Understanding Logistic Regression Analysis Assignment

Assignment Task

Research Question

Are people who live below the poverty level at increased risk of reporting depression in the last year compared to those who live above the poverty level among the 2009 US population aged 12 and older?

Statistical Methods

A logistic regression analysis was conducted for the purpose of this report, within the software ‘Jamovi’. This method allows adjustment for confounders including education, sex, race and marital status (Fig. 1). The reference level will be set as ‘living above the poverty level’ for the exposure and ‘did not report depression’ for the outcome. The logistic regression method will allow the generation of an odds ratio, a measure of association.

Discussion

The report presents a logistic regression analysis of the association between poverty level and reported depression in the 2009 US population aged 12 and older, using data from the National Survey on Drug Use and Health (NSDUH) (2). The primary finding is that people who live below the poverty level have higher odds of reporting depression in the last year compared to those who live above the poverty level, after adjusting for education, sex, race and marital status confounders. This finding is consistent with previous literature that has suggested a bidirectional causal relationship between poverty and depression exists (3). The wider literature argues that mental illness can stem from poverty through avenues like financial stressors, poor physical health, exposure to violence, and a low societal standing (4). However, there is also evidence to indicate mental illness can induce poverty by impairing cognition, reducing productivity, limiting income sources and escalating healthcare costs (3, 5). Consequently, the report results suggest that both poverty and mental illness should be addressed with wide-ranging strategies in order to significantly enhance individuals' well-being and societal health at large.

However, there are some limitations of the data and the analysis which should be acknowledged, due to the potential impact on the results’ validity. Initially, there are notable chunks of missing data, particularly in the outcome variable (reported depression). This has led to a smaller sample size, and also could foster potential bias if the data is not missing at random. As previously hypothesised, the gap in the data may stem from the participants' hesitance to divulge personal medical information, especially in a face-to-face interview setting (1). Individuals diagnosed with depression, for example, might be more inclined to skip or refuse to answer the question than those without it (6). The implication of this is that the logistic regression analysis could underestimate the actual relationship between poverty and depression by excluding a significant number of depressed individuals from the sample.

Another limitation of the report is that it relies on self-reported measures of poverty and depression, which may be subject to measurement error and social desirability bias. For instance, people may over-report their income to appear more affluent or successful (7). Similarly, people may under-report their depression symptoms or diagnosis to avoid stigma or discrimination (8). As a result of this measurement error, the estimated association between poverty and depression may be inflated or diminished. Objective measures of poverty and depression, such as household expenditure or consumption data for poverty, and standardised clinical assessments or biomarkers for depression may be able to amend this shortcoming. However, these measures may not be available or feasible in large-scale surveys such as NSDUH.

Finally, a further significant limiting factor of the report is that it does not account for potential confounding for covariates outside the NDSUH dataset. Aspects like genetic factors, personality traits, social support, significant life events, substance use, and access to health care could influence both exposure to poverty and vulnerability to mental health issues, thus potentially affecting the observed association between poverty and depressions (3, 9). Therefore, it is feasible confounding may still exist and bias the results.

In conclusion, the report provides evidence for a positive association between poverty level and reported depression in the 2009 US population aged 12 and older, following adjustment for confounders. As such, it is apparent that strategies to improve psychological and financial support across the United States are necessary for the welfare of the population. However, this evidence should be interpreted with caution due to some limitations present across the analysis, such as missing data, measurement error, and potential external confounding.

This Statistics has been solved by our PHD Experts at My Uni Paper.