Capítulo 5. Resultados obtenidos
5.3 Conclusiones
The results of this dissertation should be considered in light of a number of limitations. There are at least six types of limitations that can affect the interpretation of the results.
First, as has been the case in several studies on gangs (Decker, 1996; Decker &
Lauritsen, 2002; Gordon, 2000; Miller, 2006; Pyrooz & Decker, 2011), the sample size is smaller than would be ideal. The fact that the current study only has 73 participants limits the statistical power of any multivariate analyses. The desistance logistic regression model, for example, only had 22 participants in the predicted category.
However, the impact of this limitation was mitigated by restricting the number of variables in each statistical model to the most relevant in each context (i.e., those variables identified in bivariate analyses as the most strongly associated with each dependent variable). It should be also noted that the study sample of 73 incarcerated gang members is the largest of its kind in the Canadian context.
Second, the findings of the dissertation are based on the retrospective, self-reported data of incarcerated youth gang members in British Columbia and had accuracy and generalizability issues. The sample cannot be considered as representative of gang
members in general, and the specificity of the British Columbia setting should be kept in mind when considering cross-national generalizations. In addition, the retrospective self-report technique can impact the recall of some information, the temporal association, and can be affected by the perceptions bias (Metts, Sprecher, & Cupach, 1991). For example, it can be hypothesized that when participants were asked to identify the motivations that led them to join gangs, the motivations they reported could have been more related to the ones that were keeping them in their gangs at the time of the interviews and actually different from the ones they had when they joined their gangs.
Another limitation is associated with the data collection process. For the Youth Group Activities questionnaire, only trained graduated students with gang research knowledge were allowed to conduct the interviews. The majority of the interviews (n = 60) were conducted by two main research assistants, including the author. It has to be acknowledged that each interviewer has his or her own style of interviewing and that this can affect the nature or quality of the details that participants divulge (especially for the open-ended questions). Note also that male interviewers were not allowed to interview female gang members as per policy of the youth custody center.
A fourth limitation is associated with the use of cross-sectional data to examine the turning points in gang membership, even though longitudinal data would have been best to address such issues. The findings were interpreted along causal models, but the research design did not allow for a causal design that would eliminate potential confounding factors. However, the retrospective data used in the study were organized around major events for which the sequence was known. For example, the participants were asked questions about pre-membership characteristics in order to predict the initiation event, which, by definition, occurs later. The sequential order was known for all of the key dependent variables (e.g., age of entry, gang initiation, gang criminal activities, and gang desistance). Thus it became possible to make some inferences by taking into consideration the temporal order for those main turning points.
The fifth limitation is related to the fact that participants were not asked to provide
to verify the internal validity of the information provided. As mentioned by Hagedorn and Devitt (1999), even when this information is known, it is not rare that different members of the same gangs report different information regarding gang composition and activities (see, also, Descormiers & Morselli, 2011). This discrepancy can be affected by length of membership, status of the individual within the gang, and gang organizational structure.
Some participants may not be aware of some information about their gangs, either because they have been too recently affiliated and/or they did not have the status to be informed of such details or simply because the gang structure deliberately keeps the the same gang is higher in the current study, but it is impossible to verify with the current data. The general impression of the interviewers is that a single gang in B.C. at the time of the study could have generated multiple members in the sample and that less than a handful of those could have been interviewed for this study. However, it remains an important limitation of the study.
The sixth limitation is associated with the measurement of some variables. First, the nature of the connection who introduced a participant to gangs is unknown. Though it can be hypothesized that those connections were individuals involved in the criminal milieu, it would have been very informative to have known if they were gang involved, members of other criminal organizations, or something else. Second, the format of the motivations variable was to allow the participant to mention more than one answer.
Consequently, it was impossible to determine what the most important motivation was for those who named more than one. Third, participants were asked to divulge whether they had recently desisted from their gangs, but they were not asked to specify when that happened. This also impacts the possibility of determining the duration of membership for all the participants of the study. Fourth, the dissertation did not include the individual’s involvement in criminal activities, but only reported the gang criminal activities, as the Youth Group Activities questionnaire was initially designed. Although it was not the main objective of the dissertation, this aspect limits the ability to empirically
test the selection and facilitation causal models as they were created to verify the influence of gang membership on the individuals’ propensity to commit crimes. Fifth, the initiation event and the desistance process were qualitatively examined by only one open-ended question in the questionnaire. Because this was exploratory, no other questions were developed to provide more specific details. Not having had the opportunity to probe participants further on these topics limits the breadth of information found on these themes. Sixth, participants were asked whether their gangs were in contact with other gangs, but they were not asked to specify the number of those connections, only the presence of. The number of different connections was only asked for interactions with other criminal organizations. Though the results of the study showed no statistically significant association, it can be hypothesized that a better measurement for criminal social capital would have been related to their interactions with other gangs (Descormiers & Morselli, 2011). Some of the ways in which this can be done in future studies are explored in the concluding chapter.