Reactive Proactive Aggression Questionnaire and its factorial invariance according to sex in Peruvian adolescents.

ABSTRACT

Introduction: in this research, we analyze the validity, reliability and invariance of the Reactive/Proactive Aggression Questionnaire. Method: the study was an instrumental sort of. We assessed 2803 adolescents, 48.9% female and 51.1% male from Arequipa City. The Reactive/Proactive Aggression Questionnaire (RPQ) was applied. Results: the test obtained adequate levels of goodness of fit and the bifactorial structure was confirmed by a confirmatory factor analysis. The test presented scalar and strict invariance based on the sex of the evaluated students, so men showed greater reactive and proactive aggressiveness than women. Reliability measures were satisfactory. Conclusion: the RPQ is valid with a two-factor structure; it is also invariant depending on the sex of those evaluated.

Key words: reactive aggression, proactive aggression, factorial invariance, latent means.

INTRODUCTION

Violence is, in its many manifestations, one of the most widespread health problems globally since it affects the physical and mental health of individuals. In Latin America, in addition to being one of the regions in the world with the highest level of violence, criminality and aggressive behaviors in teenagers between the age of 14 and 17, has increased.(1)

On an individual level, several biological factors intervene in the expression of aggressiveness. For example, hormones and neurotransmitters, or lesions in certain brain structures such as the amygdala and the prefrontal lobe(2). On a psychological level, low self-esteem, impulsiveness, personality traits and certain cognitive distortions are present in aggressive teenagers(3). These conditioning factors might appear since childhood. Growing up in violent family environments(4) and communities surrounded by crime and other illegal activities(5), results in the outlining of certain behavioral patterns that are shaped differently during

adolescence.(6)

Adolescence is a stage in life that involves profound biological and psychosocial changes. Teenagers face new challenges such as the definition of their personal identity and the construction of a life project, which can lead to experience certain existential crises, making them prone to a wide variety of risky behaviors, such as alcohol and drug use, unwanted pregnancies, crime, etc.(7) Through some local studies, it has been possible to assess various psychosocial aspects of adolescents, so that out of 3979 adolescents who have been involved in bullying others, 65% have been influenced by their peers(8). In this last group, a high percentage show signs of alterations in executive functions and show low indicators of social

cognition.(9)

It was also reported that life satisfaction, the ways in which adolescents between 14 and 18 years old cope with stress and feelings of despair are related to their meaning of life(10). Also, family structure and parental interactions were found to have an impact on antisocial behaviors(11), and more recently, through a sample of 1225 adolescents between 11 and 18 years old from Arequipa, it has shown to have an impact on

depression.(12)

On the other hand, among the psychological theories that explain aggressiveness, some of them focused on frustration as a cause and motivating component, while on the other hand, Bandura’s theory was based on the explanatory framework of social learning(13). More recently, Dodge and Coie(14) distinguished two types of aggressive behavior: reactive and proactive. Reactive aggression refers to aggressive behaviors that are responses to provocations or threats, while proactive aggression refers to deliberate and to planned aggression. In the first case, reactive aggression can be circumstantial, such as domestic violence or passion crimes(15), whereas proactive aggression involves criminal behaviors(16). Furthermore, reactive aggression is a type of impulsive or emotional aggression, while proactive aggression is a type of instrumental or rational aggression; but both are a part of an integrating model of

aggressiveness.(17)

Evidence suggests that reactive aggression is associated with a variety of negative emotions psychopathological symptomatology such as fear, irritability, hostility, depression and somatization(18); while, proactive aggressiveness is related to violent behavior, anxiety, impulsivity, distortion of reality and psychopathy(19). Studies carried out in Spanish-speaking adolescents indicate that reactive aggression and proactive aggression, or both, are associated with antisocial behavior(20); but it has also been reported that proactive aggression tends to decrease with age, that it is more frequent in males(21), and that it is associated with various forms of cognitive distortions.(22,23)

This theory of reactive and proactive aggressiveness has not been investigated in Peru, despite the fact that aggressiveness and violence have been recurring themes in national psychological research, especially related to violence against women(24), relationships within family members, aggressive behavior in adolescents(25) and variables related to violence at school(8). The absence of studies on reactive and proactive aggression in Peru is due to the lack of duly validated instruments that evaluate this variable within the corresponding theoretical model. Therefore, this research aims to analyze certain psychometric properties, such as validity, reliability, and invariance, of the Reactive/Proactive Aggression Questionnaire (RPQ), which was constructed by Raine et al. the 2006, who reported adequate indicators of construct validity and convergent validity, as well as satisfactory reliability indexes (?> .8), in adolescents from the United States.(19)

This test was validated for the Spanish-speaking population by Andreu, Peña and Ramírez in 2009(26) In this study, the instrument was applied to 732 adolescents from Madrid and two models underlying the confirmatory factor analysis performed were reported: a unidimensional one and a bifactorial one. The following year, Andreu(27) launched the standardized version with the corresponding rating scales. There is also a validated version of this test that is based on teachers’ statements to assess students’ aggressive behavior(28). This version has a bifactorial structure that explains 78% of the total variance of the test and presents high reliability indexes for each of its dimensions.

In Peru, in 2018, the Reactive/Proactive Aggression Questionnaire was validated in a sample of 822 adolescents from Huaraz, a city located in the department of Ancash in the northern highlands of Peru(29). The author reported a two-factor structure with adequate internal consistency indexes for the reactive aggression scale (?= 733) and the proactive aggression scale(?= .772). More recently, in 2021, Castañeda-Bernal et al.(30) have reported the psychometric properties of this instrument in a sample of 344 institutionalized and non-institutionalized adolescents from Lima. The results of this study show that the test has a bifactorial structure with good adjustment and reliability index, calculated using the internal consistency method and McDonald’s Omega test, of .797 and .837 for the two resulting factors: reactive aggression and proactive aggression.

METHODOLOGY

Participants

The sample consisted of a total of 2,803 high school students from the city of Arequipa. 48.9% were females between 13 and 19 years old (M= 15.72; SD= 0.817); and 51.1% males between 13 and 19 years old (M= 15.81; SD= 0.878). Participants were chosen was through an intact sampling of several public and private schools in the city of Arequipa, by convenience selection.

Instruments

The Reactive-Proactive Aggressiveness Questionnaire (RPQ), created by Raine et al was used(19). Its purpose is to evaluate the kinds of reactive and proactive aggression, and therefore, the motivation that each subject presents, according to the underlying theoretical model. This questionnaire contains 23 items, of which 11 asses reactive aggression and 12 asses proactive aggression. The response format is on a Likert-type scale: (0) never, (1) sometimes, and (2) often. The instrument can be managed individually or collectively.

Procedure

For the investigation purposes of this study, there was an adequate planning with the directors selected establishments. in order to obtain permits to apply the test and organize the evaluation schedules. Before the evaluation, a document explaining the characteristics of the research, was sent to the parents, so they could consent to their child to being submitted to the tests . In the same way, the students were informed of the objectives of the study, and voluntarily participated. The criteria to keep the anonymity and confidentiality of the participants have been met. The evaluation was carried out in a single session with an approximate duration of 15 minutes. It has also received the approval of the Ethics Committee of the National University of

San Agustín.

Data Analysis

First, a descriptive analysis of the items was performed and the compliance with the statistical assumptions of normality (univariate and multivariate) and absence of multicollinearity was evaluated. Taking into consideration that the RPQ does not have previous studies that clearly define the underlying factor structure in Peruvian samples, it was evaluated through a confirmatory factor analysis (CFA). To evaluate the invariance of proactive and reactive aggression evaluated by the RPQ between men and women, a Multiple Group Confirmatory Factor Analysis (MG-CFA) was applied using the R program version 3.6.1(31), specifically the packages lavaan version 0.6.5(32) and semTools version 0.5.2(33), while the graphs of the models were made through the package semPlot version 1.1.2(34). The robust weighted least squares estimation method with adjusted mean and variance (WLSMV) was used, due to the categorical nature of the variables under study(35). Theta parameterization was also used according to the recommendations of Muthen and Muthen(36). The analyzes were performed on the polychoric correlation matrix, which estimates the continuous variables underlying the items of an ordinal nature.(37)

To measure the factorial invariance of measurement, a progressive evaluation of three stages was carried out considering the configural invariance (the fit on both samples without adding any constraints to the model), scalar (invariance in factor loadings and thresholds; ?i and ?i) and strict (invariance in factor loadings, thresholds and error variances; ?i)(38,39). In the evaluation of configurational invariance, according to the criteria recommended by Mueller and Hancock(40) , the ?2 and the absolute indices RMSEA and SRMR were used. Values ? 0.06 are considered optimal. CFI and TLI were also used, for which values above 0.95 are considered

adequate.(41)

To evaluate the scalar and strict invariance, it was considered that variations of the CFI (?CFI) ? 0.01, RMSEA (?RMSEA) ? 0.015, SRMR (?SRMR) ? 0.03 are adequate to accept the invariance(42). Changes in ?2 were also taken into account; although this was done with caution as this indicator is highly sensitive to small deviations from a “perfect” model in large samples. In addition, the latent means were compared, since the scalar invariance is fulfilled. Finally, internal consistency was estimated by calculating McDonald’s ? and Cronbach’s ? coefficients.

RESULTS

Descriptive analysis

Table 1 shows that the majority of indicators (14 items) presented a distribution close to normality with values of asymmetry and univariate kurtosis less than ± 1.5. Multivariate normality was also evaluated using the Mardia index, obtaining a value of 769.73 for the total sample, which indicates that there is no multivariate normality. Likewise, the absence of multicollinearity was verified, as no values higher than r= .90 were observed between the items of the correlation matrix.

Confirmatory factor analysis

The bifactorial model of the RPQ was analyzed by means of a CFA, in which it was found that all the goodness-of-fit indexes were not adequate.[?2= 2721,91; p< 0,001; ?2/gl=11,88; CFI= 0,914; TLI= 0,905; RMSEA= 0,062; SRMR= 0,062]. The modification indexes indicated that errors had to be correlated between items 1 and 3, 1 and 5, 1 and 7, 5 and 7, 7 and 8, all of then belonging to the dimension of reactive aggression, which improved the values of fit indexes: ?2= 2023,33; p< 0,001; ?2/gl=9,03; CFI= 0,938; TLI= 0,930; RMSEA= 0,054; SRMR= 0,054.

Measurement Invariance analysis

First, the fit of the base model without restrictions was evaluated in men and women separately, finding that the factorial structure was similar in both groups, including correlated errors

(Figure 1); furthermore, the goodness-of-fit indexes were relatively similar (Table 2). After that, the configural invariance of the RPQ between the groups was analyzed, resulting in adequate values as a whole, ?2= 1978,16; p< 0,001; ?2/gl= 4,42; CFI= 0,945; TLI= 0,938; RMSEA= 0,049; SRMR= 0,056. It should be noted that this model will serve as a reference for a comparison with the restrictive models.

The next step was to analyze the scalar invariance, with restrictions on the factor loadings and thresholds, finding suitable fit indexes: CFI = 0,947; TLI= 0,943; RMSEA= 0,047; SRMR= 0,056. When comparing this model to the metric invariance model, it is observed that ?CFI= 0.002; ?RMSEA= -0.002 and ?SRMR= 0.000, values, show that the RPQ does present a scalar invariance, this indicates that both male and female respondents interpret the items in the same way. Finally, the strict invariance was analyzed, which had restrictions on factor loadings, thresholds, and residuals; and showed adequate fit indexes: CFI = 0,954; TLI= 0,953; RMSEA= 0,043; SRMR= 0,058. When contrasting the fit indices of this model with those of the scalar invariance model, it was found that ?CFI= 0.007; ?RMSEA= -0.004 and ?SRMR= 0.002; which shows that the RPQ presents strict invariance.

Comparison of latent means according to sex and reliability

The male sample showed a higher latent mean in the reactive aggression component, than the female sample (z= 2.936; p= 0.003), with a very small effect size, d= 0.11; Cohen’s U3= 0.544 and a probability of superiority of 0.531. This means that 54.4% of the male sample presents higher scores in reactive aggression than the average of females. In addition, given a random selection of students, there was a 53.1% probability that males would have a higher score than females in this factor.

In the case of proactive aggression, it was found that men presented a higher latent mean than women (z= 7.925; p< 0.001), with a small effect size, d= 0.29; Cohen’s U3= 0.614 and a probability of superiority of 0.581. This means that 61.4% of men have higher scores in proactive aggression than the average of women. In addition, given a random selection of students, there was a 58.1% probability that males would have a higher score than females in this factor.

The reliability of the test was measured using the internal consistency method, with both the reactive aggression factor (?= 0.807; ?= 0.805) and the proactive aggression factor (?= 0.848; ?= 0.837) in the general sample and showed adequate reliability indexes. Similarly, the reactive aggression factor (?= 0.798; ?= 0.795) and the proactive aggression factor (?= 0.823; ?= 0.804) presented adequate reliability in the sample of women. Equally, both the reactive aggression factor (?= 0.819; ?= 0.817) and the proactive aggression factor (?= 0.854; ?= 0.846) proved to be reliable in the male sample.

DISCUSSION

Aggressive behavior has been described in many ways, one of them being proposed by Dodge and Coie(14), who make a distinction between reactive and proactive aggressive behavior. Reactive aggressiveness is provoked and constitutes an emotional and impulsive response that overwhelms the person’s control and leads to aggression, as a consequence of experiencing negative emotions, such as mainly anger frustration(17). Proactive aggressiveness does not require provocation, but is associated with certain primary or secondary benefits, which the person subjectively perceives, such as power and control; it is therefore a rational and planned form of aggressive behavior.(18)

The efficacy of this theoretical model has been assessed in forensic, psychiatric, psychopathological, educational, social, and family contexts, with clear implications in the treatment and prevention of aggressiveness and violence. However, this presupposes having adequate instruments and understanding that aggressiveness is a variable that is subject to cultural variability(43) and to the sex of the people(6). In this sense, the present study aimed to analyze the validity and reliability, as well as the invariance according to sex in a sample of adolescents from the city of Arequipa, of the Reactive/Proactive Aggression Questionnaire, created by Raine et al.(19)

Thus, through various studies with Spanish-speaking samples from Spain(26,27), Uruguay(21) and Peru(29,30), the Reactive/Proactive Aggression Questionnaire has shown to have adequate psychometric properties, confirming the structure of two factors with satisfactory reliability indexes (?< .7) for each of them. In our study, the two-dimensional factorial structure only obtained adequate goodness-of-fit indicators when statistical corrections and modifications were made. Furthermore, the analysis of latent mean differences in both factors indicates that there are differences according to the sex of the subjects. Therefore, the levels of reactive and proactive aggressiveness are higher in men than in women, which is in line with previous studies on aggressiveness(2,6). In this sense, it should be noted that studies on the Reactive/Proactive Aggression Questionnaire have been preferably carried out in men(19,21), Therefore, there is little evidence on reactive and proactive aggressiveness in female adolescents with this instrument, and even less factorial invariance studies.

Due to the aforementioned, the present analysis becomes a substantial contribution to the understanding of aggressive behavior in Peruvian adolescents, but on the other hand, it is necessary to assess aggressive behavior relating it to other associated variables, such as alcohol consumption(44), emotional disorders(12) and family structure(11). It would also be convenient to assess several forms of invariance based on age, place of origin and other sociodemographic variables that would shed more light on the manifestation of aggressive behavior in adolescents, but in a differentiated way.

REFERENCES

1. 1. Latinobarómetro. Informe 2018. Santiago, Chile: Corporación Latinobarómetro; 2018.
2. 2. Gil JA, Pastor JF, De Paz F, Barbosa M, Macías JA, Maniega MA, González LR, Boget T, Picornell I. Psicobiología de las conductas agresivas. Anales de Psicología. 2002; 18(2): 293-303.
3. 3. Garaigordobil M, Durá A, Pérez JI. Síntomas psicopatológicos, problemas de conducta y autoconcepto-autoestima: Un estudio con adolescentes de 14 a 17 años. Anuario de Psicología Clínica y de la Salud. 2005; 1: 53-63.
4. 4. Quiroz N, Villatoro JA, Juárez F, Gutiérrez ML, Amador NG, Medina-Mora ME. La familia y el maltrato como factores de riesgo de conducta antisocial. Salud Mental. 2007; 30(4): 47-54.
5. 5. Herrera D, Morales H. Comportamiento antisocial durante la adolescencia: Teoría, investigación y programas de prevención. Rev. Psicol. (Lima). 2005; 23(2): 201-47.
6. 6. Garaigordobil M. Conducta antisocial durante la adolescencia: Correlatos socio-emocionales, predictores y diferencias de género. Psicología Conductual. 2005; 13(2), 197-215.
7. 7. Corcuera P, de Irala J, Osorio A, Rivera R. Estilos de vida de los adolescentes peruanos. Lima: Universidad de Piura; 2010.
8. 8. Bellido F, Rivera R, Salas JC, Bellido V, Peña N, Villasante G, Casapía, Y. Influencia de los pares en la manifestación del bullying en estudiantes de secundaria en Arequipa Metropolitana. Interacciones. 2016; 2(1), 33-42.
9. 9. Rivera R. Funciones ejecutivas y cognición social en adolescentes agresores, víctimas y espectadores en contexto de bullying. Rev. Psicol. (Arequipa Univ. Catól. San Pablo). 2018; 8(1): 41-66.
10. 10. Huamani JC, Arias WL, Núñez AL. Predictive model of purpose in life in adolescents of public educational institutions from Arequipa City. Panam. J Neuropsychol. 2018; 12(2). doi: 10.7714/CNPS/12.2.209
11. 11. Rivera R, Cahuana M. Influencia de la familia sobre las conductas antisociales en adolescentes de Arequipa-Perú. Actualidades en Psicología. 2016; 30(120), 85-97. doi: 10.15517/ap.v30i120.18814
12. 12. Rivera R, Arias WL, & Cahuana M. Perfil familiar de adolescentes con sintomatología depresiva en la ciudad de Arequipa, Perú. Rev. Chil. Neuropsiquiat. 2018; 56(2): 117-126.
13. 13. Bandura A. Pensamiento y acción. Fundamentos sociales. Barcelona: Martínez-Roca; 1987.
14. 14. Dodge K, Coie JD. Social information processing factors in reactive and proactive aggression in children´s peer group. Journal of Personality and Social Psychology. 1987; 53: 1146-58.
15. 15. Fernández J, Echeburúa E. Trastornos de personalidad y psicopatía en hombres condenados por violencia grave contra la pareja. Psicothema. 2008, 20(2): 193-8.
16. 16. Andreu JM, Peña E. Escala de conducta antisocial y delictiva. Anales de Psicología. 2013; 29: 516-22.
17. 17. Andreu JM. Propuesta de un modelo integrador de la agresividad impulsiva y premeditada en función de sus bases motivacionales y socio-cognitivas. Psicopatología Clínica Legal y Forense 2009; 9: 85-98.
18. 18. Dodge K, Lochman JE, Harnish JD, Bates JE. Reactive and proactive aggressive in school children and psychiatrically impaired chronically assaultive youth. J. Abnorm Psychol. 1997; 106: 37-51.
19. 19. Raine A, Dodge K, Loeber R, Gatzke-Kopp L, Lynam D, Reynolds C, Stouthamer-Loeber M, Liu J. The Reactive-Proactive Aggression Questionnaire: Differential correlates of reactive and proactive aggression in adolescent boys. Aggress Behav. 2006; 32: 159-71.
20. 20. Penado M, Andreu JM, Peña E. Agresividad reactiva, proactiva y mixta: análisis de los factores de riesgo individual. Anuario de Psicología Jurídica. 2014; 24: 37-42.
21. 21. Fares N, Cabrera JM, Lozano F, Salas F, Ramírez JM. Agresión reactiva y proactiva en niños y adolescentes uruguayos. Psicopatología Clínica Legal y Forense. 2012; 12, 47-68.
22. 22. Roncero D, Andreu JM, Peña ME. Procesos cognitivos distorsionados en la conducta agresiva y antisocial en adolescentes. Anuario de Psicología Jurídica. 2016; 26, 88-101.
23. 23. Peña ME, Andreu JM. Distorsiones cognitivas: Una revisión sobre sus implicaciones en la conducta agresiva y antisocial. Psicopatología Clínica, Legal y Forense. 2012; 12: 85-99.
24. 24. Castro R, Cerellino LP, Rivera R. Risk factors of violence against women in Peru. J. Fam Viol. 2017; 32(8): 807-15. doi: 10.1007/s10896-017-9929-0
25. 25. Mayorga E, Ñiquen M. Satisfacción familiar y expresión de la cólera-hostilidad en adolescentes escolares que presentan conductas antisociales. Revista de Investigaciones Psicológicas. 2010; 1: 87-92.
26. 26. Andreu JM, Peña E, Ramírez JM. Cuestionario de agresión reactiva y proactiva: Un instrumento de medida de la agresión en adolescentes. Revista de Psicopatología y Psicología Clínica. 2009; 14(1): 37-49.
27. 27. Andreu JM. Cuestionario de Agresividad Impulsiva y Premeditada en Adolescentes (CAPI-A). Madrid: TEA Ediciones; 2010.
28. 28. Cosi S, Vigil-Colet A, Canals J. Desarrollo del cuestionario de agresividad proactiva/reactiva para profesores: estructura factorial y propiedades psicométricas. Psicothema. 2009; 21(1): 159-64.
29. 29. Abanto AL. Evidencias de validez del cuestionario de agresión reactiva y proactiva (RPQ) en adolescentes de Huaraz. (Tesis de pregrado). Escuela Profesional de Psicología. Universidad César Vallejo, Trujillo, Perú; 2018.
30. 30. Castañeda-Bernal R, Correa-Rojas JD, Malvaceda-Espinoza EL. Invarianza factorial del Reactive/Proactive Aggression Questionnaire (RPQ) en adolescentes limeños institucionalizados y no institucionalizados. Evaluar. 2021; 21(3): 35-48.
31. 31. R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria; 2019. URL: https://www.R-project.org/.
32. 32. Rosseel Y. lavaan: An R Package for Structural Equation Modeling. Journal of Statistical Software. 2012; 48(2): 1-36.
33. 33. Jorgensen TD, Pornprasertmanit S, Schoemann AM, Rosseel Y. semTools: Useful tools for structural equation modeling. R package version 0.5-2; 2019. URL: https://CRAN.R-project.org/package=semTools
34. 34. Epskamp S. semPlot: Path Diagrams and Visual Analysis of Various SEM Packages’ Output. R package version 1.1.2; 2019. URL: https://CRAN.R-project.org/package=semPlot
35. 35. Brown TA. Confirmatory factor analysis for applied research. 2nd ed. New York: Guilford Press; 2015.
36. 36. Muthén LK, & Muthén BO. Mplus User’s guide. 7th ed. Los Angeles, CA: Muthén & Muthén; 1998–2012.
37. 37. Lee SY, Poon WY, Bentler PM. A two-stage estimation of structural equation models with continuous and polytomous variables. British Journal of Mathematical and Statistical Psychology. 1995; 48(2), 339-358. doi: 10.1007/BF02294660
38. 38. Schroeders U, Wilhelm O. Equivalence of Reading and Listening Comprehension across Test Media. Educational and Psychological Measurement. 2011; 71(5), 849-869. doi: 10.1177/0013164410391468
39. 39. Mueller RO, Hancock GR. Best practices in structural equation modeling. En Osborne J. (Ed.) Best practices in quantitative methods (pp. 488-508). Thousand Oaks, USA: Sage. 2008.
40. 40. Hu L, Bentler PM. Cutoff criteria for fit indexes in covariance structural analysis: Conventional criteria versus new alternatives. Structural Equation Modeling. 1999; 6: 1-55.
41. 41. Chen FF. Sensitivity of goodness of fit indexes to lack of measurement invariance. Structural Equation Modeling. 2007; 14: 464-504.
42. 42. Putnick DL, Bornstein MH. Measurement invariance conventions and reporting: the state of the art and future directions for psychological research. Developmental Review. 2018; 41: 71-90.
43. 43. Torregrosa MS, Inglés C, García-Fernández JM, Ruíz-Esteban C, López-García KS, Zhou X. Diferencias en la conducta agresiva entre adolescentes españoles, chinos y mexicanos. European Journal of Education and Psychology. 2010; 3(2): 167-76.
44. 44. López KS, da Costa, ML. Conducta antisocial y consumo de alcohol en adolescentes escolares. Revista Latinoamericana de Enfermedades Mentales. 2008; 16(2): 1-7.