The effects of social networks on tobacco use among high-school adolescents in Mexico

Efectos de redes sociales sobre el uso de tabaco en adolescentes de preparatoria en México

Guadalupe Ramírez-Ortiz, D CS^I; Ramiro Caballero-Hoyos, D CS^II; Guadalupe Ramírez-López, D C^III; Thomas W Valente, PhD^IV

^ICentro de Seguridad Social Manzanillo, Instituto Mexicano del Seguro Social. Colima, Mexico
^IIUnidad de Investigación en Epidemiología Clínica, Instituto Mexicano del Seguro Social. Colima, Mexico
^IIIUnidad de Investigación Epidemiológica y Servicios de Salud del Adolescente, Instituto Mexicano del Seguro Social. Jalisco, Mexico
^IVDepartment of Preventive Medicine, Keck School of Medicine, University of Southern California. USA

Corresponding author

ABSTRACT

OBJECTIVE: To identify the effect of centrality in social network positions on tobacco-use among high-school adolescents in Tonala, Jalisco, Mexico.
MATERIALS AND METHODS: Longitudinal sociometric social network data were collected among 486 high-school adolescents in 2003 and 399 in 2004. The survey included: social network components, smoking and sociodemographic characteristics. Social network measures of centrality were calculated and multivariate logistic regression was used.
RESULTS: Ever used tobacco (OR= 44.98), marginalized-low stratum (OR= 2.16) and in-degree (OR=1.10) predicted tobacco use. Out-degree (OR= 0 .89) and out-in-degree (OR= 0.90) protected against tobacco use.
CONCLUSION: Nominating more friends rather than receiving such nominations was protective for tobacco use. Popular students, those receiving many nominations, were at higher risk for tobacco use. Involvement of leaders with capacity to influence might be an efficient strategy for dissemination of preventive messages.

Keywords: social networks; tobacco use; adolescents; centrality; friendship; peer pressure

RESUMEN

OBJETIVO: Identificar el efecto de posiciones de centralidad de la red social sobre el uso de tabaco en adolescentes de preparatoria en Tonalá, Jalisco.
MATERIAL Y MÉTODOS: Estudio longitudinal de redes sociales sociométricas. Participaron 486 bachilleres (2003) y 399 (2004). La encuesta incluyó: componentes de redes sociales, tabaquismo y características sociodemográficas. Se calcularon medidas de centralidad de redes sociales y utilizó regresión logística multivariada.
RESULTADOS: El consumo alguna vez de tabaco (RM= 44.98), estrato socioeconómico marginado-bajo (RM= 2.16) y vínculos recibidos (RM=1.10) predijeron el tabaquismo; mientras que los vínculos enviados (RM= 0.89) y la diferencia entre vínculos enviados y recibidos (RM= 0.90) protegieron contra el tabaquismo.
CONCLUSIÓN: Nombrar más amigos que ser nombrado por otros protegió contra el tabaquismo. Los estudiantes populares, aquellos con muchos nombramientos, tuvieron mayor riesgo de ser consumidores. La inclusión de líderes con influencia podría ser una estrategia eficiente en la diseminación de mensajes preventivos.

Palabras clave: redes sociales; uso de tabaco; adolescentes; centralidad; amistad; presión de pares

An important factor associated with adolescent's smoking is having friends that smoke.⁶ To understand factors associated with adolescent tobacco use, socialnetwork methods were used to measure an adolescent's position in the social network. Social network methods can be used to measure social relations and interactions that influence tobacco use.^7-9 Also, it is a useful guide for the development of interventions for prevention and treatment of addictions.^10-12 The social network model is based on relationship systems and communications. The basic data for analysis are the links between nodes (or actors). Studies of social network analysis on smoking have been analyzed from two perspectives: 1) social influence from others derived from group interaction¹³ or sociometric positions¹⁴ and 2) actor centrality.^15,16 Social influence studies have repeatedly documented that being exposed to smokers increases smoking risk. In addition, studies have shown that tobacco use was higher among students having links with group members, being group members, liaisons,^17,18 dyads,¹⁴ or even being isolated.^13,14,19 In one notable 32-year cohort study, smokers were increasingly moved to the periphery of the social network, whereas non-smokers moved to the center.²⁰

Perhaps the most common indicator extracted from social network data is centrality. The term "centrality" is restricted to the idea of "central actor". It indicates positions in which actors occupy a prominent place or strategic position in the network.²¹ Central people may have a greater influence on the opinions and behaviors of others and at the same time may be influenced by others in the network.²² In this regard, central people can induce the persuasive influence of other peers by signaling cultural acceptability for the behavior.¹⁰

Centrality, measured as the frequency a person was named as a peer, has been associated with substance use. Valente²³ reviewed studies on the effect of school-based social networks on substance use and found that use is the result of the interaction among peers and their degree of centrality.²⁴ In this interaction, peer influence²⁴ and the normative effect of substance-using friends and close relatives were important.²⁵ Another network study considered two different definitions of centrality: popularity, or the number of friendship nominations received from others (in-degree) and expansiveness, or the number of friendship nominations sent to others (out-degree).¹⁵ The study showed that popular people were more likely to be and to become smokers.¹⁶ Some studies have found an effect of centrality measurements (indegree, betweenness and Bonachich power centrality) on psychoactive drugs use.²⁶

Whether the effect of centrality measurements such as out-in-degree and out-in-closeness may explain this relationship has not been evaluated. The aim of this study was to identify the effect of central positions (inand out-degree, inand out-closeness, out-in-degree and out-in-closeness) in social networks on current tobacco use among students of a high-school. The findings will provide information for planning strategies for the prevention of tobacco use.

Material and Methods

Study design and sample: A longitudinal study was conducted in order to collect sociometric social network data²¹ in one high-school in Tonalá, Jalisco, Mexico. Tonalá is located in the Guadalajara metropolitan area but some semi-urban traits persist and the lowest strata of the social pyramid predominate. Tonalá High-school had a student body of 2,650 students in 2003, and 2,702 in 2004, from first to sixth semester. From June to July 2003, a total of 490 students from first and second semester were invited to participate in the study. Of them, 486 (99.2%) accepted and 399 were followed-up in 2004.

Procedures: School officials and students gave their written consent. The project was approved by the Local Health Research Committee at the Mexican Social Security Institute.

At the time frame of the study, school regulations prohibited smoking

Variables: Tobacco use was measured with the following questions: have you ever smoked? (yes/no) and do you smoke currently? (yes/no).

Social network data were collected by asking for the name and sex of each person's six best friends in the school, and communication frequency according to the Pearson and Michell¹³ format. Social network indicators included:

Density (D): number of links in the total networks, expressed as a proportion of the maximum number of possible relationships within the networks. Density formula is l/n (n-1)/2 where n is the number of nodes (students) and l the number of lines present.²¹

Subgroup density: proportion of connections between actors of an asymmetric valued matrix that share an attribute.²⁷

Centrality measurements proposed by Freeman²⁸ and Valente²⁹ were calculated and included in-degree, out-degree, in-closeness and out-closeness. Also, two variables of difference were generated: out-in-degree (difference between out-degree and in-degree) and out-in-closeness (difference between out-closeness and in-closeness). Measurement's definitions are described in Table I.

Peer pressure was defined as the subjective experience of feeling encouraged by people of one's own age to do certain things regardless of whether one wants to do them.³⁰ The measure included 11 items with a 5-point scale ranging from "strongly disagree" to "strongly agree".

Socioeconomic stratum was evaluated according to Basic Geostatistical Areas

Statistical Analysis: Social network structure analysis was performed using NetMiner II 2.4.0.¹ Chi-square tests were calculated to evaluate prevalence changes and logistic regression to evaluate associations. The dependent variable was current tobacco use in 2004 and the independent variables were the centrality measurements in 2003 (in-degree, out-degree, in-closeness, out-closeness, out-in-degree and out-in-closeness), peer pressure in 2003, ever tobacco use, occupation, age in 2003, socioeconomic stratum, and sex. A HosmerLemeshow test was used to evaluate goodness-of-fit. Multicollinearity was not observed (variance inflation factor < 2.0 in both models). Statistical analyses were performed with SPSS 15.0 and Stata 9.0.

At baseline (2003), 486 (of 490, 99.2%) freshmen highschool students participated in the study. The baseline refusal rate was less than 1.0%. After one-year (2004), 399 (of 486, 82.1%) students were followedup. Attrition was due to students exclusion due to failing grades or school absenteeism (78 or 89.6%), and voluntary withdrawal from school (9 or 10.3%).

There were no attrition differences by sex and socioeconomic level in the follow-up; however, greater attrition was observed in students working and studying (31.0%) in comparison with those that remained in the study (21.2%, p= 0.036). There was greater baseline ever tobacco use (63.5%) among those lost to follow-up than those who remained in the study (46.4%, p= 0.003). In-degree of those who withdrew from the study was lower (7.29), than among those who remained (8.60, p= 0.025).

The mean age at baseline was 15.7 years (15-19 years), whereas for the follow-up, it was 16.6 years (15-19 years). The increase of students working and studying was 15.1% (p < 0.010).

Current tobacco use was 9.5% at baseline and 13.8% at follow-up, an increase of 4.3 percentage points (p= 0.029). Tobacco ever use was 49.4% at baseline, 49.9% at follow-up (0.5 percentage points increase, not statistically significant).

Changes in current tobacco use by socio-demographic variables during follow-up are shown in Table II.

Overall network density means increased from 0.011 + 0.14 to 0.015 + 0.16 (in 2004), indicating that 1.1% of all possible connections among network members existed during the first year, and 1.5% in the second. Table III reports sub-group density rates in 2003 and 2004 indicating that densities were greater within ho- mogeneous groups (smokers or nonsmokers only) than among heterogeneous groups (smokers-nonsmokers and nonsmokers-smokers). The differences of density among smokers and nonsmokers in both years were statistically significant.

Peer pressure was higher among smokers both years. At one year of follow-up there was a reduction of 5.37 among non-smokers (p< 0.001). Smokers had a lower out-degree in comparison with non-smokers at baseline, however at year two it was the opposite (p= 0.014) and an increase of 0.63 was found among non- smokers (p= 0.013). Smokers had a higher in-degree in comparison with non-smokers at baseline; this was the opposite in the second year (p= 0.018) with a 0.78 among non-smokers (p= 0.013).

Non-smokers had a higher out-closeness in both measurements; only in 2003 there was a difference (p < 0.001). At follow-up there was an increase of 2.57 among smokers (p= 0.021). In-closeness was higher in non-smokers in both measurements with an increase of 2.46 at follow-up (p= 0.001). At baseline the out-indegree was lower among smokers (p= 0.005). However, at follow-up it was higher, with an increase of 3.96 (p=0.011). At both times non-smokers had higher out-in-closeness with a reduction of 1.96 in average (p= 0.016) (Table IV).

Models of centrality positions associated with current tobacco use are shown in Table V. In the first model, in-degree, out-degree, in-closeness and outcloseness were evaluated. Tobacco ever use in 2003 and in-degree were associated with current tobacco use in 2004 (OR= 42.55 and OR= 1.10, respectively). In contrast, out-degree was a protective factor for current tobacco use in 2004 (OR= 0.89). In the second model, out-in-degree and out-in-closeness were evaluated. Variables associated with tobacco use in 2004 were: tobacco ever use in 2003 (OR= 44.98), marginalized-low stratum (OR= 2.16) and out-in-degree (OR = 0.90).

Discussion

Having more nominations to peers (out-in-degree) rather than receipt of these nominations was a protective factor for tobacco use. In addition, simply as naming a high number peers (out-degree) was also protective for tobacco use. On the other hand, being named by peers (in-degree) was a risk factor to becoming a smoker. The protective effects of naming peers indicates that being integrated into school-based friendship networks at this developmental stage has benefits for avoiding risky behavior. Conversely, the positive in-degree association indicates that popular students have a greater probability of becoming smokers. This indicates that current tobacco use is popular; therefore, widespread use is expected in this student social network in the future.

Results for out-in-degree related to tobacco consumption have not been reported before, and indicate that the difference between naming friends and being named may be an important indicator for social position that has an influence on risk behavior.

Our results about in-degree related to tobacco use are similar to findings reported by others.^15,16 Valente and others found that students who were popular in the sixth and seventh grades in the US had greater probability of becoming smokers than less popular ones. Moreover, popular students at schools with high tobacco-use prevalence were more at risk to smoke; this indicates that tobacco use can be attributable to the students' position in the network structure to the extent that position indicates the person's power and susceptibility of being influenced,^15,16 or the process of selecting peers with attributes similar to oneself.³¹ In this regard, a popular person is one who receives connections or friendship nominations and is a vertex of high in-degree.^32,33 In a cohort study, smokers moved to the network periphery perhaps due to the advent of public health campaigns against tobacco consumption and rise in anti-tobacco norms.³⁴ Using networks sociometric positions, smoking was higher among dyads and isolates and it was lower among highercategories of popularity. Maybe, in these instances, cohesive groups applied peer pressure in the opposite direction to enforce non-smoking behavior.¹⁴

Our results about marginalized-low stratum related to tobacco use are similar to others. These studies demonstrated differential tobacco consumption according to socioeconomic strata, with a significant association with the consuming peer 's normative influence.^35,36

Subgroup density results show that there are subgroups of smokers and nonsmokers within the student network, allowing for the hypothesis that greater cohesion among current tobacco consumers over time suggests that dense social ties can reinforce the use norm over time.²⁶ This hypothesis also suggests an interaction context in which mutual influence may occur that favors use⁶ leading to the formation of subcultures in which tobacco use is a part of their identity. This may influence adolescents in the group to have access to cigarettes, to approve use, and to have mutual emotional support, not unlike that which occurs with other substances.^26,37 The subculture may also contribute to the development of other risk behaviors.³⁸

Ever tobacco use predicted a greater risk of current tobacco use. Ever users were more likely to be lost to followup. It is possible that a higher current tobacco use in the follow-up measurement might have increased the associations we find between current tobacco use and centrality since in-degree was also associated with loss to follow-up. Current tobacco-use prevalence in this study was greater than national prevalence^2,3 and less than current-use prevalence among Mexico City highschool students^4,5 which can be explained to the fact that use is greater among youth in contexts of greater urban development.

Limitations: attrition was caused mainly by student dropouts, which was not possible to control. Also, it was not practical to follow-up adolescents who did not remain in the study, since they were no longer exposed to the student network. Attrition in the follow-up is accompanied by differences in in-degree and tobaccoever-used participants who remained in the study and those who did not, which may cause a selection bias.³⁹ Also, smoking, in our study, was defined as current tobacco use, while others consider it as smoking at least one cigarette every day in the past 30 days. In our case, as stated by others, we considered that any use is abuse.⁴⁰

The findings of this study describe the formal student network structure that could be complemented by exploring the possible influence of networks outside the school such as the family,^41,42 and neighborhood friends.^34,43,44 It would also be worthwhile to analyze the formation of subgroups by use patterns: light smoker, moderate smoker, and heavy smoker.²

Although our results show that popular students have a higher probability of smoking, this position has been used in interventions to reduce tobacco consumption. Opinion leaders are selected based on in-degree position because they have a prominent position in social networks structure, and may influence towards healthy behaviors.²⁹ In these interventions opinion leaders are trained to direct educative interventions at the informal interactions with their peers.^10-12 Also, recently approved Mexican regulations restrict tobacco use in public places, and favor non-smoking promotion in schools, which we hope will contribute to a more effective control.

Acknowledgments

This study was supported by grant IMSS-FP-2003/148. We thank authorities, students and teachers at Tonalá High-school of University of Guadalajara.

References

1. Kuri-Morales PA, González-Roldán FJ, Hoy MJ, Cortés-Ramírez M. Epidemiología del tabaquismo en México. Salud Publica Mex 2006;48:91-98.         [ Links ]

2. Secretaría de Salud. Instituto Nacional de Estadística Geografía e Informática (SS/INEGI). Encuesta Nacional de Adicciones 2002. Aguascalientes: SS/INEGI, 2004.         [ Links ]

3. Secretaría de Salud. Instituto Nacional de Salud Pública (SS/INSP). Encuesta Nacional de Adicciones 2008. Cuernavaca, Morelos: SS/INSP, 2009.         [ Links ]

4. Villatoro J, Medina-Mora ME, Fleiz C, Amador NG, Bérmdez P. La encuesta de estudiantes de enseñanza media y media superior de la Ciudad de México: Noviembre 2003. Prevalencias y evolución del consumo de drogas. Salud Mental 2005;28:38-51.         [ Links ]

5. Villatoro J, Gutiérrez M, Nieves M. Encuesta de estudiantes de la Ciudad de México 2006. Prevalencias y evolución del consumo de drogas. Salud Mental 2009;32:287-297.         [ Links ]

6. Center for Disease Control and Prevention. Youth tobacco surveillance United States 2000. MMWR Morb Mortal Wkly Rep 2001; 50: 1-84.         [ Links ]

7. Hoffman BR, Sussman S, Unger JB, Valente WT. Peer Influences on Adolescent Cigarette Smoking: A Theoretical Review of the Literature. Substance Use & Misuse 2006;41:103-155.         [ Links ]

8. Luke DA, Harris JK. Network Analysis in Public Health: History, Methods, and Applications. Annu Rev Public Health 2007;28:69-93.         [ Links ]

9. Smith KP, Cristakis NA. Social networks and health. Annu Rev Sociol 2008; 34:405-429.         [ Links ]

10. Valente TW, Hoffman BR, Ritt-Olson A, Lichtman K, Johnson A. Effects of a social networks method for group assignment strategies on peer-led tobacco prevention programs in schools. Am J Public Health 2003;93:1837-1843.         [ Links ]

11. Audrey S, Cordall K, Moore L, Cohen D, Campbell R. The development and implementation of a peer-led intervention to prevent smoking among secondary school students using their established social networks. Health Educ J 2004;63:266-284.         [ Links ]

12. Campbell R, Starker F, Holliday J, Audrey S, Bloor M, Parry-Landgon N, et al. An informal school-based peer-led intervention for smoking prevention in adolescence (ASSIST): a cluster randomized trial. Lancet 2008;371:1595-1602.         [ Links ]

13. Pearson M, Michelle L. Smoke rings: social networks analysis of friends groups, smoking and drug-taking. Drugs Educ Prev 2000;7:21-37.         [ Links ]

14. Pearson M, Sweeting H, West P, Young R, Gordon J, Turner K. Adolescent substance use in different social and peer contexts: A social network analysis. Drugs: education, prevention and policy 2006;13:519-536.         [ Links ]

15. Valente TW, Unger JB, Johnson A. Do popular student smoke? The association between popularity and smoking among middle school students. J Adolesc Health 2005;37:323-329.         [ Links ]

16. Alexander C, Piazza M, Mekos D, Valente T. Peer, school, and adolescent cigarette smoking. J Adolesc Health 2001;29:23-30.         [ Links ]

17. Fang X, Li X, Stanton B, Dong Q. Social networks position and smoking experimentations among Chinese adolescents. Am J Health Behav 2003;27:257-267.         [ Links ]

18. Henry DV, Kobus K. Early adolescent social networks and substance use. J Early Adolescence 2007;23:346-362.         [ Links ]

19. Ennett ST, Bauman KE. Peer group structure and adolescent cigarette smoking: A social networks analysis. J Health Soc Behav 1993;34:226-236.         [ Links ]

20. Christakis NA, Fowler JH. The collective dynamics of smoking in a large social network. N Engl J Med 2008;358:2249-2258.         [ Links ]

21. Scott J, ed. Social networks analysis. A Handbook. Second ed. London: Sage Publications, 2000.         [ Links ]

22. Valente TW, Davis RL. Accelerating the diffusion of innovations using opinion leaders. Ann Am Acad Polit Soc Sci 1999;566:55-67.         [ Links ]

23. Valente TW. Social networks influences on adolescent substance use: An introduction. Connections 2003;25:11-16.         [ Links ]

24. Malcon Mc, Menezes B, Maia S, Chatkin M, Victora CG. Prevalência e factore de risco para tabagismo em adolescentes na América do Soul: uma revisão sistemática da literatura. Pan Am J Public Health 2003;13:222-228.         [ Links ]

25. Urquieta J, Hernández-Ávila M, Hernández B. El consumo de tabaco y alcohol en jóvenes de zonas urbanas marginadas de México. Un análisis de decisiones relacionadas. Salud Publica Mex 2006;48:30-40.         [ Links ]

26. Ennett ST, Bauman KE, Hussong A, Fairs R, Foshee VA, Cai L. The peer context of adolescent substance use: Findings from social networks analysis. J Res Adolesc 2006;16:159-186.         [ Links ]

27. McPherson M, Smith-Lovin L, Cook JM. Birds of a feather: Homophily in social networks. Annu Rev Sociol 2001;27:415-444.         [ Links ]

28. Freeman L. Centrality in social networks. Conceptual clarification. Social Networks 1979;1:215-239.         [ Links ]

29. Valente TW. Social networks and health: Models, methods and applications. New York: Oxford University Press, 2010.         [ Links ]

30. Santor DA, Messervey D, Kusumarca V. Measuring peer pressure, popularity, and conformity in adolescent boys and girls: predicting school performance, sexual attitudes, and substance abuse. J Youth Adolescence 2000;29:163-182.         [ Links ]

31. Hall JA, Valente TW. Adolescent smoking networks; the effects of influence and selection on future smoking. Addict Behav 2007;32:3054-3059.         [ Links ]

32. Jansson I. Popularity structure in friendship networks. Soc Networks 1999;21:339-359.         [ Links ]

33. Dingues MM, Oetting ER. Similarity in drug use patterns between adolescents and their friends. Adolescence 1993; 8:253-266.         [ Links ]

34. Mercken L, Snijders TAB, Steglich C, de Vries H. Dynamics of adolescent friendship networks and smoking behavior: Social network analyses in six European countries. Social Science & Medicine 2009;69:1506-1514.         [ Links ]

35. Caballero R, Madrigal E, Hidalgo A, Villaseñor A. El consumo de tabaco, alcohol y drogas ilegales, en los adolescentes de diferentes estratos socioeconómicos de Guadalajara. Salud Ment 1999; 22(4):1-8.         [ Links ]

36. Mathur Ch, Stigler MH, Cheryl L, Perry, Arora M, Reddy KS. Differ ences in prevalence of tobacco use among Indian urban youth: The role of socioeconomic status. Nicotine Tob Res 2008; 10(1):109-116.         [ Links ]

37. Pearson M, West P. Drifting smoke rings: Social networks analysis and Markov Processes in a longitudinal study of friendship groups and risktaking. Connections 2003; 25:56-76.         [ Links ]

38. Kelsey YL, Whittemore AS, Evens AS, Thompson WD, ed. Methods in observational epidemiology. New York: Oxford University Press, 1996.         [ Links ]

39. Kirke DM. Chain reactions in adolescents' cigarette, alcohol and drug use: Similarity through peer influence or the patterning of ties in peer networks? Soc Networks 2004;26:3-28.         [ Links ]

40. Medina-Mora ME. Los conceptos de uso, abuso, dependencia y su medición. En: Tapia-Conyer R, ed. Las adicciones:dimensión, impacto y perspectivas. México: El Manual Moderno, 2001:21-44.         [ Links ]

41. Leatherdale ST, McDonald PW, Cameron R, Brown KS. A multilevel analysis examining the relationship between social influences for smoking and smoking onset. Am J Health Behav 2005;29:520-530.         [ Links ]

42. Kaduri P, Gilreath T, Mbwambo J, Kilonzon G, Flisher AJ, Matthews SA. Social networks' influence on tobacco use among students in Dar Es Salaam, Tanzania. Promot Educ 2005;12:66-70.         [ Links ]

43. Chuang YC, Ennett ST, Bauman KE, Foshee VA. Neighborhood influences on adolescent cigarette and school use: Mediating effects through parent and peer behaviors. J Health Soc Behav 2005;46:187-204.         [ Links ]

44. Friedkin NE. Norm formation in social influence networks. Soc Networks 2001;23:167-189.         [ Links ]

Corresponding author:
Dra. María Guadalupe Ramírez-Ortiz.
Centro de Seguridad Social Manzanillo, Instituto Mexicano del Seguro Social.
Av. Tonalá 121. 45400 Tonalá, Jalisco, México.
E-mail: ishaoro@hotmail.com

Received on: September 19, 2011
Accepted on: February 1, 2012

Declaration of conflict of interests: The authors declare that they have no conflict of interests.
1 Cyram NetMiner II: Software for network analysis 2.4.0. Seoul: Cyram Co, Ltd., 2004.