Spanish Validation of the Code of the Street Scale Among Salvadoran High Risk Youth and University Students

Background:This study examined the factor structure of the Street Code Attitudinal Scale (SCAS) within two Salvadoran populations: High-risk and gang-involved youth (HRGIY) and university students.  It has been demonstrated that SCAS is a valid and reliable instrument used to measure an individual’s adherence to violence, known as Code of the Street (CotS; Stewart & Simons, 2006, 2010).  Within communities plagued by poverty, discrimination, and mistrust of law enforcement, CotS has been theorized as a set of unspoken rules that dictate how individuals earn and maintain respect through acts of violence (Anderson, 2009).  SCAS has yet to be validated in Spanish within the context of Latin American communities, specifically in countries with severe gang and drug-related problems.  As such, this study validated and compared factor structures of the SCA within gang members and university students from the same municipality in San Salvador.  

Methods:The nine-item SCAS was translated into Spanish using a multistep method recommended by Rothgeb, Willis, and Forsyth (2005).  Responses to the SCAS by males between the ages of 16 and 25 were retrieved from datasets associated with two larger studies, one involving HRGIY and the other involving university students.  Initial descriptive statistics and reliability analyses were conducted to examine the behavior of individual items.  Confirmatory factor analysis (CFA) was then used to explore the psychometric properties of the SCAS.  CFAs were run separately in the gang and university samples.  The model fitting in each was used in a multi-group CFA to test whether model fit was comparable in the two populations. Finally, criterion-related validity analyses of the reduced CSAS were conducted.

Results: Reliability analysis revealed that inclusion of the reverse coded item in the SCAS reduced the reliability of the overall scale in both populations and had low item-to-total correlation.  This item was left out of the subsequent CFAs. A model that used the remaining 8 items was found to have acceptable fit in both populations (University Sample: [χ²(18) = 37.280; RMSEA = 0.0492; SRMR = 0.0371; CFI = 0.983; and GFI =0.970]; Gang Member Sample: [χ²(18) = 61.841; RMSEA = 0.0882; SRMR = 0.0445; CFI = 0.974; and GFI =0.953]).  In the multi-group CFA, however, the model fit significantly worsened once the factor loadings were constrained to be invariant (Δχ²(8)=18.61; p<0.05), suggesting the factor structure of CotS varied between the two samples. Subsequent analyses identified potential sources of differences, and criterion-related validity was established separately for each group.

Implications:  This study supported that a modified version of the SCA Scale (Stewart & Simons, 2006, 2010) is valid and reliable for use within Spanish speaking communities.  Multi-group analysis revealed that the relationships between the latent variable CotS and certain individual items were inconsistent between HRGIY and university students, meaning further scale development is needed before comparisons of SCAS between demographic groups are made.  Still, the SCA can be used to advance the examination, within Latino communities, of the relationships across the multiple constructs and relationships that together make up the CotS theory.