the 2018–2019 wave.
In the second wave, more determinants become relevant. Apart from gender and education level, also income and religiosity levels have become significant. As observed for other countries in this analysis, the level of religiosity is inversely linked to participation (i.e., the less religious the individual, the more likely her participation in protests). Also, income is inversely linked to participation, meaning that being beneath the median income makes a person more likely to have participated in a demonstration. Such a result is unique among the countries analyzed in this paper, where, when significant, income is always directly linked to participation (i.e., people having an income above the median are more likely to participate).
Table 9 - Demographic determinants of participation in Algeria (2011 wave)
Table 10 - Demographic determinants of participation in Algeria (2018-2019 wave)
The first element emerging from the analysis is the persistence of contentious politics within Arab societies eight years later the eruption of the so-called Arab Spring. Overall, the share of protest participation is two percentage-points higher, a result that reflects mainly the data from Egypt and Algeria. In this regard, the most surprising element is Egypt’s increased share of participation, which speaks against the claims of stability projected by the current regime.
In what follows, I summarize the results for each demographic determinant included in the analysis, in order to highlight constants and exceptions in their influence over participation.
Gender. The ensuring significance of the gender determinant—in favor of male participation—in almost all the regressions run in this paper reflects a patriarchal culture that curbs women's participation in protest movements that is still extremely widespread in the region. At the aggregated level, the gender determinant’s coefficient even increased between the two waves, suggesting that this gap is unlikely to disappear any time soon. Hence, since predictably Arab women have similar levels of discontent and grievances than their male counterparts, it is possible to say that, due to this cultural bias, current protest movements in Arab countries do not make effective use of a vast potential reservoir of additional participation.
Age. Despite having been labeled as “youth revolutions,” the movements in 2011 do not see age as a strong determinant of participation (with the notable exception of Tunisia). However, age has gained significance in the 2018-2019 wave. This seems to confirm, for example, what has been noted about the recent spark of demonstrations in Egypt in September 2019, when witnesses reported on the rather young age of protesters. Many of them were probably too young to have participated in the 2011 movement. This may indicate that the new movements that have emerged in 2018-2019 do not have as main protagonists the same individuals who participated in 2011.
Urban/Rural divide. Most of the reports about protests usually come from urban centers. The most powerful images of the 2011 uprisings were those coming from places such as Cairo’s Tahrir Square or Tunis’s Boughiba Avenue. However, the analysis of the data on protest participation shows that at the aggregated level the urban/rural divide was not a significant determinant of participation neither in 2011 nor 2018–2019 while it was significant only a few instances, such as Egypt (for both waves) and in Jordan (although weakly and only in the second wave). These results suggest the presence of vast, although sparse, protest movements in the countryside of most Arab countries that are significantly unreported and unresearched.
Education level. The significance of education level as a determinant of protest participation is the most consistent element emerging from the analysis. It was significantly linked to participation at the aggregated level in both waves and in most country-level analyses. In part, this may stem from the fact that educated youth is among the most suffering social groups in most Arab countries. However, the fact that age does not often appear as a strong determinant limits the explanatory power of this consideration. In general, this result confirms the findings of the literature on contentious politics, which tend to establish a direct and strong link between education and willingness (and capability) to participate in contentious actions.
Income: belonging to the upper part of the income distribution (above the median) has emerged as one consistent determinant of participation in both waves at the aggregated level and in several country-level analyses. Such results confirm the more recent strands of literature on the topic, such as resource theory and relative power theory. In contrast, these results contradict some of the typical narratives that have been reported by newspapers—and sometimes even by academic articles—on the 2011 uprisings, which depicted the protesters mostly as members of the most economically marginalized social groups.
Religiosity. The results regarding the role of individuals’ religiosity in determining protest participation are perhaps the most intriguing aspect emerging from this analysis. In fact, especially in 2011, the emergence of Islamist movements as powerful (sometimes the most powerful) actors from the uprisings could have been interpreted as proof of the strong role played by religiosity in the protests against (usually) rather secular authoritarian regimes. Nevertheless, an inverse link between religiosity and participation emerges in this analysis, remaining consistent throughout most tests conducted in this paper. Perhaps even more surprisingly, such a link is less strong in the 2018–2019 wave, which relates to protest movements in which Islamist movements have not had any meaningful role, at least so far. This element may be linked to the results regarding income and education level. In fact, more affluent, better-educated people are often also those less likely to be very religious. However, the dynamics between religiosity and participation in contentious politics in the MENA region need more scrutiny also in light of these results.
In sum, it is possible to affirm that the core participants in the protests in both waves are educated, secular citizens belonging to the middle and upper-middle class. However, the differences between the results of the two wave highlight a reduced (although still significant) role for income and religiosity in determing participation. These results seem to confirm the literature defining the middle class as the most important social group when it comes to organize and participate in contentious politics. Moreover, since the survey question regarding participation encompasses a period of three years in both waves, it is possible to affirm that the picture emerging from this analysis portays a situation of persistent and increasing level of contention throughout the region. This may indicate that we are in presence of an ongoing process of social and political transformation that may produce more contentious episodes, destabilization, and political change for years to come.
 1- Male; 2- Female
 Respondents were asked to write their age in numbers
 A binary variable indicating whether the respondent’s income is beneath or above her country’s median income: 1- Beneath the median income; 2- Income equal or above the median income
 The question includes 7 possible answers: 1- Illiterate; 2- Elementary; 3- Preparatory/Basic; 4- Secondary; 5- Mid-level diploma; 6- BA; 7- MA or above
 The original binary variable (1- Urban; 2- Rural) included in the survey was transformed into a new binary variable in which Urban is equal to 1 and Rural equal to 0
 The question includes 4 possible answers: 1- Always; 2- Most of the time; 3- Sometimes; 4- Rarely
 Mark R. Beissinger, Amaney Jamal, and Kevin Mazur, “Who Participated in the Arab Spring? A Comparison of Egyptian and Tunisian Revolutions,” Princeton University. Http://Www. Princeton. Edu/~ Mbeissin/Beissinger. Tunisiaegyptcoalitions. Pdf, 2012.
 Pamela Abbott, Andrea Teti, and Roger Sapsford, “The Myth of the Youth Revolution: The Role of Young People in the 2011 Arab Uprisings,” 2017.
 The New York Times, Why Islamism Is Winning.
 The New York Times, Egypt's Harsh Crackdown Quashes Protest Movement.
 Carnegie, Deconstructing Egypt's Protests.
 Middle East Institute, The Lesson from Morocco and Jordan: Reform or Perish.
 The Jordan Times, Cybercrime bill redrafted with new hate speech definition, reduced slander penalty.
 Edward McAllister, “Immunity to the Arab Spring? Fear, Fatigue and Fragmentation in Algeria,” New Middle Eastern Studies 3 (2013).
 See, e.g. Frederick Solt, “Economic Inequality and Democratic Political Engagement,” American Journal of Political Science 52, no. 1 (2008): 48–60.
 See, e.g. Sidney Verba, Kay Lehman Schlozman, and Henry E. Brady, Voice and Equality: Civic Voluntarism in American Politics (Harvard University Press, 1995).
 See, e.g. Solt, “Economic Inequality and Democratic Political Engagement.”
 See, e.g., Barrington Moore, Social Origins of Dictatorship and Democracy/Barrington Moore (London: Penguin, 1977); Samuel P. Huntington, The Third Wave: Democratization in the Late Twentieth Century, vol. 4 (University of Oklahoma press, 1993).