Using the clustering method for a heterogeneous reading of Lebanese youth

ABSTRACT

How to better study youth in ways that can capture their complex subjectivities? While qualitative methodologies succeed rather well at unpacking youth’s holistic selfhoods, quantitative tools are often more rigid at apprehending their multifaceted lives. Some tools such as multiple correspondence analysis do allow for a more nuanced and thorough understanding, and are particularly opportune in contexts where the studied group is heterogeneous in terms of social class and sectarian origin – such as the Lebanese case. Building on these attempts at quantitatively measuring youth’s multidimensional attributes, this paper analyzes a survey of Lebanese youth conducted in 2015 within the framework of the Power2Youth (P2Y) study to generate an intricate reading of young people, using the k-means clustering method. We use this technique to generate five youth groups: i) potential migrants, ii) secular youth, iii) school-to-job youth, iv) conservative students, and v) maturing youth. The paper discusses how each cluster relates to politics and religiosity, as well as to views on women’s roles and rights, highlighting the high variability within and across clusters. We conclude with reflections on how the clustering method may be useful to furthering research agendas and quantitative methodologies examining youth’s attitudes to political change in post-colonial contexts.

KEYWORDS:

• Youth
• youth attitudes
• clustering
• quantitative methods
• political change
• Lebanon

Acknowledgments

This paper has benefited from the research assistance of Christophe Maroun. We thank three anonymous reviewers for their valuable comments that greatly enriched the paper.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. We would like to thank an anonymous reviewer for pointing out this alternative typology of political participation, developed by Ekman and Amnå (2012), which is certainly relevant to research further in al-Hirak context, similarly to how Onodera et al. (2018) did in their important article on forms of engagement among Arab Mediterranean youth. Indeed, Ekman and Amnå (2012, p. 292) prompt us to acknowledge less direct and ‘latent’ forms of participation in our studies of political participation and prospects for political change, which they conceptualize as ‘civic engagement’ and ‘social participation.’ Such forms include voluntary work, lifestyle-related politics, as well as activities related to personal interest and attentiveness to politics. However, such types of engagements are difficult to capture through the Power2Youth quantitative survey, which defined, perhaps too strictly, political participation through manifest and formal means (e.g. voting, membership, signing petitions, participating in protests, striking, etc.).

2. This section draws heavily on the findings of the working paper of the first work package of the P2Y study, authored by one of the authors (Harb, 2016).

3. For more information on the survey, which was also administered in Palestine, Morocco, Tunisia and Egypt, see N. Sika ‘Beyond the Impasse? Youth Agency in Times of Crisis,’ this issue. FAFO’s team supervised the survey administration with local partners across countries. FAFO’s researchers were an integral part of the P2Y team, who collectively designed the survey, over several iterations. As such, this minimized disconnections between the survey’s research design, and its ground implementation.

4. Because of the project’s framing both conceptually and methodologically, the survey was only administered to Lebanese youth, and as such excludes Syrian, Palestinian, and other non-Lebanese youth residing in the country. The authors are aware of the limitations this poses for a discussion of youth in Lebanon.

5. We realize there is a contradiction between the definition of youth put forward in the article, which states that youth is not only defined by age cohorts, and the survey’s definition. Evidently, this is the constraint of quantitative data collection, which requires closed-ended definitions of variables. We try to address this limitation by integrating several survey variables in our analysis of youth.

6. The questionnaire did not ask about household income but rather the overall economic situation of the household, which is a subjective assessment by the household itself or the respondent, i.e. this is a self-classification (as opposed to an ‘objective’ measure such as income).

7. Using the k-means clustering method, first random cluster centres are assigned. Each point is assigned to the cluster with the nearest cluster centre using Euclidean distance in the seven dimensions deployed here. Cluster centres are then re-assigned through multiple iterations to minimize the sum of distances from each observation to its closest cluster centre. A cluster centre therefore represents average characteristics of observations in its group – young people in this case (See Table A1 in appendix).

8. Participation in political action is an index variable made up of six variables that measure whether respondents have ever taken part in any of the following forms of political action: signing a petition, joining a boycott, attending peaceful demonstrations, joining strikes, online activism through Facebook or blogging or other means, or participation in election campaigns. For this composite variable, respondents were grouped into those who took part in any political action (one or more), and those who never took part in any (See Table A2 in appendix).

9. The survey questions mostly used a Likert scale (‘strongly agree’ to ‘strongly disagree,’ and “very likely to ‘not likely at all’). For religiosity, the scale included three choices: very religious, religious, or not religious. This may be slightly reductionist, but the P2Y team agreed this might make it clearer and easier to answer than with five. Perhaps for such a questionit might be even clearer and easier to answer with three choices than with five. For the question on migration, respondents a series of questions on how likely they were to travel abroad in the coming five years (three questions: one on each of the Middle East, Europe, and Rest of the World). These were followed by a series of questions about potential reasons for travel. We extracted persons who are somewhat likely or very likely to travel to any of the three regions for the purpose of work or permanent residence, and created a variable for willingness to migrate. If we were to re-design the questionnaire, perhaps it would serve us better to ask a direct question about the likelihood to migrate for work or permanent residence on a scale from 1 to 5. However, given the available data, we opted to extract useful and accurate information about young individuals’ willingness to migrate, who (as we are arguing here) seem to have rather distinct attributes.

10. Detailed statistics reveal that 93% of the conservative cluster agree that men and women should receive equal pay compared to 84% in the school-to-job cluster; 82% of the conservative students agree that men and women should have equal inheritance rights compared to 69% of the school-to-job youngsters; and 86% of the conservative students compared to 70% of the school-to-job cluster agree that women and men should have equal divorce rights.

11. Note that existing political parties fielded the following number of candidates: 63 by the Strong Lebanon bloc (Tayyar, Maronites), 35 by the Future Movement (Sunni), 26 by the Lebanese Forces (Maronites), 18 by Amal Movement (Shiites), 15 by Kataeb (Maronites), 14 by Hezbollah (Shii),and 7 by the Progressive Socialist Party (Druze).

12. The seat was won by Paula Yacoubian, who is a famous female media anchor, well known to the Lebanese public – revealing that people who voted for her, selected her mostly based on familiarity with her name and face.

13. An ongoing research study undertaken by the Lebanese Centre for Policy Studies and Harvard University in Lebanon and Tunisia explores if various youth categories share a commitment to national political community, are able to tolerate members of different groups, and investigates views on citizenship, and the role of the state.

Additional information

Funding

This work was supported by the European Commission Power2Youth Grant Agreement Number 612782.