Analyses of Google internet searches, reveal deep and otherwise elusive sentiments that are all too frequently unobserved by traditional research methods. Internet users often search topics they would be unwilling to ask their closest and most intimate confidants about. Thoughts about suicide, depression and the means one might use to take one’s own life are prominent examples – people may keep these thoughts very private but reveal them to google. Spring boarding from this central insight this paper seeks to determine whether suicide rates are higher in places where google searches about depression and suicide are common. Building on prior research, this project will assess the extent to which Google searches predict the 50-state suicide rate above and beyond traditional approaches. Using factor analysis three clusters of search terms were identified: suicide, depression, and gun-related. Suicidal and depressive search terms are found to be significantly associated with the suicide rate over the 11-year period (2006- 2016) even after controlling for divorce, unemployment, mental state, binge drinking, gun ownership, and yearly effects. These findings suggest that internet search data represent a, practical, and cost-effective method for studying suicide and other large-scale sociological phenomena. As fields such as public health, epidemiology, and psychiatry effectively incorporate internet and big data-driven methodologies, it is imperative that sociologists carefully consider its capabilities given its increasing utilization and potential.

Studies on the effects of ethnic diversity in neighborhoods and cross-ethnic contact on individual health outcomes have had mixed results regarding whether this effect is positive or negative. The Coalitional Safety Index (CSI) model (Boyer et al. 2015) proposes that these inconsistent results are related to differences in the perceived level of safety and its regulatory function to the stress sensitivity during cross-ethnic contact. Using the American quota samples of the Coalitional Psychology Survey (Boyer and Firat 2015), I found that the level of in-group cohesion, one of the two factors for judging the perceived safety level, moderated the direction of the moderation effect of cross-ethnic contact on the association between neighborhood diversity and health outcomes in the following way: low levels of in-group cohesion lead to negative health outcomes, while high levels of in-group cohesion lead to positive health outcomes. The perceived level of out-group threat, another factor of CSI, did not show this significant effect. These results suggest that, despite some limitation, the basic concept of CSI model (Boyer et al. 2015) may provide an integrated framework encompassing macro-scale (neighborhood diversity), meso-scale (cross-ethnic contact), and micro-scale (individuals' cognitive processes involving the coalitional sense of safety).

Suicide and fatal drug overdose are a major public health and social problem in the US. Google search term data may help researchers understand variations in area-level suicide and fatal drug overdose rates. The extent to which an area Googles about suicide and drug-related topics may be indicative of the overall social environment in the area. This project tests if there is an association between suicidal Google searches and the suicide rate, and also tests if there is an association between drug-related Google searches and the drug overdose rate. This project tests these associations using two-way fixed effects linear regression to control for unmeasured confounding, while controlling for the major predictors of each outcome. The unit of analysis are Designated Market Areas (n=209) and the analyses observe a seven-year period (2010-2016) for a total of 1,463 unique observations. Results show that suicidal Google searches are associated with suicide rates, and drug-related Google searches are associated with overdose rates. These findings demonstrate that Google searches may provide researchers with an additional lens into the innermost thoughts, attitudes, and behaviors of individuals at an area level. These findings suggest that Google search data can assist in real-time suicide and drug overdose monitoring and prevention.

A critical problem facing the health of the U.S. is the opioid epidemic which has been unfairly addressed due to racism and the divide of color. Prescription opioid use has seemingly affected different demographics at different periods and rates, with growing racial disparities. Much evidence suggests a discrepancy in the government and public response to this drug epidemic in comparison to the criminalization of the drug epidemic of the 1980s, having to do with the race/ethnicity of a person using drugs, with communities of color facing incarceration and Whites facing recovery services. A strong theoretical tradition in medical sociology says that health outcomes are built on social factors, specifically racialized social systems that reinforce white privilege through social relations and practices. Data from the State of California (CA), over much of the period of the epidemic (2008-2018), was used to test whether this happened for opioid prescriptions and deaths in CA counties(n=58). To investigate some of the dynamics of this growing problem, I formulate and empirically test hypotheses derived from fundamental cause theory (FCT) as a contributing explanation for opioid prescription and overdose (OD) death trends in California. This theory proposes that disparities by socioeconomic status (SES) and race-ethnicity persist or are reproduced because access to treatments and interventions are unequally distributed, flowing more freely to those with greater access to flexible resources of knowledge, money, and power, prestige, and beneficial social connections. Multiple regression was run to investigate the extent to which counties that were predominately populated by Whites, had greater access to opioid prescriptions over time. An investigation of OD death trends in CA counties overtime was also investigated. Results were consistent with FCT, involving resources that determined the extent to which whites were able to avoid multiple disease outcomes involving the opioid epidemic.

Social epidemiologists contend that economic inequality is an important driver of population health. However, results from empirical research that examines the link between income inequality and population health are equivocal. In this dissertation, I revisit the income inequality-health debate. I develop a model that envisions income inequality as a key component of economic stratification that intersects with positional factors, like socioeconomic status, to influence population health. Drawing empirical evidence from both cross-national, comparative and within-U.S. data, I calculate multiple-imputation fixed effects, Prais-Winston, and hybrid panel estimators to examine who pays income inequality’s health tax at the global, country, and individual level. Findings from the three studies indicate that several factors, including economic development, political exclusion of non-elites, educational attainment, gender, and early-life poverty, interact with income inequality to produce health disparities. The results suggest that income inequality does not act as a sort of pollution from which no one can escape but as a feature of economic stratification reflective of unequal distribution of resources that has a differential impact on health for vulnerable and privileged groups at multiple levels of analysis.

A disturbingly high number of adolescents experience mental illness, with many delaying seeking help for years. In response, public health inspired-help-seeking model and interventions argue that mental health problems should be appropriately recognized and labeled using psychiatric terminology. Accordingly, labels enable adolescents to understand that their symptoms are attributable to an authentic condition that can be treated. This is seen as an empowering and hopeful experience that promotes help-seeking. Adolescents often self-label, or cognitively self-identify as someone who has a “mental illness” or a “mental health problem.” They also receive parental labels, which occur when their parents similarly identify them. These labels do not always co-occur; rather, it is typical for adolescents to self-label in the absence of a parental label (discordant self-label), receive a parental label in the absence of a self-label (discordant parental label). In some cases, adolescents may self-label while receiving a parental label at similar times (concordant labels). Chapter 2 investigates the effects of con/discordant labels on three distinct types of help-seeking: formal, informal, and school-based. The findings reveal that discordant self-labels are inconsequential for help-seeking while discordant parental labels may harm adolescents’ help-seeking efforts. Only concordant labels were beneficial for help-seeking. The following chapter expands this investigation to consider whether sociological processes inform adolescents’ help-seeking decisions. While a public health perspective touts the benefits of labels, a sociologically oriented perspective considers how the associated stigma could undermine adolescents’ help-seeking efforts by eroding self-concept. To examine this, Chapter 3 explores self-concept (self-esteem and mastery) as a mediator in the relationship between con/discordant labels and help-seeking. The findings paint a complex picture of how adolescents may arrive at the decision to seek help. Opposing a public health perspective, the findings show that labels diminish adolescents’ self-concept. However, the findings contrast with sociological perspectives – a diminished self-concept is associated with a greater likelihood of engaging in certain help-seeking behaviors. This dissertation provides novel findings concerning how the con/discordance of labels plays a significant role in adolescents’ decisions to seek help. Furthermore, it challenges conventional thinking on how labels’ effects on self-concept shape adolescents’ help-seeking decisions.

Rationale

Objective

Method

Results

Conclusion