Exploring the Non-Spatial and Spatial Role of Locus of Control Among Heroin-Injecting Mexican American Men

Background and Purpose. Drug abuse is a critical public health issue linked to a broad range of adverse social and health consequences which disproportionately impact minority groups. Very little is still understood about the drug abuse behaviors of Hispanics, particularly the potential impacts of spatial or environmental factors. Locus of control (LOC), i.e. the extent to which an individual believes certain behaviors are contingent upon internal or external factors, strongly predicts drug abuse behaviors. However, it is not clear how its influence varies across different geographies or contexts. This paper examines the non-spatial and spatial effects of LOC on Mexican American heroin users and their perceived concerns about injecting heroin in the near future.

Methods. This study focused on a sample of heroin-injecting Mexican American men aged 45 and over, drawn from a larger research study based in Houston, TX. We used ArcGIS 10.0 software to geocode and map the home addresses of respondents for whom we had either an exact address or an approximate location based on the nearest cross streets (n=160 out of 227). Our LOC measure had six questions, asking respondents whether they believed that injecting drugs was contingent upon internal or external forces (summative scale; higher scores indicating greater internal LOC). Logistic regression models were used to assess the effects of LOC on two outcomes: the chances that participants would inject drugs in the next six months; and how worried they were about injecting drugs in the next six months. Logistic geographically weighted regression (GWR) models were conducted to explore how these effects varied spatially.

Results. Non-spatial results showed that participants with greater internal LOC were less likely to report a very high or medium chance of injecting drugs in the next six months (OR=0.22, p<0.05), after controlling for other sociodemographic measures and current vs. former drug user status. In addition, higher internal LOC also reduced the likelihood that the participants would be quite or extremely worried about injecting drugs in the next six months (OR=0.54, p<0.05). The GWR results confirmed that the associations between LOC and the two outcomes varied spatially: the local parameters ranged from -1.92 to -2.37 (median=-1.97) and -0.42 to -1.17 (median=-0.84), respectively. For both models, the local effects were largely significant (96-97% of p-values less than 0.05), which is consistent with the global results shown in the non-spatial models. The associations between LOC and the chances of injecting drugs were largest in the northwest parts of Houston. In contrast, the strong associations between LOC and worrying about injecting drugs were concentrated in the southern part of the city, including several neighborhoods where Hispanics represent 85% or more of the population.

Conclusions and Implications.  The results highlight the important role that geography and context play in how we understand and address individual drug abuse behavior. Contextual factors, such as the spatial accessibility of treatment programs, availability of drugs, presence of positive or negative social capital, social networks, and the ethnic or generational make-up of the communities, should be further explored.